v-xEPA
         United States
         Environmental Protection
         Agency
Policy
(2126)
Washington, DC 20460
EPA231-R-98-006
August 1998
        Technical Methods for

        Analyzing Pricing Measures to

        Reduce Transportation

        Emissions

-------

-------
                  05
Dear Colleague:                               :

State transportation and air. quality planners have requested the Environmental Protection Agency
(EPA) and the Department of Transportation (DOT) for assistance in how to quantify the impacts of
transportation pricing measures, in their regional transportation models. They need this assistance to
develop  regional   transportation  plans,  transportation  improvement programs,  and   state
implementation plans.  This report, jointly ftinded by EPA and DOT, responds to those inquiries and
provides technical  assistance on best  practice  approaches for analyzing various transportation
pricing policies.        .                                                        v

This document is intended strictly to provide technical recommendations and does not advocate the
use of any specific policy measures.  Individual states and regions are always encouraged to select
those policies that best match local priorities in meeting National Ambient Air Quality Standards.

Transportation pricing mechanisms have been recognized as having the potential to significantly
improve air quality, reduce congestion, bolster transit, and improve regional economic performance.
Despite this-potential, implementation  of pricing measures has  been limited  and  substantial
institutional barriers remain.  This report will assist non-attainment and  maintenance areas in the
demonstration of the emission reduction and travel impacts of various market-based transportation
policies.      '          .                      -.          -;     •    > " •  •

Many of the techniques outlined in this document are not commonly used by Metropolitan Planning
Organizations (MPOs).  However, incorporation of these techniques into regional transportation
models would allow the explicit analyses of pricing options.  The recommendations in .this report
provide a means  for areas to receive credits in  State Implementation Plans (SIPS) for  pricing
measures.  The modeling enhancementsimay also be useful for demonstrating conformity.  ,

The methods discussed in this  report are not required for areas considering the adoption of market-
based strategies,  but are thought 'to be useful  tools to  help areas assess the impacts of  such
measures.  We hope that transportation and air quality modelers using this document find it useful
and productive for assessing the impacts of various transportation pricing options.  Please let us
know of any improvements or additonal information that might be useful as we. strive to develop
technical assistance for state and local planners.     ,                         •

                                   Sincerely yours,                          ; .  .
                                                  -=^>f
                                                -.. ?*<:..
                                                         /
Maryann Froehiich             '
Director, Office of Policy Development
                                                 Kevin Neanue
                                                 Director, Office of. Envircrmert arc
                                                > "Isrrirc

-------

-------
                                                              Technical Methods for Analyzing Pricing Measures
                                                                          to Reduce Transportation Emissions
 Table of Contents
 1.0  INTRODUCTION.
    1.1. GUIDANCE PURPOSE ............................ ... .......... . ................. ...,....„ ....... .    ..          .....               i_i

 • 1.2 DEFINITION OF MARKET-BASED MEASURES ... ..... ..... ...... ...... ;. .......... . ..................... .....                   \_2
       ••••••'                 .                                   -                .
    1.3 GUIDANCE ORGANIZATION..., .......... . ........................... . ................ ;-.... ....... . ...............      ;;               j_3

 2.0 OVERVIEW OF MEASURES AND ANALYSIS REQUIREMENTS ............ .............. . ............. .....     2-1

 B 2.1 INTRODUCTION.... ........... . ......... . ....... . .......... ............... ...... : .............. .......            :..........            2-1

 B 22 OVERVIEW OF MARKET-BASED MEASURES... ........... . ............... .. ........... . ..... ..............;......                2-2'
   2.2.7 Description of Market-Based Measures.. ........ ...... ....... . ...... , ......................................      ;..            2-2
   2.2.2 Impacts of Market-Based Measures on Travel and Vehicle Use .......... „,. ................ ..;..„ ............... ...    „. 2-70
   2.2.3 Relationships Between Travel and Emissions ................... . ............. ... ........... ...       :.............        2-75
   2.2.4 Overall Advantages and Disadvantages of Market-Based Measures.. ......... ...... :... ...... . ........ „. ..... .....    2-20

 • 2.3 EVALUATION OF MARKET-BASED MEASURES .......... .......... .....:..... ............... . ................ ........            2-23
   2.3.7 Analytical Process ............. ... ...... , ........ . ..... ... ...... ... ........................... ..................                    2-23
   2.3.2^ Identifying and Assessing Areas of Impact ....... ............. ..... -......-..' ..................................................      2-2P
   2:3.3 Analysis Guidance .......... : ..... : ..... . ................... ...... ........ . ...... .......   ....    .-^              •       •  •  2-42

 • 2.4 PACKAGING MEASURES FOR BEST EFFECT......... .............. ......................... ........ .........        ...          2-6!

 3.0 GUIDELINES FOR ANALYSIS OF MARKET-BASED MEASURES ............ . ...................................... .. 3-1

 • 3.1 INTRODUCTION... ............................. .. ............... „; ....... . ........... : ................... . .................... .     •           3_j

 • 3.2 EVALUATION GUIDANCE FOR INDIVIDUAL PRICING MEASURES.... ........... . ...... ............ ........... ........... '...:...-...... 3-3
   3.2.7 Parking Pricing;.^. .............................. ;... ...................................           .                          j_^
   3.2.2 Modal Subsidies ......... . ........................ ; ...................................... :....:......                        •    3-73
   3.2.3 Pump Charges ............................... . ....... ............................... .". ...............                             3_20
   3.2.4 Emissions Fees ...,.; ..................................... . ..... '. .......................... .    ......'..                         3-2
-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
Table of Contents
   (continued)
4.0 IMPLEMENTATION ISSUES..
                                                                                        .4-1
 • 4.1  INTRODUCTION	4_j

 • 4.2  EPA POSITION ON DOCUMENTATION	4.3
   4.2.1: Public Policy Context.	             4.3
   4.2.2 Legal Authority	;                _    ^
   4.2.3 Administrative Capacity	  4.4
   4.2.4 Technology Availability	                4.5
   4.2.5 Revenue Generation and Reinvestment	~            .
Appendix A: Modeling Pricing Measures in a Traditional Modeling Environment -
A Case Study Approach	
.A-l
Appendix B: The STEP Analysis Package: Description and Application Examples	B-l
                                                               U. S. Environmental Protection Agency

-------
                                                        Technical Methods for Analyzing Pricing Measures
                                                                   to Reduce Transportation Emissions
List of Figures
     2.1  Schematic Overview of Four-Step Planning Process....:..........	:	'........	  2-24

     2.2a Analysis Requirements For:  Parking Pricing..	...........	'.	  2-45

     2.2b Analysis Requirements For:  Modal Subsidies..:	.;............	  2-49

     2.2c Analysis Requirements For:  Pump Charges	;.	        2-52

     2.2d Analysis Requirements For:  Emissions Fees	.'.-	...I;...'....;'.'.     2-55

     2.2e Analysis Requirements For:  Roadway Pricing	....:...	                   2-58
U. S. Environmental Protection Agency

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
 List of Tables
      1.1  Guidance Overview	[                               1_4

      2.1  Overview of Market-Based Measures	,	   	•        2-4

      2.2  Potential Magnitude of Market-Mechanism Impacts	         2-5

      2.3  Characteristics and Comparative Advantages of Pricing Measures	  2-16

      2.4  VOC Emissions for Typical 1987 Automobile	 2-18

      2.5  Anticipated Impact of Measures in Analysis Hierarchy	  2-30

      3.1  Impacts of Parking Pricing	           3.4

      3.2  Impacts of Modal Subsidies	         3_13

      3.3  Impacts of Pump Charges	,	        3_2Q

      3.4  Impacts of Emission Fees	    3_2g

      3.5  Impacts of Roadway Pricing	         3.31

     3.6  Example of VMT Adjustment	,	  3.70
IV
                                                            U, S. Environmental Protection Agency

-------
                                                                   to Reduce Transportation Emissions
 1.0    Introduction
•  1.1  Guidance Purpose

                                •';           '•    .         • ••'      •          '"•'.'
     State transportation and air quality planners have requested the Environmental Protection Agency
     (EPA) and the Department of Transportation (DOT) for assistance in how to quantify the impacts
     'of transportation  pricing measures in their regional transportation  models.   They need this
     assistance to  develop regional transportation plans, transportation  improvement  programs, and
     state implementation plans..  The modeling enhancements may also be useful for demonstrating
     conformity. This report, jointly funded by the EPA and the DOT, responds to those inquiries and
     provides technical assistance on best practice approaches for analyzing various transportation
     pricing policies.                                  '                  ,

     This document is intended strictly to provide technical recommendations and does not advocate the
     use of any specific policy measures. Individual states and regions are always encouraged to select
     those policies that  best match local priorities in meeting National Ambient Air Quality Standards.

     Market-based approaches to transportation policy can provide many potential benefits ranging
     from reduced  congestion and improved regional economic performance to reductions in emissions
     of criteria  pollutants.   These  policies  have qften not been  analyzed since  many  existing
     transportation planning  models are not able  to provide adequate analyses of these issues.  This
     document will enable regional and state transportation planners  to consider a broad selection of
     innovative market-based policy measures that may provide significant air quality benefits.

     Although market mechanisms can be applied in all areas of environmental policy, this guidance is
     rooted in EPA's authority for implementation of the Clean Air Act Amendments of 1990.  It thus
     discusses only methodologies for assessing air quality benefits.  Specifically,  this  guidance
     describes methodologies cities and states can use to analyze the air quality benefits of market-
     based transportation policies. The intent of the guidance is to:         ,

           •   Assist transportation, planning, and air quality professionals at state and local levels in
              analyzing the  emissions reduction potential of market-based approaches; and

           •   Provide  state air agency and U.  S. EPA field staff with technical guidance for reviewing
              analyses of market-based measures proposed as part of state and local air quality
              planning processes.                                     ,           '

     Neither consumer nor traveler response to new policies is ever perfectly predictable. Nonetheless,
     both traditional and more recent travel demand modeling provide powerful tools, for analyzing the
     likely response to changes  in transportation prices. This  guidance discusses those tools, and
     highlights analytical steps necessary tp assure that policies  are well-formulated  and evaluated
     properly from a technical standpoint. It then briefly discusses non-technical implementation issues.


U.S. Environmental Protection Agency                  •          -  .                     ,           j.j

-------
Technical Methods for Analyzing Pricing-Measures*
to Reduce Transportation Emissions
     Proper use  of methodologies  described in this  document will generally satisfy EPA  that
     appropriate analysis  has been conducted.  EPA has no  intent  to  limit areas from using other
     methodologies that might also provide an accurate assessment  of the consequences of market-
     based measures, so Iqng as these methodologies are fully described in an  application for SIP
     credits.   No  guidance can anticipate all local conditions and  EPA is  eager to help potential
     implementers of market-based measures solve any challenges they face.  Because this guidance
     cannot cover all topics related to market-based policies, it also refers readers to studies that provide
     more detailed information on particular topics. As guidance, this  document describes expected
     practices, but establishes no legally binding requirements.

     It should be noted that any analytical method used to estimate emissions benefits for SIP .credits
     must meet all requirements for SIP purposes. It should be further noted that any credits granted for
     implementation of market-based mechanisms will reflect the  uncertainty inherent in the methods
     used.
     1.2   Definition of Market-Based Measures


     The term "market-based measures" here refers to pricing mechanisms which reduce transportation
     emissions. That is, they harness the marketplace.

     This guidance document discusses five major categories of measures:

              1.   Parking Pricing—removing or reducing subsidized or free parking privileges;

              2.   Modal Subsidies—reducing the cost of non-single  occupancy vehicle  (SOV)
                  modes;

              3.   At-the-Pump Charges (hereafter abbreviated to Pump Charges)— changing prices
                  at the fuel pump to increase the marginal cost of auto use (including vehicle miles
                  traveled fees);

              4.   Emissions Fees—levying fees directly on vehicles or their use in a way that reflects
                  their emissions production; and

              5.   Roadway Pricing—charging road users directly for travel on specific roadways,
                  with potential price differentials based on time of day, congestion levels, or miles
                  traveled.

    Market-based  measures typically include actions that either (1) change the cost of specific trips
    (e.g., roadway pricing,  parking charges, and transportation subsidies) or (2) change more general
    or cumulative characteristics (e.g., charges on vehicle emissions or miles traveled).
                                                                 U. S. Environmental Protection Agency

-------
                   '     ''""*     ™-v»-' -'-"•• ••'•'• "•••'••'     fechnicM"MM6'ds for AMyzing Pricing Measures
               • •  •              '                                    to Reduce Transportation Emissions


      1.3   Guidance Organization


      Effectively using market forces in setting transportation policy requires careful work in policy
      design,  community  involvement, planning,  and  implementation.  This  guidance' addresses
      prospective analysis  and discusses other  considerations only  as  they  affect EPA's ability to
      approve a SIP containing market-based measures. Specifically, this guidance describes:       \

               •,  The scope and nature of market-based measures that affect travel patterns; and

               •  Methodological approaches for estimating the impacts of market-based measures on
                  travel emissions when these measures are included in.State Implementation Plans
                  submittals.

      Following this introduction, the guidance is made up of three substantive chapters and two
      technical appendices, described below and in Table 1.1.    -

      Chapter 2 is directed to persons in managerial and decision-making positions with regard to the
      planning or implementation of air quality improvement measures.  This audience includes  state
      and regional transportation officials, planning directors, members of air quality or transportation
      steering/review committees, and EPA regional staff.

      Chapter 2 also provides a summary of key points for more technically-oriented readers too busy to
      review  the full-guidance  document.  Chapter 2 defines and describes market-based  measures,
      reviews the analytic challenges involved in estimating  travel and air quality impacts of market-
      based  measures,  describes characteristics of current  planning and analysis techniques, and
      summarizes standards of accepted basic  practice, including  suggestions  for  more  advanced
      practices that would yield greater accuracy and confidence.                          '

      Chapter 3 is directed at the practitioner and, is much more specific in its technical guidance on
     evaluation techniques  for market-based approaches. Interested managers and decisionmakers also
     are encouraged to read this chapter. Analogous to Chapter 2, Chapter 3 describes five groups of
    .market-based measures, and discusses how these measures are presumed to affect travel behavior
     and  emissions.   The  subsequent discussion  presents  detailed,  step-by-step   methodological
     guidelines and procedures that can be used to evaluate the transportation and emissions impacts of
     each measure. With these analytical recommendations, Chapter 3 describes the expected impact
     and treatment of the measure at each stage of the travel behavior modeling hierarchy.

     Chapter 4 discusses implementation issues. Topics include: putting market:based measures into
     the public policy context; addressing issues of legal  authority,  administrative capacity, and
     technology availability; and consideration of revenue generation and  reinvestment opportunities.

     Appendices A and B  provide technical examples of (A) a traditional modeling approach, and (B)
     an advanced modeling approach.  Appendix B also discusses some  policy issues as part of its
     discussion of how ito approach modeling questions.
U. S. Environmental Protection Agency  '              .                       ,                        1-3

-------
 Technical Methods for Analyzing Pricing Measures'-
 to Reduce Transportation Emissions
 Table 1.1  Guidance Overview
                                                 Provide overview on measures and summarize analytical
                                                 issues
   iii   "iii .'! iii «iH»	««  APPendix A, Appendix B
                     Chapter 4
Compare measures and give step-by-step guidelines ort
analysis
Address implementation issues: legal, administrative, and
technological
1-4
                                                                    U. S. Environmental Protection Agency

-------
                                                     Technical Methods for Analyzing Pricing Measures
                                                               to Reduce Transportation Emissions
2.0    Overview of Measures  and
           Analysis  Requirements
     2.1   Introduction
     This chapter serves several purposes.  It first provides an introduction to market-based measures:
     what they are, specific  examples  and applications, and how they are  effective in reducing
     emissions. Second, this chapter discusses the importance of accurate evaluation of the impact of
     these measures, with specific attention,to the challenges that may be posed by current analysis
     tools and data.  Third, the chapter presents guidelines for effective evaluations, both.to support
     local assessment and acceptance and to allow EPA to grant emissions credit for these measures
     should they appear in a State Implementation Plan submittal.'               ••-..'

     This chapter is specifically directed to persons in managerial and decisiorimaking positions in
     planning  or implementing air  quality improvement measures, including  state  and regional
     transportation  officials,   planning  directors,  members  of  air  quality  or  transportation
     steering/review committees,  and EPA regional staff. This chapter discusses the importance of
     considering market  measures, articulates EPA's  understanding of the methodologies used to
     evaluate their impacts, and offers insights and suggestions as to methods and procedures that
     should lead to enhanced evaluations. This chapter does not directly address implementation issues;
     these are covered in Chapter 4.   .                                  ,
    : This guidance manual does not deal directly with emissions from freight transportation.   This is an
      important emissions source, particularly of NOX and. paniculate matter. While many measures in this
      manual would affect freight, the manual does not discuss the models and data for evaluating freight
      impacts. See Federal Railroad Administration, Federal Highway Administration, and US Environmental
      Protection Agency, Air Quality Issues in Intercity Freight: A Guidebook for Estimating the  Travel and
      Emissions Impact of Intercity Truck and Rail Intermodal Freight Activity and Effects of Improvement or
      Control Strategies, draft final report, prepared by Cambridge Systematics, Inc., et al, December, 1996.
U. S. Environmental Protection Agency
                                                                                     2-1

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions


 •  2.2  Overview of Market-Based Measures


                                                .•.'-.                '   '           '       -.-.«#
      2.2.1 Description of Market-Based Measures
                                               1                     '                            ,           a
      In general, a market-based measure adjusts the price of a given commodity in the marketplace to
      accurately reflect the actual cost of producing and using that commodity.  In the realm of public
      policy and emissions programs specifically, this means incorporating air quality costs in prices
      paid by consumers.

      The prices currently paid for transportation deviate from full cost pricing in two ways. First, prices
      are  distorted by financial subsidies,  such as employer subsidies for employee  parking, transit
      subsidies from non-transit revenue sources, and taxes collected from non-highway drivers and used
      to underwrite highway infrastructure. Prices also are distorted from failure to include external
      costs, such as air quality and health costs.                        ,

      Mechanisms offered for consideration in this report are once-removed from the ideal, since it is
      difficult to alter all policies that affect the current marketplace all in one step. These strategies are
      more properly regarded as pricing mechanisms, which begin showing users the true cost of travel
      and  emissions."

      A number of strategies  apply pricing to influence travel and other emissions-related decisions.
      This guidance document discusses five major categories of measures. Summarized in Table 2-
      1 and discussed in greater detail in the text below, they include:

         •          Parking pricing;
         •          Modal subsidies;
         •          Pump charges;
         • •         Emissions fees; and
         •          Roadway pricing.

      The  magnitude of the impacts of these measures could be considerable.  Estimated impacts of a
      sample of market mechanisms are presented in Table 2-2.2
     Not all of these measures are purely "market based."  For example, subsidies reduce prices to
     below economic costs.  Such subsidies may, however,  alter comparative prices to better reflect
     relative social benefits and/or differences in social and economic costs.
       See Greig Harvey  et.  al., "Transportation Pricing for California: An Assessment  of the Air Quality,
       Congestion, Energy, and Equity Impacts," Vol. 1: Summary Report, Draft/California Air Resources Board^
       1995. This article can provide further details on the potential magnitudes of pricing strategy impacts. The
       numbers presented in Table 2-2 are based on tables prepared in this document.
2-2
                                                                  U. S. Environmental Protection Agency-

-------
                                                               Technical Methods for Analyzing Pricing Measures
                                                                           to Reduce Transportation Emissions


      This section describes various categories of pricing strategies and presents their major features; it
      is followed by a discussion (Section 2,2.2) of how these measures are expected to affect'travel
      behavior and emissions.
U. S. Environmental Protection Agency
                                                                                                    ,2-3

-------










8
3
(/)
CO
CD
S

"O
CD
CO
CO
m
•+!*
11 8*
is ifl
i co
* S
&0 ••
*s T5
•S ®
°* i 5"
fl.2
H g. ^^
a I§ ^
x iSi
fe §j?


Shift work trips to alternate modes, work
arrangements (telecommuting,
compressed work weeks) or off-peak
periods, leading to reduced peak-period
vehicle trips and VMT.

05
•^
cl
«••%

| |
0) O
Employee parkin
Rates to favor sh
over long-term

Shift travel to alternate modes and/or
destinations, leading to reduced vehicle
trips and VMT.





CO
O3
•22
g
CO
Q.
"S
CD
CD
.i>
CO
Q.
1
Jo
"55
jQ

CO
s
1 -
•i; co
CO CO
O 03
ll
01 a.











CO
^
o
£
OJ
c
"i

Increase use of subsidized non-SOV
modes, leading to reduced vehicle trips
and VMT, particularly in peak period.


CO
03
^
"eo
=3
0)
CD
•0
Parking "cash ou
HOV subsidies
Non-motorized m
1
CO
O CO
£ 8
O i—
co "53
CO f-
CD *

§ ™
~ 0
! "i
CO >,
CO -°
> CO
P CO
I-"8
M E










CO
."*
TJ
"Z5
•3 " •
to
CO
•d
o

Raise overall cost of driving for all
travelers, causing shift jn modes,
. reduced trip lengths, and more grouping
of trips, leading to reduced vehicle trips
and VMT.


en
o
03
i|
to .2
8 ro
.52 0
Per gallon fuel pr
different cost allo
CO
n
CO •—
*-. c°
O c
=; CO
fD £

£ o
ll
o 1 g
to R •«
si I
CD CO CO
~ "to ,£
cc 8 .s




"S-
£
CO
^e
,3
75 — .
«S, jo
to -*S
CD M
OJ ^""T
JZ ?*"
C3 us*
U>
E «
^ *m
y^ j™£ ,

Reduce ownership and use of older,
higher-emitting vehicles.


CO
o
co Ic
CO CO
o ^
"2 g
Age-based regist
(periodic fee inde
age and VMT)

Reduce ownership and use of higher-
emitting vehicles and encourage better
vehicle maintenance.

TJ
8>
3
CO
CD
X
1—
Emissions fee (VI
emissions rate)
£2
CO CO

O "to
CD .{2
o £
2 co
93 ^D

f-" •«—
O m
to >
-8
"fl -I
JS 0 =3
2 CLT3
S- O 0
£^ 1— L_
£ Q. 0.

- s*f

f







O>

u.
 J>
O3 O
Congestion pricin
linked to congest!
of day
m»
^
*O
a—
«o_
so to
O3 .03
O3 ^^
CO 'o •
-§ 45
||

-------
J  §
•3 til
.
   1

a
       Si
        OS
                                        O)
                                        CM
                                        CO
                                        CD
                                       CM.
                                       CO
 O CO
 ^ CM
                                                 CO -q;
                                                 O CM
 CM T-.
.*- CO
                                                O)  CO
                                                O  CM
 CM_ »-
 i- CO
                                                
ng
ri
r
king Charg
Regionwide Congestion P
de Employee Pa
                                                              .CM  o>
                                                              ^-  CD
f-;  10
CM'  oi
                                                              CM  co
                                                              CM'  i^-'
                                                              cq  co

                                                                  CM
                                                              co  r^
                                                              CM  <3)
              I--.  CM
              CM  C7>
              CO  CD
              CM  O>
                                                              CO T-
                                                              CM CO
                                CO  r-
                                CD.CO
                  CD   •


                  ITS CO
                                           to

                                           CO
                                                         1^

                                                         CM
                                                             t--;,
                                                             CD'
                                coiq
                                O CO
                                                                                CM t^;
                                                                                O ,CM.
                 CM  CO
                 o"  co
                 <•> co
                 o.co
                                CM  O>
                                d  CM'
                                           to
                                           •*
                                           in
                                           co
                            iv.
                            IO
                                 •>3;
                                 10
                                .cq
                                10
                           CD
                           •*
                                                        cq
                                                        co.
                                                           ..
                                                             O
                                                             rf.
                                                             'CM
        CXI
        co
        co
        CD
                                                                        8
                                                                        -
se
n
n
Regio
2a) $1.00 per day
2b) $3.00 per day.
re
lo
lo
                                                                                  -1
                                                                               E  S
                                                                                      -
                                                                                                 CB
                                                                                                 Q.
                                                1
                                                                                          «o«
                                                                                                               0
                                                                                                              -S-

                                                               t3
pe
pe
—  o> o>

§
 •n  O O
 £  iq p
=  O CM
 O  <& <&
 CO  ^ ^
 OS  CD ^2
(j)  CO CO

Jo"
                                                                        8
•5
_CB

 I
UJ
                                                    |i
                                                    g  |.
                                                    iS
 o
 u _
"5. o

 I
HI


                                                       1
                                                       ,3
                                                                                 o
                                                                                 CD
                                                                  5
                                                                                                                                 I  •
                   s
                   s
                   g
                   o
                   £
                                                                                •i  §
                                                                                3  ~8
                                                                                 §  s

                                                                                                                                       i  &

                                                                                                                                                                     §
                                                                                                                                                                     s
                                                                                                                                                                     §


-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
     Parking Pricing
            Parking Pricing—removing or reducing subsidized or  free parking
            privileges .that serve as an incentive to drive.
     Free or discounted vehicle parking creates a strong incentive for vehicle ownership and driving for
     all types of travel, both work and non-work. National data suggest that nine out of 10 workers in
     metropolitan areas park for free or at below-market prices, yet employer-paid parking is treated as
     a tax-exempt fringe benefit by federal and state tax laws, a subsidy value of up to $170 per month.3
     Free parking provides a similar attraction for non-work travel for shopping and other purposes in
     suburban and outlying areas, in  contrast to parking in downtowns and activity centers which is
     priced, but where there also  exist alternatives to driving.  Internalizing the cost of parking would
     reduce the demand for private vehicle travel in favor of alternatives.

     There are many ways in  which  parking pricing  can be introduced, the major variation being
     whether it is applied exclusively to work travel, or to non-work travel as well.

     Pricing commuter parking is  perhaps the most common application of parking pricing. Employees
     may be seen as more captive to their location than non-work  travelers, and because  of their
     repetitive  travel to  the same site  during peak hours, they  usually have  access to  the best
     ^transportation alternatives.  In many instances, employees already pay a parking charge if they
     work in a downtown or activity center; a small percentage may experience some parking cost as a
     result of the employer's involvement in a trip reduction program. Employee parking pricing as a
     regional pricing mechanism can take many forms, including:
           •  Imposing a parking tax on all or a portion of the parking supply;
           •  Shifting public policy to eliminate or reduce tax exemption of parking subsidies;
           •  Eliminating early bird/all-day discounts, or replace these with peak-period surcharges;
           •  Introducing a minimum parking charge for all employees (i.e.,'where all parking is
              raised to some threshold level, say $3 per day);
           •  Requiring parking charges at sites that meet  certain criteria (e.g.,  with  100 or more
              employees, where vehicle trip rates exceed a standard, etc.);
           •  Replacing free parking with a transportation allowance applicable  to  various modes,
              coupled with introduction of parking rates;
           •  Offering the value of the subsidized parking space back as a cash credit to persons who
              would agree not to drive (popularly termed " Cash Out");
           • . Offering discounted parking for High Occupancy Vehicles (HOVs); and
           •  Levying "area"  entry fees on Single Occupancy Vehicles (SOVs), particularly in peak
              periods (these are generally considered more of a "congestion pricing" action).
     Attaching parking prices to non-work travel is more problematic.  Parking charges  for non-work
     travel include:
    3 This amount is adjusted periodically by the IRS in $5 increments to adjust for inflation.
                                                                  U. S. Environmental Protection Agency.

-------
                                                          TechnicaTMetnodsfor Analyzing Pricing Measures
                                                   1          '        to Reduce Transportation Emissions

               Imposing [higher] charges at major traffic generators, such as malls and activity centers,
               shopping centers, hospitals,,educational institutions, and entertainment sites;

               Levying a parking tax on public or private parking operators;
               Restricting off-site parking in residential areas and/or installing meters or increasing
               metered rates; and

               Restricting parking supply or raising parking costs at residential locations.
      Modal Subsidies
            .Modal "Subsjdies^nlerwriting-part^ of the. cojst jot non-SOV modes;
            thus  reducing^ their^fJTOe relative  taj Sp\£ m"6aes> tojncrease  theic
                            *        *    st<" ~"' /• J"~  ^  "    ^     "'  ", <->•      '  "•'•
      By lowering the cost of alternative modes such as transit or ridesharing relative to the automobile
      or by providing direct (or indirect) incentives for non-motorized modes (bicycle and walking), the
      attractiveness of these modes increases and their usage  would be expected to increase. ' As
      alternative mode use increases, a reduction  in vehicle trips and vehicle miles traveled (VMT)
      would be  expected to reduce emissions'.   If subsidies lead to proportionately higher, use of
      alternative modes  for work trips  in  peak  periods, then congestion- relief also  might occur,
      potentially with additional emissions benefits.

      There are a number of ways in which modal subsidies  can  be  introduced or enhanced through
      public policy initiatives:

               •  Lowering the usage cost of non-SOV alternatives, for example, reducing transit
                  fares, or exempting HOV users  from various tolls and fees;
               •  Allowing employers to grant (or employees to receive) tax-exempt, subsidies for
                  carpool  or  non-motorized mode use, or  to grant larger  subsidies to transit or
                  vanpool  users;                                                     '     , .

               •  Offering tax credits to employers to cover their cost of providing subsidies; and
               •  Directing implicit subsidies to  rideshare users in the form of gasoline allowances,
                  parking discounts, vanpool capital or insurance assistance or maintenance, etc.
                                    \           •''',-•                            .
     Reliance on modal subsidies as an emissions reduction strategy  has  two major problems.  First,
     they require a net outlay of resources to support, which if not financed by a specific revenue'
     source^ will fall as a burden on employers or state/local governments.  Second, they may further
     distort current imperfections of the marketplace, by adjusting for existing subsidies through the
     addition of new subsidies, and potentially inducing demand on highways with freed-up capacity.
     While this may, in principle, begin to equal out the subsidy treatment across modes, it does not
     serve the purpose of making current prices a better reflection of costs, and hence, may just further
     skew the  operation of the marketplace.  While these limitations are important, subsidies are
     advantageous hrthat travelers view them as a benefit, rather than a cost.      ,          .'  .   '
U. S. Environmental Protection Agency
                                                                                              2-7

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
     At-the-Pump Charges (and their relative, VMTfees)
            At-the:Pump Charges—Using the pump price of fuel to increase the
            price of driving and thus discourage private vehicle  (especially SOV)
            use.	;	'	
      By raising the price of fuel, this measure reduces emissions by increasing the overall cost of travel
      for all private vehicle trips, thus affecting the frequency and length of such trips.  A pump-based
      charge can be based on numerous factors. For example, pump-based charges can be:

               •    Structured to cover a more complete proportion of the cost of road construction and
                   maintenance;

               •    Used as a medium to collect insurance payments under a pay-as-you-go insurance
                   coverage policy, as is being considered in some states; and/or

               •    Structured to account for social costs, such as emissions  (often through links to
                   VMT).

      Experience has demonstrated that fuel price increases can be controversial at both the national and
      state  levels.   Some states may, however, see some benefits  in  designing systems that more
      accurately reflect the marginal social costs of vehicle usage. As with all the policies discussed in
      this document, state and regional decision makers are encouraged to develop the best strategies for
      meeting emissions reduction targets.


      Emissions Fees
            Emissions Fees—-Levying fees directly on individual vehicles or their
            use in a manner that reflects their emissions production.  I
     This category of pricing measure consists of charges based on vehicle characteristics, including
     emissions production, rather than on travel behavior.  It attaches a higher cost to owning high-
     emitting vehicles, discouraging ownership and usage, as well as encouraging proper maintenance.
     Exactly which is affected more—ownership or use—depends on the structure of the particular
     mechanism.

     Emissions fees may be based on;

               •   Vehicle age;

               «   Vehicle type; and/or

               •   Vehicle emissions, using:

                   —    Estimated emissions; or
                   -    Measured emissions, as determined at an annual emissions test.

        In most cases, emission fees also would be linked to VMT, based on odometer readings.
2-8
                                                                 U. S. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                            ,                                          to Reduce Transportation Emissions


      Generally, emissions fees are imposed as part of a vehicle registration fee, and tied to required
      emissions tests.  While annual fees may be most common, fees also may be assessed on a more
      frequent basis or in the form of a VMT-based .emissions fee at the fuel pump.  This version of the
      measure potentially could have a greater impact on travel.


      Roadway Pricing
            Roadwa| Pncing—lm|3osfpg| direct, usen;chargesjof travel H
            roadways; with poten«a| price differentialsi'd^ndjhg'ori^pe ofVay or
            jsopgeslfori teyels,:  .,,, - '1 \  : ,  -  y^    ' '"""'"-  '"     '; \      """
      Fuel taxes and vehicle registration fees are the-conventional means for traveler financial support of
      the  highway system.  These measures,  however, neither  cover the entire cost of supplying that
      transportation capacity, nor serve as any indicator of the level of service which is being demanded
      from the transportation system at any given time.           ,

      Road pricing establishes a more direct and immediate link between the cost of roadway use and
      payment for that use through:

               •   On-the-spot collection of fees or tolls;                     •                       .
               •   Pre-paid permits; or                  ,                                 ,

               •   Electronic monitoring and periodic billing.

      Manually collected tolls may be somewhat counterproductive as the resulting delays at toll booths
      can  increase emissions levels. However, electronic toll collection systems solve this problem.

      Roadway pricing can be narrowly focused, such as installation of tolls on individual  bridges or
      tunnels.   Alternatively,  roadway pricing can be more broadly implemented  by,  for example,
      applying them to a system  of facilities that define a corridor, to a particular functional class of
      facilities such as all expressways, or—in its most extensive form—to all. travel into, within, and
      through an area.                                            ,'•••,»'

      In the specific case of congestion pricing, the usage fee is assessed at the marginal cost for each
      additional vehicle. This places a premium on the privilege of travel during peak demand periods.
      Travelers may face fee schedules ranging from peak-only fees to fees that are variable by time of
      day, facility,  or .level  of use.   Congestion pricing  provides incentives for  travelers to take
      congestion costs into account when making trip decisions, thus leading to more efficient use of
      facilities and avoiding construction of expensive new capacity.  By reducing congestion, such
      pricing also provides higher levels of service to those willing  to pay for it as well as auxiliary
      benefits, such as providing better access for public safety vehicles.4
       Congestion pricing creates greater economic efficiency in the sense that it increases the net benefits to
       society.  This is because congestion results in significant  costs to society, and these costs can be greatly
       reduced if fees are used to reduce congestion. The funds collected through fees then can be returned to the
       public in another form.  The fee is thus not a true cost to  society, but simply  a transfer, while the cost of
(Footnote continued on next page...)
U. S. Environmental Protection Agency
                                                                                               2-9

-------
      Seeing an increase in travel cost based on roadway prices, the traveler may either pay the increase
      in cost or opt for alternatives.   For work travelers, where the destination is largely fixed in the
      short run, the choices include different modes,  different  routes, or, in the case of congestion
      pricing, different times of day.  Non-work travelers face the same options, as well as the ability to
      choose destinations that can be reached without payment of tolls, or that provide more services for
      the given outlay.


      2.2.2 Impacts of Market-Based Measures on Travel and Vehicle Use

      In response to a properly designed and implemented program of market-based measures, travelers
      can be expected to do any or all of the following:

            •    Change travel mode to walk, bike, transit, carpool;
            •    Change destination;

            •    Change travel route;

            •    Change time of day of travel;

            •    Change the pattern  of trip making (e.g., consolidate trips or group them into chains);
            •    Substitute telecommunications for travel;

            •    Adopt alternative work schedules or other arrangements to reduce the number of trips;
            •    Shift patterns  of vehicle use in  multi-vehicle  households and change maintenance
                practices; and

            •    Make changes in residence and employment location (and thus change pattern or rate of
                growth of community).

      Each of the  described market-based measures, with  the  exception  of emissions fees,  affects
      emissions  by  first affecting vehicle travel and usage patterns. Emissions fees, if imposed through
      an annual  or  periodic fee, would be expected to have only an indirect impact on travel patterns,
      primarily  by  affecting  vehicle  ownership.   Although  the  other four measures influence  travel
      patterns, they nonetheless vary in the specifics of their impacts on travel and emissions.  In brief:

                »,   Parking  pricing reduces  or  redirects vehicle  trips by increasing  costs  at
                   destinations, often through the  levy of  visible  out-of-pocket charges at the site.
                   Parking pricing is most effective when it is applied regionally.

               •   Modal subsidies increase the use of less-polluting modes through  a reduction in
                   their relative price.

               •   Pump charges  reduce total vehicle travel by raising the price of fuel.
       congestion has been reduced. Those road users who value their trips the most are willing, to pay congestion
       fees in return for reduced congestion and will benefit from the fees on balance.  Those road users less
       willing to pay for peak-period travel will  prefer switching modes, routes, or time of travel, and the
       disbenefits they experience under congestion pricing will be smaller than the net benefits to the first group
       of users.  Regardless of the equity implications of this approach, it generates net benefits to society.
2-10                                                                U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                       .                       .                        to Reduce Transportation Emissions

                •  Roadway pricing reduces vehicle trips along given roadways  by raising the en-
                   route cost of travel, either on a per-mile, per-facility, or per-entry basis (Roadway
                   pricing also can affect total travel if it is implemented on all roadways in a given
                   area).                        .          '          -

                •  Congestion pricing is a variation of roadway pricing that reduces vehicle travel on
                   congested facilities at congested times of day.
                   .--•••      '  -    •       -   - •      '          r    '.  '  •  '        •         . '   •  •
      There is ample evidence to substantiate that travelers do respond to cost changes.  Much depends
      on the level of the charge and the availability of substitutes. The higher the income of travelers  '
      the less likely travelers are to respond to modest changes in the price of fuel, parking or vehicle
      ownership.                                             .                      "  '

      The relative impact of pricing mechanisms depends on the pricing instrument itself and how direct
      a signal it provides; charges which are periodic and disassociated with travel behavior are unlikely
      to be linked in a conscious decision-making process. The availability of realistic options also is an
      important consideration, although more options exist (e.g., carpooling, deliveries, telecommuting)
      than are commonly acknowledged and options would  increase substantially if driving prices are
      raised. Revenues derived from pricing also can enhance'the number,and quality of these options
      Under the current system of subsidized and indirect pricing, however, these  options are not likely
      to develop.                                                '  -  .

      The balance of this section summarizes the key  assumptions underlying market-based measures
      and describes expected impacts of each measure in greater detail.

      Impacts of Parking Pricing

      Parking pricing provides direct signals to the traveler on the cost of'driving!  Because a parking
      charge is often collected at the point of use,  it has an explicit linkage to  the particular trip  Many
      studies have concluded that such an " out-of-pocket" cost has a greater impact on decisionmaking
      than costs that are averaged over a longer time period or multiple uses.

      If parking  pricing is implemented at a site that must be  accessed (e.g. a work site) it is difficult to
      avoid or ignore, particularly in the short run. Faced with new or increased parking charges  -work
      travelers  would  be expected to  consider shifting to  non-SOV modes, telecommuting, and/or
      compressed work week arrangements.  The effect on emissions would be through reduced vehicle
      trips and VMT, primarily during congested peak travel periods.

     Non-work travelers faced with parking charges would  be  expected to shift modes or travel less,
      such as by grouping trip functions to accomplish more on the same trip tour, (the effect is  likely to
      be stronger if there are no free parking areas). However, non-work parking charges,  also may be
      avoided simply by traveling to another destination, which may entail more travel as a result of
     being further away.

     It can  be physically difficult to institute-parking charges at a high proportion of destinations,
     particularly for small  employers and  for non-work situations.   Ensuring that the  charges are
     actually  implemented, and  not  diminished by  employer, or merchant subsidies,  requires
     enforcement.  One other characteristic of parking pricing is that it has less impact on  longer trips,
     since parking charges represent  a smaller  proportion  of overall trip  costs;  this may .be.
U. S. Environmental Protection Agency                                         ''	"	~ 11

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions


     advantageous  in addressing  certain pollutants  since  pricing would have  a greater  impact in
     reducing short trips and associated cold starts.
                                                            1     f '','•,
     The effectiveness of any parking pricing arrangements depends on the level of the price change
     and  the uniformity of the  application.  Major differences in rates across a region (e.g.,  suburb
     versus core) or exemption of many sites from pricing are expected to encourage shifts in choice of
     destination to avoid or minimize the charge, even if such a change involved more travel. This is
     particularly a concern  with non-work travel.  Such a result could have an undesirable effect on
     total emissions.


     Impact of Modal Subsidies                          .

     Modal subsidies have the advantage that they are comparatively easy to implement. The traveling
     public is not likely to resist being offered a discount, in contrast to paying a charge.

     The disadvantage of modal subsidies is that they must be paid for from some source, thus raising
     the question of who bears this cost (employers,  governments, taxpayers,  etc.), although revenues
     raised through  other pricing measures could  serve  as  the  funding base for these subsidies
     (potentially a  mechanism  for addressing equity concerns).  A troubling aspect of subsidizing
     alternative travel modes is that it may exacerbate current  market pricing distortions, by trying to
     overcome one subsidy with another. Evidence also suggests that travel behavior is less influenced
     by a cost incentive than a disincentive.  In  light  of these characteristics, modal subsidies are best
     seen as a strategy supportive of some other active pricing mechanism.


     Impact of Pump Charges (including VMTfees)

     Pump charges  are one of the  most broad-reaching pricing strategies  available.  Implemented
     directly through the cost of fuel, a pump charge  closely tracks vehicle usage rates, bears upon the
     travel  decisions of all travelers  and  trip purposes,  and  can be implemented with relative
     technological and administrative ease. In general, fuel price surcharges cause travelers to:

               •  Switch modes;

               •  Restructure travel to emphasize shorter trips  or grouping of trips;

               •  Seek a wider variety of options for satisfying activity needs, including  non-travel
                  alternatives  and  locational preferences  for   more  compact  and   integrated
                  developments; and/or

               •  Purchase more fuel-efficient vehicles.

     The last of these will tend to eventually offset the VMT benefits of higher fuel prices, although
     decreased energy consumption is still beneficial.

     For  work travel, the primary response of a pump-based charge is a shift from driving to other
     modes or to alternative work arrangements (e.g., telecommuting or shortened  work weeks).  For
     non-work travel, destination shifts or errand/trip chaining might be more common.
2-1.2                                                         •       U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                                     to Reduce Transportation Emissions

     Because fuel price surcharges increase cost burdens in direct proportion to the number of miles
     traveled, they would be expected to affect longer trips more than shorter ones. As such, they can
     be complementary to fixed charges like parking pricing, which diminish in influence as trip length
     increases as parking prices become a smaller proportion of total trip costs. Higher pump charges
     also  could encourage less rapid  acceleration,  better  maintenance  and  other  driving-style
     improvements to reduce fuel consumption,
                                    ••'.-.'       S
     Impact of Emissions Fees

     Emissions fees vary in their impacts, depending on  how they are designed.  Most simply,  a
     surcharge is added to the standard  vehicle registration fee that reflects vehicular emissions, 'in
     simple applications, requiring almost no  additional administration, the fee is indexed to the type
     and year of vehicle as a proxy for contribution to air-borne pollutants,  Depending on the level of
     the surcharge, this approach would be expected to discourage ownership of older and presumably
     higher-emitting vehicles.                                                                  '

     More targeted  applications link charges to each vehicle's  annual VMT, or base charges on
     emissions rates as actually measured at the time, of inspection, these approaches have a more
     direct effect on ownership and usage rates of high-emitting vehicles.  Any effect on daily travel
     behavior, however, would depend on how closely and with what impact the  owner links the fee to
     the level of travel; fees that are administered infrequently have a more tenuous tie to daily travel
     decisions.  The primary advantage  of emissions  fees  is that  they target  emissions production
     directly, and cut across all ownership and use  situations.  The primary disadvantage is that the
     simpler applications may not produce the desired results, while the more complex (e.g., real-time
     emissions rate and VMT) are more challenging to implement.

   ,  Addressing particulate  emissions  through emissions  fees poses  a unique challenge, because,
     reentrained dust emissions depend on the  location and road type (e.g. unpaved versus paved) more
     than the vehicle type.  Emissions fees tailored to these particulates would be difficult to design.
     Fees based on VMT could impact particulates, but could not be readily targeted to high-emitting
     activities or vehicles in the same way NOX or VOC-based fees could be, for example

     Depending on the specific fee structure to be implemented, the following consumer responses to
     emissions fees might be anticipated:                                                        '

              •  Consumers  will replace high-emitting vehicles with cleaner  ones (also creating
                  greater pressure on manufacturers to produce such vehicles);

              •  If the  ownership fee is very high, households will reduce the number of vehicles
                  owned;

              •  Multi-vehicle households will make greater use of their cleaner vehicles; arid
              •  Where fees are based on actual measured emission's rates, consumers will maintain
          ,        vehicles to the highest possible standards.               '

     Reductions also may occur in overall rates of travel, depending on how much of the fee is based on
     VMT usage, and  how frequently  the fee is  levied  and/or  people  are reminded of thfe  fee ..
     (infrequent/periodic fees may not be strong disincentives to daily use).        '
U. S. Environmental Protection Agency                                                      -     -    • 2-13

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions


      Impact of Roadway Pricing

      Roadway  pricing is  an attractive approach  because it is a classic application of direct user
      charges. Travelers are charged for the number of trips they make and, in the case of congestion
      pricing, for the congestion cost that they impose on the transportation system. Road pricing serves
      as a fairly immediate  signal to travelers to consider alternatives to driving, including travel by a
      different mode, time of day, or route.  Congestion pricing offers one of the few direct mechanisms
      to reduce congestion and to address the emissions impacts specifically caused by congestion.5

      Road pricing schemes are most effective  when  they have  full coverage of the road  system
      (including  collectors and local roads). However, it  is probably more  feasible to price controlled-
      access facilities. Pricing systems need to consider possibilities for route diversion onto un-priced
      arterials and local streets. Traffic-calming strategies and changes in signalization could reduce the
      efficiency of alternatives and make the priced alternative clearly superior and worth the price.
      To avoid road pricing, travelers may shift to any of the following:

                •   Unpriced routes that are ill-equipped to accept such traffic or entail  greater VMT  to
                    reach intended destinations; or

                •   Other destinations that are further away  and/or not as serviceable by non-SOV
                    options.
       It should be noted that, there might be cases where congestion relief allows speeds to increase to levels
       where emissions rates are actually higher, at least for some pollutants. In other cases, speeds may increase
       such that emissions rates are reduced.
2-14
                                                                     U. S. Environmental Protection Agency

-------
                                                         Technical Methods for Analyzing Pricing Measures
                                                                    to Reduce Transportation Emissions


     Comparison of Characteristics Among Measures

     Table 2-3 summarizes the major characteristics of the market-based measures described above.
     Each measure has: distinct strengths and weaknesses. No single measure adopted by itself is likely
     to achieve all program objectives. In fact, the most effective program-will be one that combines
     complementary measures, thus improving overall traveler acceptance and maximizing  emissions
     impact.                                                      •

     Moreover, impacts can vary among the different specifications of an individual measure and
     differences  in implementation. This  calls for careful definition and evaluation  of individual
     measures, identification of potential benefit-cost synergy among measures, and analysis of how
     individual measures can be integrated into an overall system plan.          '


     2.2.3 Relationships Between Travel and Emissions   . '  ...


     Market-based measures are implemented to achieve two types of social objectives: increased travel
     mobility, measured in terms of transportation access, level  of service and  travel trip time; and
     reduced vehicular emissions.  The ultimate  impact of market-based measures on emissions  is '
     closely tied to program setting and pollutants to be addressed.

     Areas attempting to achieve compliance with of maintain National Ambient Air Quality Standards
     (NAAQS) need plans to reduce ozone, carbon monoxide (CO), or particulate matter (PM).  Areas
     with ozone  violations must reduce the component  pollutants of volatile organic  compounds
     (VQCs), oxides of nitrogen (NOx), and CO.  These pollutants have different characteristics, stem
     from different sources and conditions, and are typically addressed by slightly different strategies.
     Program design is complicated by pollutants' different relationships to factors such as speed range,
     flow conditions, number of cold starts,,VMT, and-vehicle ownership or use characteristics.

     Only careful analysis can identify and evaluate the overall air quality impacts of pricing measures.
     However,  knowing the particular circumstances that affect the various pollutants is important in
     defining the group of strategies likely to be most effective in meeting a particular area's objectives.
     This section describes the emissions that are commonly the focus of market-based measures and
     identifies transportation  variables associated  with each, thus pointing out the travel behavior that
     must be influenced to best reduce emissions of each  type of pollutant.
U. S. Environmental Protection Agency                              ~         !  '       '~~~   !     2-15

-------
1










i5
3
to
CO
CD
S
D)

• MM
_.O
CL
u—
ative Advantages o

CO
CL.
8 E
i °
1 1
£? c5
I i
^-» § nHr
f 1 .52
f «§ ,8
-S a °
^ .§ co
•flo
"•S M
*§! [5 CO
5 « Ol
.8 Ja OJ
Jl^ 2
£ 0 |Si

-• ,









i
S
mS&
a










§
J*
1




o>
c
I
J=
1
*o
"2 *-
o o
^M EM
S —



 •£
° Ii"
"g >
O QJ t—


§Q^ O5
.1 "^-
Difficult to implement in m
Creates demand for lower
be more distant (causing t
Relatively easy to evade
Difficult to enforce
• • • •
cn +*

•yj :t± ,Q3
8 =. 5E
"® CD 3=
o en c/j

,2° ^ =-
2 ™ ">
*3 ;£ .—
0 „, =
> Relatively direct signal
• Relatively strong signal (
> Affects shorter trips mor<
cold starts)
> Results in better trip plar
chaining use)
«» w ^ *»



o
ca
1 CD
•a fo
ca _o
13 -C
1-8
O ~^

cn
e
0
c
O)
JS '
CO
a.
CD "to
o to
E o S
>> "o .2
ja j= -a
C/3 cQ
CO CD -_
III
"« E ^

o> ,c >
^~~ _. - •;— ;
Requires societal outlay
Further distorts market pri>
cheaper than cost
Can cause sub-optimal sh
carpool)
Incentive may be less effe
. . . .
ro <*-
S —

CD O
1 |

Io5 °^

1 J
E? P.
• Easy to implement (throi
cost)
• Shift to modes that pollul




CD
CO ~
O CD
ECL
j=
"co o
c: v-
03 >
% o.
~ J 2 .2
: S Q.
(A
O
-a
m
J2
3
CO
"c5
"o
S.
S
.0
1
'c
cu
•J3

^_. CD
CD 0>
E 1
.£= c
§ CO
CD "D
*-; QJ

Must be substantial to hav
VMT impact can be dimini!
replace current fleet. .
• •
:"S
— _ ^—^ •—
2 -S S
.§. S "o
E m c

*r- "qT C
c > O
§ J§ -a"
"5. - CD
E •£ 9-
> Technologically easy to ii
administrative costs
1 Affects all travel (non-wo
1 Encourages shorter, grot
trips
Increases fleet efficiency



O i 	
	 1 —
ii
"S 3
"° 1§
to JO
8 j32 ,
CD '£
CO. i

ck Q.

Q>
S?
to
'*>
CL
£
Q.
CD
cr
"co ^_^
1 s
IE
-t= CD
5» Q.
o
CO -S^
CO T •

-^ S
co "
CD ifc
Infrequently levied fees w
More effective fees are dil
• •
f 1
E o
"CO Q.
.2 -^
CO S=Z &
o o £

.co CD :Q oj
111 1
0 CD C O>
• Direct impact on source
• . Reaches all vehicle own
• Technical implementatic
versions
• Encourages better main
emissions rates



co"
CD .Q.
° m -£=
-£= •> 52
§2 o g
co g- •£ =s
c .E QJ o
I§ ^i
O ^^ • O -C
-C^ CD 0> CO >•
-^ § ro -gj §>
o g ^ £ 3
QD -2 •=; "£* O
£l I i-%

CO >,
CO ~ > j^
S JO CD CO
C3 £ ' c= T3
C. ^^ CD- •*—
_ co So
2-2 o5 g
§ -§ g.J
"£ CD CD tl

-------
                                                         Technical Methods for Analyzing Pricing Measures
                    '"'.-,.            ,           •  , •  j                to Reduce Transportation Emissions

            "'""                  '             '        • '   '   •
      Volatile Organic Compounds (VOCs)
                ' •      '      ..'"''       *K -                        •
     VOCs are unburned hydrocarbons  (HC), typically  generated  by gasoline combustion  engines.
     VOC emissions per vehicle have been'. substantially abated through automobile technology and
     fuels programs over the last decade, though it remains a problem in many areas and threatens to be
     a long-term issue as both VMT and the number of vehicular trips .continue to rise.  VOC emissions
     are influenced by speed range, flow conditions, number of vehicle trips and, VMT.

     Engines produce VOCs in these elements of operation:

               •   Cold Start Emissions-^oVA start emissions,occur wfien a vehicle is first started and
                   the catalytic converter is cold.   For a given vehicle and weather conditions, cold
                   start VOC emissions  are the same for all trip lengths,  thus they  are higher as a
                   percentage of total VOC emissions for "short" trips. As noted in the table below,
                   cold start emissions represent about 65 percent of total VOC emissions for a 5-mile
                   trip, 53 percent of total VOC emissions for a 10-mile trip, and 39 percent for a 20-
                   mile trip.

               •   Running Emissions—Running emissions occur once the catalytic converter has
                   warmed  up. VOC running emissions vary with the number of miles driven and the
                   speed of travel. Assuming the same speed conditions, a 10-mile trip would  produce
                  twice the VOC emissions of a five-mile  trip.  VOC emissions generally decline
                   rapidly with speeds up to about 25 mph, fall more gradually  up to 55 mph, and then
                   rise moderately over 55 mph. A vehicle traveling in free-flowing traffic emits only
                   one-third as much VOC emissions as one in congested  traffic. As VOC  running
                   emissions are proportionately related to trip length, they increase as a proportion of
                  total emissions as trip  length increases. The running emissions for a 5-mile trip are
                   roughly 21 percent of total trip emissions, 35 percent for  a 10-mile trip, and 52
                   percent for a 20-mile trip.

               •  Hoi Soak—Hot soak emissions occur  after the vehicle is turned off when  residual
                  heat in the engine causes fuel  in the gas line and carburetor to evaporate. These
                  emissions also are independent  of trip  length; they therefore represent a greater
     .V             share of short-trip than long-trip emissions (approximately  14 percent of a 5-mile
                  trip,  12 percent of a 10-mile trip, and 9 percent of a 20-mile trip).

     While longer vehicle trips produce less emission per mile, they still, of course,, produce more total
     emissions.  A 20-mile trip, for example, produces 35 percent more emissions than a 10-mile trip
     and 64 percent more than a 5-mile trip.            -                             !   •
U.S. Environmental Protection Agency                  '                         .'                  2-17

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
                Table 2.4 VOC Emissions for Typical 1987 Automobile,
                      Normal Operating Conditions (grams of HC)
5-Mile Trip
Cold Start
Running Exhaust
Hot Soak
Total
Grams/mile
,9
3
2
14
2.8
10-Mile Trip
9
6
2

-------
                                                         Technical Methods for Analyzing Pricing Measures
                                                                    to Reduce Transportation Emissions
      Nitrogen Oxides
      Nitrogen oxide (NOX) is an important pollutant in the generation of ozone and in some cases of
      PM.  Although NQX emissions are a product of gasoline-powered engines for vehicles, diesel
      engines are the most significant producers of NOX emissions. Diesel engines power heavy-duty
      trucks and also a variety of off-road sources, including rail, air, and marine transportation. Buses
      are also powered by ,diesel engines, and contribute heavily to NOX emissions. HDD  trucks
      contribute as much as 35 percent to 50 percent of regional mobile source NOX emissions and 10
      percent to 25 percent of total regional NOX.7

      While NOX emissions from passenger vehicles .have been  substantially  curtailed by  catalytic
      converters,  commercial diesel  vehicles are still  significant NOX emitters.  Given progress in
      reducing VOCs for the 1996 15 percent reduction demonstration, NOX has now become a problem
      focusjn many areas.                                     ~           -               .        .
                                            \    ,       ""''•.'••
      NOX emissions obey different travel relationships than .VOCs.  NOX emissions are high at speeds
      under 10 mph, fall slightly to speeds of about 20-25 mph, rise slowly to speeds of about 45 mph
      and then increase rapidly with speeds above 45 mph. Hence, strategies that  relieve congestion and
      improve  operating speeds to over 10 mph are generally effective in reducing CO, VOCs and NOX
      emissions;.programs that increase speeds above  25 mph will reduce CO and VOCs but may
      increase NOX emissions (depending on pre-program speeds); and speeds over 45 mph result in
      substantial NOX emissions, though VOCs and CO emissions continue to decline until speeds of 55
      mph are reached. Current models  indicate that PM emissions are not much  affected by speed,
      except that higher speeds lead, to more travel and therefore more PM.

      Selection of an emissions-reduction strategy based on travel patterns and speed must accommodate
      divergent effects across pollutant  types.  Moreover, the  ratio of NOX to VOC  reductions is
      important.  As the ratio between a region's VOCs and NOX increases, the appropriate strategy for
      controlling ozone changes. At a VOC/NOX ratio of 10 or less, for example, VOC controls are
     more effective, but at VOC/NOX ratios of 20 or more, NOX control is more effective. In between
     these  extremes,  NOX and VOC policies are  more comparable in their  effectiveness.  These
     variations also must be taken into account in program design and evaluation of program impacts.
     7 "Air Quality Issues in Intercity Freight," see footnote 1.
U. S. Environmental Protection Agency
                                                                                           2-19

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
     Particulate Matter

     EPA estimates show that about 26.4% of PM-10 sources are due to transportation. This excludes
     miscellaneous  sources and is based on direct emissions,  but road dust kicked up  by traveling-
     vehicles is another significant source. Tailpipe emissions also create secondary PM-10 emissions
     from the oxidation of NOx and SOx to nitrates and sulfates in the atmosphere; also, some of the
     VOC compounds are also oxidized to secondary particulates.

     PM-2.5, or fine particulates, result from two major types of sources: 1) the incomplete combustion
     of fossil and biomass fuels in motor vehicles, boilers, furnaces  and open  burning sources and 2)
     gaseous emissions such as SC>2, NOx, and VOCs  from boilers and other fossil fuel combustion
     processes that transform in the atmosphere to form sulfate and nitrate particles. The resuspended
     soil  component of PM-2.5 is generally  less than 5-15%. EPA is  continually reviewing  and
     improving estimates of the emission factors and inventories of PM-2.5.

     Revisions to the particulate matter  standard were  announced on July 18,  1997.  The review of
     hundreds of peer-reviewed scientific studies,  published since the original PM-10 standards were
     established, provided evidence that significant  health effects  are associated with  exposure to
     ambient levels of fine particles (PM-2.5) allowed by the previous standard.  These new standards
     are  available   in   62FR   38652  or  at   the  following  world   wide   web  address:
     http://www.epa.gov/ttn/oarpg/rules.html.

     Pricing strategies to reduce sources of particulates would probably vary based on the sources
     within a region and various climatic factors.  In general, reductions in VMT, rather than specific
     congestion reduction strategies or  emissions fees tailored for other pollutants,  may  be more
     effective.


     2.2.4 Overall Advantages and Disadvantages of Market-Based Measures


     Consumers who have grown accustomed to subsidized prices (i.e., prices that do not reflect the full
     social and economic costs of their decisions) are likely to oppose any change in the status  quo
     unless  it can be demonstrated that they will be made better off (or not worse off) as a result of the
     change. Travelers are no different from other consumers and thus the introduction of market-based
     measures to date has been unpopular with the public and their elected representatives.
2-20                                                          .     U. S. Environmental Protection Agency

-------
                                                         Technical Methods for Analyzing Pricing Measures
                                                                     to Reduce Transportation Emissions


     Demonstrating the advantages of market-based pricing measures is critical to gaining their ultimate
     acceptance. Some important advantages of market-based measures, which may be used to increase
     support for their introduction, include the following:

               •   Applying market-based  measures to achieve program  objectives  (in contrast to
                   regulatory actions that compel governments or individuals to assume particular
                   courses of action), provide considerable latitude for .individual freedom of choice.

               •   By reducing driving subsidies  and shifting costs to those who are  responsible for
                ,   them, market-based  measures give  individuals more information  about their
                   transportation costs and choices.

               •   Consumers can make more rational trade-offs between how much they want to  pay
                   and how and when they want to travel under a pricing system based on true costs;
                   the  current  system of  indirect taxes and  subsidies  obscures the  information
                   necessary for such choices.                         '                      .....':

               •   Rather than  locking  in  current institutional and service  concepts, market-based
                   measures support creativity to develop new and more efficient solutions to travel  and
                   access needs currently met mainly through private vehicle travel.
               •   Market-based measures can  be applied to all types of travel,and trips, not just
                   employment-related travel.  This is in contrast to TGMs, which.have been criticized
                   aS having a  limited (and perhaps inequitable) impact because they affect  only a
                   portion of travel activity.

       -        •   Market-based measures are typically self-financing, and in fact may provide revenue
                   to support their implementation, as well as, other transportation improvements.
               •   Market-based measures can substantially reduce the cost of transportation (both
                   direct and indirect). Reduced vehicle dependency can result in improved travel time
                   and reliability in movements  of people  and goods,  reduced construction  and
                   maintenance costs, lower taxes, and fewer accidents in _ addition to lower health
                   costs, less environmental damage and clean  up across  a range of  media:  water,,
                   climate, noise, solid and hazardous wastes, and habitat.                         •
               •   Market-based measures can help states and local areas delay or offset the need for
                   expensive new transportation capacity, help finance that capacity when it is needed,
                   and result in more efficient usage. They also can help cut costs for maintenance and
                   government services.

               •   Longer-term, structural changes induced by market-based measures are critical to
                   the sustainability of transportation and air quality plans, and can help areas avoid
                   recurring updates of their SIPs in search of new and stronger controls to offset VMT
                   growth.      •  .          '.-••'
U.S. Environmental Protection Agency '                     ','..-'      "                      2-21

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
      Market-based measures are not, of course, without their shortcomings.  Perceived disadvantages
      include:

                •   Political resistance, as noted above, to any form of direct transportation pricing,
                   elimination of subsidies, or imposition of new taxes.

                •   Technological and administrative challenges and uncertainties in implementation.

                •   Concerns about fairness and equity, including, for example, whether travelers have
                   reasonable alternatives and whether they impose increased burdens on the poor or
                   other disadvantaged groups.

      These concerns are important to consider in program design, and should not be dismissed by local
      decision-makers.  Political  consensus on complementing policy changes at the local level will be
      required.  This document will allow decision-makers to evaluate  the  costs  and benefits of
      alternative market-based policy options. (Issues of implementation and practical acceptance are
      discussed in Chapter 4).8                                                  "   '
       Issues related to public involvement,  public acceptance, and equity  are also discussed in:  USEPA,
       Opportunities to Improve Air Quality through  Transportation Pricing Programs (EPA  420-R-97-004),
       September 1997.  This  document  .is  also  available  at the following world  wide  web  address:
       http://www.epa.gov/OMSWWW/gopher/Market/pricing.pdf
2-22
                                                                    U. S. Environmental Protection Agency

-------
                                                         Technical Methods for Analyzing Pricing Measures
                                                                    to Reduce Transportation Emissions
     2.3   Evaluation of Market-Based Measures


     Real-life examples that demonstrate how pricing can influence transportation choices exist: transit
     ridership clearly fluctuates with fare changes; bridge and tunnel usage demonstrate sensitivity to
     toll changes; HOV. usage responds to exclusion from tolls; and automobile owners respond to fuel
     price levels by changing their driving behavior.  Currently, planners and decision-makers do not
     have system-wide implementations of pricing initiatives at their disposal (such as those discussed
     in this guidance document) to furnish detailed empirical evidence of their effectiveness and related
     .impacts. DOT's Value Pricing Pilot Program (formerly the Congestion Pricing Pilot Program) is
     generating valuable experience and research into the potential effectiveness of pricing policies.
     For the immediate future, however, estimation of the impacts of these measures will likely come
     from simulation tools and other analytical approaches described in this section.  There also are
     examples from other countries upon which evaluations can be based.


     2.3.1  Analytical Process


     Most major metropolitan areas •perform transportation planning analysis through some variation of
     the "four step process," which derives its name from the  four basic analysis modules through
     which  it assesses travel  demand:  Trip Generation, Trip Distribution,  Modal Split, and Traffic
     Assignment. This process, which is pictured in Figure 2.1, methodically translates,spatial activity
     patterns into trip flows, which are then represented on a transportation network to reflect specific
    , travel volumes on facilities, with respective speeds and levels of congestion.

     Supporting these major elements are important " upstream" functions^  such as land use/activity
     allocations,  detailing  of the transportation  network, and  vehicle ownership forecasts,  which
     provide essential starting inputs to the process. Although these sub- functions may not be thought
     of as demand estimation procedures, they have an  important role  in  profiling travel  demand.
     Allocation of trips to peak and off-peak time of day is also an important function.

     Emissions are calculated as an end product of this process.  Analysts extract information on
     vehicle trips, miles of travel and speeds from the transportation model and link it with information
     on the regional vehicle fleet mix and emissions characteristics.  This typically is accomplished
     through application of EPA's MOBILE model (version 5a), or in  California, the EMFAC model
     developed by the California Air Resources Board (CARS).  PM-i-o emissions-are calculated using
     EPA's PART 5 Model.                    ,

     This section describes  the steps in the analysis hierarchy and issues associated with these analyses.
     This serves as a prelude to the subsequent discussion  on analysis  of specific market-based
     measures.
U. S. Environmental Protection Agency                                     •                         2-23

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
               Figure 2.1 Schematic Overview of Four-Step Planning Process
      Precursor Activities
        Regional Growth
            Forecasts
          Land Use
          Allocations
        Vehicle Ownership
                                                         Four-Step
-
-
1 wooes? ^ , ^^
Trip Generation
v^5
* x ' "'
1 r,/, "
/- ~ >^ '
Trip Distribution
^ ^^^ "^
' 1
1' , ("-
f
Mode Split
•*- " r, - -
\ y ,(
V ^
/v
-'r
//
•& *
ft-'
/, 1
- 1
-W - i
' '' •-> -.t" s'f .-.,:
~ - " ~~ ~~ ~~ ~~ ^ ^^^
.; Time of Day/ -tr ^-
Peaking - *-'
i ° , , x«««»
/ f
' \.
~~ -> ^
%•'

r .
' " * 1, '
Traffic Assignment
- x-,!~
1
.
r
Emissions
>„%; *^
V
/• s
'**), ' -
\ « ^ V
//
/
"> '.-
^"'^"
-•*1
^
'J^
j1
^^
^ 'H'«J
-
^ ^ ^Li!
s 5- j
- ^'

2-24
                                                                U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                                     to Reduce Transportation Emissions
      2.3.1,1  Steps in the Analysis Hierarchy
     Nine steps comprise the analysis hierarchy, including three precursor activities (regional growth
     forecasts, activity allocations, vehicle ownership), the components of the " four step process" (Trip
     Generation, Trip Distribution, Modal Split, and Traffic Assignment), as well as its implied time of
     day/peaking component, and, last, the emissions evaluation.


     Regional Growth Forecasts
     The regional planning process begins with forecasts of population and economic growth. Usually,
     economic growth, stated in terms of basic^ employment, is estimated first, followed by estimation
     of the population  growth that would be  stimulated by those jobs.  Then, population-serving
     employment and attendant population increases  are estimated.   Frequently,  these estimates are
     keyed to federal and state control totals.


     Activity Allocations

     Once jobs arid population information is compiled, they are located  spatially across the  region,
     based  on development trends, master plans, incomes/economic criteria, or a variety of other
     factors. Specific allocations of jobs and households are made to geographic units, generally traffic
     analysis  zones that loosely conform to  groupings  of census tracts.  This process may involve
     analytic techniques (so-called "land use models"  which reflect transportation access  and cost
     issues), but most commonly are the result of an iterative negotiating process.                     /


     Vehicle Ownership

     Travel activity forecasts are driven by the economic resources available to households*  Most
     travel models represent this through the nuniber of vehicles owned by a household, which are then
     used throughout the process.. Some model systems rely directly on income  figures^ and do  not
     estimate  vehicle  ownership.  Vehicle ownership  estimates  are typically 'done through simplistic
     cross-classification models,  reflecting differences by  income, location arid other socioeconomic
     characteristics. The estimates usually ignore land use and alternative mode availability as explicit
     factors.                                                          ,    ,,


     Trip Generation                                  .

     From the information  on population and employment activity,  actual trip making  is estimated
     .through trip generation analysis  techniques.  Factors are  applied to the population/economic
     activity levels to estimate "productions" and "attractions," where households are  assumed to
     produce  trips and  employment and commercial activity are assumed to  attract trips.  These
     estimates  of household trip  making are generally cast in terms  of person trips, though in most
     models these are motorized vehicle person trips  only  and exclude walk/bike trips. Some model
     systems are even more limited and forecast only private vehicle trips. Trip estimates also  are
     developed for each major trip purpose (work, shopping, personal business and social/recreational).
     Trip production rates are based on such: factors as household size,  income/vehicle ownership,
     location, etc.  Trip attractions are based on the type of employment activity and other factors.
     Again, land use,  density,  and transportation facilities and price/source levels are frequently  not
     entered into the forecasting process.

U. S. Environmental Protection Agency                              .                                  2-25

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions                          .                                   '


     Trip Distribution

     The next step is to match the trip productions and attractions to create connected trips for each
     zone-to-zone  (origin-destination)  pair.  This usually is  done  using a gravity model,  where
     productions and attractions for each trip purpose are matched based on the " size" of the relative
     production/attraction and inversely to the "separation" or travel difficulty between the two zones.
     This step tends to represent destination choice.                   '
     Mode Choice
     The next step in the process projects a traveler's choice of desired travel mode for the given trip.
     Generally, this entails choices among automobilej automobile passenger, and transit, since trips
     which would be candidates for walking or bicycle have been separated out at the trip generation
     stage. A small number of models do include non-motorized modes among the choice options.
     Typically, this assessment is  made for each trip purpose and each zone pair  using a statistical
     model that considers the relative attractiveness of each alternative, expressed in terms of its travel
     time and cost, and weighted by the characteristics (income, vehicle ownership, etc.) of the traveler,
     and estimates the probability  with which each mode is likely to be chosen. Some models relate
     mode choice and destination choice, but this is not common at present.


     Time of Day
     Most travel models attempt to account for the time of day that travel occurs, which is obviously
     critical to portraying peak versus off-peak travel conditions.  However, rather than treat this as an
     active travel choice, the models handle this phenomenon through peaking factors.  These factors
     simply proportion a certain percentage of trips into the peak and off-peak periods, based on survey
     observations.  A separate travel network  and assignment is necessary to reflect peak and off-peak
     conditions.  It should be noted that a fair number of areas do not have a peak (i.e., a "constrained"
     flow) network, and work only with 24-hour conditions.
     Traffic Assignment
     Once blocked into time-of-day groups, trips are then placed into the actual transportation network
     using a statistical procedure that assigns them to routes that represent the most desirable path to
     their chosen destination (generally the " least time" path). This simulates the choice of travel route
     by the traveler, although like time of day, it is a decision that the  process "makes  for them"
     through a procedure that shifts trips around until the resulting volumes on the links best represent
     observed traffic levels. Traffic assignment also is important as the part of the process where level
     of service and congestion is determined, from which travel speeds are calculated for use elsewhere
     in the process.

     The realism of the forecasts from this multi-step procedure depends on the extent to which the
     various steps are  operating  with the same  information.   Thus, a  fair amount "of  cycling of
     information among the steps' is required.  It is necessary to pass information from later  steps (e.g.,
     travel times as calculated during traffic assignment) to earlier steps (e.g., trip distribution or modal
     split) as inputs to the calculations (i.e., traveler de,cisionmaking) at those stages,. Thus, preceding
     steps contribute inputs to succeeding steps, with the process continuing  until  a defined state of
2-26                                                                If. S. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                                                          .to Reduce Transportation Emissions

      equilibrium is reached.  Model systems vary greatly in the degree and sophistication with which
      they "feedback" information among steps and reach a satisfactory equilibrium.9


      Emissions Estimation
      This step is not part of the conventional transportation  modeling approach, but is necessary when
      estimates of vehicle emissions are required for air quality analysis.  Typically, this  procedure is
      performed through a special model,  such as MOBILE or EMFAC,  which takes information on
      vehicle trips, vehicle miles of travel, and speeds, by facility type, combines it with information on
      the composition of the local vehicle fleet and its emissions characteristics, and generates estimates
      of emissions by type of pollutant. MOBILE is  used to estimate VOC, GO; and NOX emissions,
      while a similar model, PART5, is used to estimate PMiQ.  EMFAC also  includes PMio for tail
      pipe, brake wear, and  tire wear emissions, while PARTS is used for re-entrained dust emissions.
      Future. EMFAC versions may incorporate re-entrained dust  emissions (It should be noted that
      MOBILE produces emissionjorctor estimates rather than estimates of emissions amounts).

      2.3.1.2 Issues

      Two sets of issues  are raised concerning, the suitability of these analytic tools  in performing
      evaluations of market-based mechanisms. The first issue is that the four-step process has various
      limitations as a tool for examining the impacts of market-based approaches in general, and pricing
      measures in particular.  The second issue is that the analytic capabilities of candidate regions and
      agencies vary widely.
                       N                        •                    '                 i

      It is generally acknowledged that the four-step process was not developed as a tool for performing
      sensitive policy  testing.  It was originally developed and has been primarily used for generating
      vehicle traffic volumes for highway planning and sizing purposes. Over time, enhancements have
      been made that increase its capabilities and sophistication, but  a number of shortcomings still exist
      that raise concerns when examining pricing actions:

           •   Pricing is often not rigorously included in the model's structure. Typically, the only
            .   choice dimension that is sensitive to cost is mode choice.

           •   The accuracy of the pricing relationship  in the models is  challenged  by the heavy
               emphasis on travel time in model development, and the relative lack of good variation in
               pricing conditions in  most U.S. urbanized areas when compiling input data,

           • ;  It is expected that travelers will respond differently to pricing actions depending on their
               income status, although this is not explicitly accounted for in most models

           •   Based on a  structure that approximates behavior through  zonal averages, and  the
               specification of the models themselves, it often is necessary to restrict consideration of a
               measure to the most basic definition.
      Work is underway in several regions under the U.S. DOT/FH WA's Travel Model Improvement Program to
      develop a new generation of. travel  models  that more  effectively integrate these steps,  considering
      individual activity scheduling, the changing of trips ipto tours, and the potential for trip sequencing and
      time of day of travel, as well as recognizing the effects of transportation on growth patterns. Many of these
      improvements are already available to local agencies through the Texas Transportation Institute.
U. S. Environmental Protection Agency                                                 '    _          2-27

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
            •  Commercial and "through" (external-external) travel is generally not addressed in the
               application of policy actions, especially pricing.

            •  Whereas  pricing actions would  be expected to  have a pervasive effect  on travel
               behavior, ranging from  changes  in destination,  route, time of day, or even location
               choice, most models consider price  only in  relation  to  mode choice.   Also,  these
               decisions would be expected to have some interdependency, whereas the models usually
               represent behavior as a sequential-step process.

      Added to these inherent limitations, there is the further complication that overall capability varies
      widely by location.  The model systems of areas like Portland, Oregon or San Francisco are fairly
      progressive and have many of the features that would raise confidence in their  use for pricing
      measure analysis.  However, there are many other major metropolitan areas whose models are
      fairly rudimentary or which have acknowledged flaws or limitations.  Added to this are a,number
      of areas that might choose to look at pricing measures, but which have little or no modeling
      capability at all.  These might be counties or urban areas that are too small to have sophisticated
      transportation planning capabilities. The planning process for many of these areas, for example,
      may be only a "three-step" process (excluding mode choice) or less.

      This  guidance document has  been developed with appreciation for  the  demand that pricing
      measures place on the analytic  process and the wide range of analytic capabilities that exist.  It is
      EPA's desire that areas that need additional help in reaching their attainment  goals  consider
      market-based measures.  In this regard, the candidate area should endeavor to perform the most
      accurate and realistic assessment possible, within its capabilities, when  submitting such measures
      for emissions credit. EPA's decision as to how much credit to grant will be based on:  (1) whether
      the analysis supports the requested credit; and (2) whether the steps leading to implementation are
      realistic.  Obviously, the  greater the significance and contribution of  the measure to the area's
      attainment goals, the more carefully the analysis procedures must be reviewed when deciding to
      grant credit. However, EPA will work cooperatively with the performing agency to support its
      efforts and help achieve an accurate determination of credit.
     2.3.2 Identifying and Assessing Areas of Impact


     This section presents how each measure is expected to affect travel and emissions.  Discussion is
     organized by area of impact, presented in the same order as the analysis hierarchy.  This section
     also describes the analytical tools that may be used to assess those impacts. -

     Table 2-4 lists each of the five market mechanisms, as well as an assessment of where in the travel
     demand hierarchy they would be  likely  to have  impact.  Regional growth forecasts  have been
     excluded as a step, since their value is almost always exogenously determined.  Vehicle ownership
     appears in two ways:  (1) conventional vehicle ownership, which represents the number of vehicles
     owned by households as a factor in travel decisionmaking for many model systems; and (2) vehicle
     mix (shown as a  component of the emissions generation step), which represents the  types of
     vehicles owned and their composition in the vehicle fleet mix.

     Table 2-4 indicates where a pricing measure would  have an impact in this structure,  and the
     expected level of importance of that impact. Impact importance is rated using the following scale:
2-28                                                               U.S. Environmental Protection Agency

-------
                                                         Technical Methods for Analyzing Pricing Measures
                                                     •               to Reduce Transportation Emissions


               *   Primary:  The impact is major and basic to the measure; every effort'should be
                   made to evaluate its effects.

               •   Secondary: The impact is judged to be an important: effect which may even exceed
                   the primary effect and thus must be considered; however, while its formal analytical
                   evaluation (i.e., a complete regional travel demand modeling analysis) may not be
                   possible, an area  should provide documentation  to show that the effect has been
                   considered  in the most reasonable manner possible.

               •   Trace:  Some impact would be expected, but its  implications for emissions would
                   be minimal and no evaluation would be required.

               •   None:  For practical purposes, no impact would be expected.

      A draft version of the impact hierarchy outlined above was originally developed as the result of a
      meeting that included EPA and DOT staff and contractors involved in the production of this report.
      Subsequently, the draft hierarchy was reviewed by EPA and FHWA staff and modified. EPA and
      FHWA review and modification led to the hierarchy outlined above.

      Table 2-4 should be used with several caveats in mind. First of all, level of importance valuations
      in the table are not linked to time periods. For example, while the land use impacts are often, long-
      term the distinctions among primary, secondary and trace effects are generally linked to the
      ultimate level of impact expected, rather than how long it takes for that impact to fully materialize.
U. S. Environmental Protection Agency
                                                                                            2-29

-------
I













.c*
^M
2
o
if
CO
• MM
(0
"to
c
^f

.^
(0
3
to
re
«E
*5
4-»
O
(0
Q.
E
73
0)
•»-*
(0
Q.
'c
to
3
(0
1-


'<


LI, , ,„

;,:- 	 '




0
1
o
£
1
s
























en
"o
•c
a.
SK
1
i


c
0
1
tu ix

*~«
to <»



^
*s
jg
zs
CO
i
~=


nil
o
£
ra
JC
1


«t-«
Areaofimpac









^,
•a
£=
O
O
CD
CO


..^
CO
"0
o
o
CD


CO
•a
1
CD
CO








1
t—



CO
•a
8
QJ
CO



Land Use/Activity
Allocations









^,
CO
•D
1
CD
CO




8-
r


CD
h^








CD
O
s
1—



CO
•a
1
CO


Q-
Vehicie Ownershil



8
CO
cb"
•a
3
CO
CD
JE *
Stn CD
||
D~ "o




1


CO
•a
CO








*

-
Q










.1
£




CD
0


CD








1






CD
O



 CO
^ §> ^ .1 -CO-
CD c 0 W ™
£ co ^ .22 S
_CD
L^jt

"1
c5 §>
Hi








CD
O






CD
O
,
^ '
X --
S ' -
*o
i
.2
"25
u>
uu






















^
o
S
"Q
M
O
S
i
§
a
1
K
8
-§
K
•c
&.
K
&



                                                                                                                                                                                          .§
                                                                                                                                                                                          1*3
                                                                                                                                                                                         o

                                                                                                                                                                                           I
                                                                                                                                                                                         CN

-------
                                                           Technical Methods for Analyzing Pricing Measures
                        !                                              to Reduce Transportation Emissions

       Second, level of importance valuations do not account for major differences  in price levels. • For
       any of the measures to have an effect on behavior, the level of the price applied would have to, be
    .   sufficient to be felt.  At low price levels, all of the instruments would be expected to function
       primarily as revenue generators, and have only marginal effects on travel behavior.  However, the
       relative level of impact implied by the importance valuations would be expected to stand across a
       wide range of price levels.

       The analytical tools or approaches  described in this section are categorized as basic analysis or
       advanced analysis, according to how they are rated by EPA submission requirements:

                •   Basic Analysis:  Procedures  that EPA suggests be used when formal analysis is
                    required if the measure is to  be considered for emissions credit (that is,  when the
                    measure is expected to have  a primary and direct  impact or a significant indirect
                    effect).      '                                        ..-.'.•'••

                •   Advanced Analysis:  Suggested procedures beyond the basic analytical procedures
                    that EPA believes would result in better estimates of the travel and emissions impact
                    of the measure if applied.  Use of Advanced Analysis Techniques will enhance the
                    acceptability of an application for SIP credits.                 L.

       In many cases, the same analytical technique is considered basic in one circumstance (e.g., when
       the measure will have a primary impact),  but advanced in another analytical context (e.g., when the
       measure will have a secondary or trace impact); these techniques are thus categorized as "basic or
       advanced analysis:''  This consolidated discussion of analytical techniques  helps avoid undue;
       repetition  in  Section  2.3.3,  below,  on analytical requirements for  individual  market-based
       measures.            -                                       ,     •    "


       2.3.2.1 Land Use/Activity Allocations

       Some pricing measures can affect long-term locational decisions  and land use patterns and both
       short and long-term individual decisions.  These measures might either induce more concentrated
       growth patterns that might be favorable for longer term transportation and emissions management,
       or could produce  divergent trends  toward  dispersed land use patterns  and greater automobile
       dependence.       ,                                                '


       Effects of Market-Based Measures on Land Use/Activity Allocations
       Much depends on the  level and type of cost impact of the particular measure, but most of the
       measures could conceivably have an impact  on this underlying aspect of travel behavior.  Despite
       the potential impact of a change in land use/locational choice, these impacts are listed as secondary
       in Table 2-4 because of the uncertainty associated with  projecting these changes with current
       modeling tools.  -                                               -

       Parking pricing would be expected  to  induce  a  locational effect, with greater preferences for
       locations  that  provide more  concentrated amenities and  services and better transportation
       alternatives.   If applied differentially, businesses  and households might  be expected to show
   .    preferences for locations with lower parking rates.  .For example, should the policy be to raise
       downtown rates relative to suburban rates, this would be expected to increase the attractiveness of
       suburban development. If .parking were implemented uniformly, and resulted in some areas 'having


. U. S, Environmental Protection Agency               .                    •        .'•        ,          2-31

-------
     better design/amenity or offering better access through non-SOV modes, the measure also might
     induce locational shifts to these areas.


     Modal subsidies would be expected to have only a trace effect on land use decisions, unless they
     were of significant size and the public  perceived them to be permanent (in availability and  in
     relation to the cost of other modes).  At high levels, dedicated subsidies might shift locational
     preferences to areas where these price advantages could be exploited, that is, where transit or- other
     subsidized modes are more available and  useful in reaching desired destinations.

     Pump charges would be expected to  have  a concentrating (secondary) effect on land use,  since
     they function to visibly raise the cost of driving for all travel purposes.  Thus, households might be
     expected to locate in areas where less dependence on private vehicle travel v/as'required.

     Emissions fees, if scaled primarily to emissions output and not VMT, would have little more than a
     trace effect on locational decisions. However, an emissions fee that is substantial, or based heavily
     on VMT, could encourage households to own fewer vehicles and thus prefer areas with mixed land
     uses and which otherwise provide better access to desirable locations without private vehicles.

     Roadway pricing might have an important (secondary) effect.  Widespread pricing would tend to
     encourage  locating where there is  less  reliance on private vehicles.  Assuming  only  selected
     facilities would be  priced, households  and businesses would  be  expected to locate  so as to
     minimize long-term reliance on these facilities. The impact of this on  air quality is uncertain. Area-
     wide implementations of road pricing would tend to reduce VMT and encourage a greater mixing
     of uses. Congestion  pricing would be limited in influencing location  choice, particularly if fees
     were peak-only.  To  the extent that congestion pricing leads to travel time decreases, it could lead
     to more dispersed new development, could  allow better access to currently congested centralized
     developments, or both. Area cordon fees  may lead to more transit-oriented development, provided
     the priced area remains attractive relative to other locations.
     Analytical Techniques for Land Use/Activity Allocations                             ,

     Advanced Analysis
     Two approaches are generally applicable for evaluating the impacts on land use/activity allocation
     of market-based measures:

               •  Estimate impacts using, a " land use model" that is sensitive to transportation costs
                  (through incorporation of" generalized cost," for example);10 or
               •  Develop one or more growth scenarios to depict the alternate ways in which regional
                  growth might respond to changes in transportation costs of the type proposed.  This
                  might be done in conjunction with  application of land  use  models, and involve
     10
      'The DRAM/EMPAL model, for example, developed by University of Pennsylvania's dr. Stephen Putman,
      is commonly used in the United States'to model land use changes.
2-32                                                                [/ s. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
        .•''.-'.      .                                  •        to Reduce Transportation Emissions

                   development and  integration  of alternative land use/urban design configurations
                   supportive of alternatives (e.g., subsidized) modes.1"

     Because land use/activity allocation effects are identified as secondary impacts, no formal analysis
     is called for, although a statement of anticipated impacts should be included in most cases*12


     2.3.2.2  Vehicle Ownership                        '.  .


     Effects of Market-Based Measures on Vehicle Ownership and Vehicle Mix

     Pump charges would encourage  drivers to  shift to owning and/or using more  fuel-efficient
     vehicles.  This would have an important primary effect on  vehicle types, which should be captured
     in the emissions modeling step, in terms of the fleet mix used to calculate regional emissions
     factors. These more efficient vehicles might or might not be lower-emitting.

     '                          ' "          •           *  '   *
     Pump charges that serve as a medium for pay-as-you-go insurance might have an effect on vehicle
     ownership, although the direction  of the  effect would be hard to predict.   Such charges would
     significantly boost the cost of vehicle ownership and use for groups that do not carry insurance (a
     legal violation in most cases, acting to discourage ownership).  Pay-as-you-go charges, however,
     may be more economical for owners currently insured and could encourage ownership.

     Emissions fees, if significant enough, could affect the number of vehicles owned (hence a primary
     impact).  Emissions fees would likely  cause  households to shift ownership to (and possibly use)
     newer and/or lower-emitting vehicles. This would be an important primary effect that would need
     to be reflected in the emissions modeling step through adjustments to the fleet mix.
       An outstanding example of this approach is Portland, Oregon's "Region 2040" planning process, and a
      closely related analysis process by 1000 Friends of Oregon, the "Land Use, Transportation, and Air Quality
      Connection" (LUTRAQ). Region 2040  information is available  from Portland Metro, and LUTRAQ
      information from  1000  Friends of Oregon  See, for example,  1000 Friends of Oregon, "Making the
      Connections: A Summary of the LUTRAQ Project, February 1997.        ;
      2 In a relevant recent case, a US District court ruled that as part of planning for a major new highway, Illinois
      DOT would need to. develop a separate set of socioeconomic/land use forecasts based on what would likely
      occur with vs. without construction of the road. A single set of socioeconomic/land use assumptions for both
      build and no-build scenarios for evaluation of ozone precursors was deemed to fail the legal requirements of
      the National Environmental Policy Act. US District Court (Northern District of Illinois, Eastern Division),
      Case No. 96 C 4768. Sierra Club, Illinois Chapter; South Corridor Against the Tollway, Inc.; Environmental
      Law and Policy Center of the Midwest; and Business and Professional People for the Public Interest vs. US
      DOT/FHWA, Illinois FHWA, Illinois DOT, and Illinois State Toll Highway Authority.
U. S. Environmental Protection Agency                              '                                  2-33

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions

     No other market-based measures would be expected to have a major direct effect on the number of
     vehicles owned by households.  The primary effect on vehicle ownership levels for these other
     measures would be a result  of induced shifts in location (land use/activity allocations) to areas
     where fewer vehicles were required.

     Analytical Techniques for Vehicle Ownership and Vehicle Mix

     Basic or Advanced Analysis
     Analytical tools used for evaluating ownership patterns apply primarily to vehicle mix because that
     is the factor expected to be most affected by market-based measures. Approaches include:

               •   Analytic procedures such as hedonic models to forecast shifts in ownership patterns;
               •   Research studies on shifts in vehicle ownership in response to price, and application
                   of factor methods to estimate shifts; and

               •   Special analyses, such as stated preference surveys, to investigate likely responses to
                   ownership that might result from alternative fee structures.

     2.3.2.3  Trip  Generation

     Market-based  measures can have a range  of impacts on the generation of trips from households.
     Probably the key factor in terms of the impact on trip generation is the overall scope of the measure
     and  its comprehensiveness in" increasing  the overall  cost of travel by single-occupant vehicles.
     Ideally, trip generation should account for all person-trips.  Non-motorized trips are generally not
     accounted for in current modeling systems, which could constitute an  important limitation when
     evaluating pricing systems.


     Effects of Market-Based Measures on Trip Generation
     Many  analysts assume that  the number  of trips generated  by a household  is not particularly
     sensitive to transportation factors, including cost.  They suggest that a household's  propensity to
     travel  is determined by its need to satisfy specific activities,  which are dictated  by its size  and
     socioeconomic composition.  Rather than forego activities and make fewer trips in response to
     higher prices for vehicle travel, households would be expected to make less severe changes in their
     pattern of trip making — for example, more trips to closer/different destinations or, shift to other
     modes. These are important effects, but are generally dealt with elsewhere  in the travel analysis
     hierarchy.

     There  are growing challenges to this view, however, particularly  in light of new information  and
     telecommunications systems  that are reshaping our economy and communities. For example, an
     increasing number of workers are  opting to work  at home under telecommuting arrangements;
     many others have shifted to a modified work schedule. In these cases, the number of trips made to
     a physical work site is reduced and/or the time of travel changed, suggesting  that the household
     might make fewer total trips as a result.

     Similarly, shopping and entertainment needs are being increasingly satisfied through cable TV, the
     Internet, video movie rentals, deliveries, etc. There also are other ways in which households could
     shift their activity patterns to  group  trips together into trip "chains," possibly accomplishing
                                        j                   ,                            • •       '

2*34                                                               -U. S. Environmental Protection Agency

-------
                                                         - Tecfiniqai Methods for Analyzing Pricing Measures
                                                                      to Reduce Transportation Emissions
      multiple functions in a single trip. Such adjustments to  activity patterns could show up in overall
      trip rates in response to higher travel costs, although their impact is presently not well understood.

      Parking pricing could influence trip generation in many different ways and depends heavily on the
      details of the pricing changes. For example,.requiring employers to charge employees for parking
      (or  instituting a "cash-out" program)  could have primary impacts on work trips but, probably,
      virtually no impacts on non-work trips. If non:work parking were priced, one could expect far
      larger impacts, though their character  might differ. Impacts  of non-work parking pricing might
      emphasize long-term  land use changes by retail  locations.  Since land use effects are captured
      elsewhere  in .the modeling system, trip generation would probably only have a trace impact for
      non-work trips.            "                 '•''".

      Modal subsidies would only have a trace impact on trip generation.  This would probably tend to
      affect work travel more than non-work travel. Minor increases in carpooling and transit use would
      be likely.    •                               '

      Pump charges could have a secondary impact on trip generation.  In the short run, this would be
      reflected by better scheduling of activities and vehicle usage within households (often a household
      with more than one vehicle will increase use of the more fuel efficient vehicle).

      Roadway pricing could have a primary impact on trip generation if applied area-wide.  An increase
      in the  overall cost of travel would be expected  to  result in  household activities  being more
      efficiently  scheduled  and coordinated. Limited roadway pricing, such as tolls  on individual
      facilities or limited congestion pricing would only be expected to have a trace impact.

      Emissions fees are not expected to have more than a trace impact on trip generation. Note that if
      these fees are mileage-based, arid if the  mileage-based component of fees is significant, they could
      have a greater effect.


      Analytical Techniques for Trip Generation

      Basic Analysis           .                                      .                   .
      Aside from uncertainty regarding the actual impacts, practical concerns  have been raised regarding
      the modeling conventions used to assess impacts. Frequently, trip generation models estimate only,.
      those household person trips that may be served by vehicle. They do not include trips that would
      be made by walking or bicycle.  Therefore, in situations where a household  may respond to a
      change in travel conditions by shifting some trips to pedestrian or bike  travel, this shift would not
      be captured accurately in the trip generation analytical step.                             .

      Some regions employ models that are  even more restrictive;  they consider only trips made in
      vehicles, not person trips,  typically, these are smaller areas where  the  private vehicle  is the
      overwhelmingly dominant mode and the planning process is less complex in that they are not
      concerned with transit or non-motorized alternatives.  Thus, if a  region's  trip generation  model
      estimates all person trips, the effect of pricing on trip generation would be expected to be at a trace
      level, since allowance would be made for vehicle trips to shift to non-motorized modes.

     However, if trig .generation analysis is based only on vehicle or vehicle-eligible trips, then pricing
     would be expected  to show at least a secondaryeffect, at least in  the case of pump charges, but
U.S. Environmental Protection Agency                    •                               '   ,          2-35

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions

     possibly also for parking pricing and modal subsidies, because vehicle trips would shift to non-
     motorized modes would " disappear" from the inventory.

     Advanced Analysis
     In general, if the  trip  generation analysis deals only  with motorized-vehicle  trips, then an
     adjustment should  be made to the vehicle trip  generation  estimates  to reflect  "shifts to non-
     motorized or non-automobile-driver options. Wherever possible, trip generation analysis should be
     performed  separately for each purpose  and income strata, to support subsequent analysis of trip
     distribution and mode choice.

     Moreover,  if land use analysis suggests shifts in activity or land use patterns,  then  the vehicle
     ownership  and trip  generation analytical steps should be repeated to reflect relocation of activity
     and its association with potentially different rates of trip making and vehicle ownership.

     2.3.2.4 Trip Distribution


     Effects of Market-Based Measures on Trip Distribution
     All market-based measures^  except emissions fees, would  be expected to influence choice of
     destination, that is, trip distribution. As noted earlier, if emissions fees are mileage-based, and if the
     mileage-based component of these fees  is significant, they could have a greater effect. The same
     caveat applies to all references to the effects of emissions fees throughout this section. Parking
     pricing would be expected to influence destination choice at a secondary level. Increased parking
     rates would increase the  attraction to destinations that provide better access to desirable locations
     without private vehicles. If parking rates were  uneven in their application, travelers would be
     attracted to comparable destinations with zero or lower parking rates, unless the priced area also
     offered better access, more services or a higher level of amenities.
 I                                     '              /'            _-,•"'
     Modal subsidies would be expected to  have a secondary effect. Lowered costs of selected non-
     SOV  modes  (through subsidies)  would  be expected to increase  the  attractiveness  of those
     destinations accessed by subsidized modes.

     Pump charges would have a secondary effect on destination choice. By raising the comparative
     costs  of private vehicle travel, this measure would be expected to increase  the  attractiveness of
     closer destinations and/or destinations easily reached by alternative modes.
                                                                   U. S. Environmental Protection Agency

-------
                                                          •••^••^•^^•^•^•^^••1
                                                           Technical Methods for Analyzing Pricing Measures
                                                                       to Reduce Transportation Emissions
      Emissions fees could have a trace effect on choice of destination. If the mileage-based component
      of the charge were significant, then it could encourage travel patterns that would reduce VMT,
      although the imposition of this fee on a periodic/annual basis would be expected to dampen its
      effect as a signal to daily travel.            .

      Roadway pricing would be  expected to have the largest  effect of all  pricing  measures  on
      destination choice, constituting & primary impact.  Prices would tend to induce consumers to reduce
      vehicle  trips, to  reduce the length of trips, and  also to  consolidate destinations. If prices were
      imposed on select facilities, travelers would be expected-to re-examine their options  (particularly
      for non-work travel) and choose to:

               •   Travel to alternative (possibly more distant) destinations not subject to pricing;
               •   Travel to  alternative  destinations served  by  priced facilities that  offer good
                   alternatives, special amenities, or are closer than  their original choices; or

               •   Travel to the same or alternative destinations, where they obtain quicker access at an
                  • acceptable price using priced facilities (more likely among higher income travelers).

      Area-wide road pricing would result in shifts to closer destinations. Isolated road-pricing systems
      could, in some cases, result in increased travel.


      Analytical Techniques for Trip Distribution

      Basic Or Advanced Analysis                                                           :  >  '  •
      Areas that have trip distribution models that incorporate price  sensitivity directly through  a
      generalized cost procedure can account for the effects of market-based measures explicitly. Many
      trip distribution models are sensitive to travel time but not cost. Some agencies address this in their
      models  by incorporating a generalized cost formula. This characterizes the full cost of travel by
      private vehicle from each origin to each destination in terms of both the travel time "cost"  and the
      monetary cost (use of travel time equivalents  is discussed in Section 3.3.5.3.).  Some models also
      consider the time and money cost of transit, where it is available,  and average it into the overall
      measure of generalized cost.                              -    .     ."

      If an agency has a model system with a generalized cost procedure that can be made to reflect the
      cost of market-based measures, this should be used to perform a revised trip distribution analysis
     that incorporates the proposed  cost measures.  Moreover, trip distribution analysis should be
     performed separately by trip purpose and income strata.
U. S. Environmental Protection Agency                                                                 2-37

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions

     2.3.2.5 Mode Choice
     Effects of Market-Based Measures on Mode Choice
     All of the pricing measures, with the exception  of emissions fees, would be expected to have a
     primary effect on the choice of travel mode.

     Parking pricing would. clearly serve to discourage driving, and, in particular, driving alone, in
     favor of other alternatives, and will shift some .shorter trips to walking or biking if suitable
     alternatives and access exist.

     Modal subsidies would create a direct attraction to the mode or modes being subsidized.

     Pump charges would encourage use of alternative modes that would allow the traveler to avoid or
     reduce the higher cost of private vehicle travel.

     Emissions fees could have & trace effect on mode choice through the mileage-based component of
     the charge.  Again, however, 'the effect would be greatest if fees are collected frequently and/or
     there are frequent education and reminder efforts.
                                                »
     Roadway pricing would encourage use of alternative modes to avoid or reduce the cost of priced
     facilities, although this   effect  would  be  in  combination  with consideration  of alternative
     destinations, routes and times of day at a much higher level than with the other measures.
     Analytical Techniques for Mode Choice

     Basic or Advanced Analysis
     With the exception of emissions fees, all market-based measures would be expected to  have a
     primary effect on mode choice.  The performing agency can conduct the basic analysis with any of
     the three following options:

               •  Accepted mode choice model;

               •  Accepted " quick response" methods; or
               •  Elasticity methods.

     An agency with access to a mode choice model is expected to use that model to  conduct any
     required analyses of the impact of market-based measures. It may be permissible to adjust the
     factors included in these models or their coefficient values, provided adequate documentation is
     supplied on the nature of and need for the adjustment.

     Areas without a mode choice model but with access,to person trip tables may consider acquiring
     one of several "quick response" methods, which are widely available. These include: the TDM
     Evaluation Model from  the  Federal  Highway Administration, the QRS planning  system,  and
     various methodologies developed to support CMAQ project analyses. Information on these tools is
     given in Section 3.3.6.9.
2-38                                                               U.S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                                     to Reduce Transportation Emissions
      If an agency does not have reasonable access to either a mode choice model or one of the above
      surrogate procedures, analysis may be based on application of elasticities from research studies or
      from the mode choice model of similar urban areas. Section 33.6.9 provides references on sources
      for such elasticities and guidance toward their proper application.

      Advanced Analysis                          •                             .     .,   .   '
      Advanced analysis of impacts on mode choice would use current generation, more sophisticated, -
      mode choice models such as "nested"  multinomial logit models, which permit testing of a wide
      array of alternative modes. The range of transportation service and socioeconomic variables that
      they incorporate offers greater flexibility in relating strategies and more realistic estimates of travel
      response. Some models have separate coefficients for categorically different cost'items, such  as
      transit fares, parking subsidies or parking costs, and HOV costs,, and should result in more" accurate
      estimates of response to corresponding policies.  Models that reflect the interaction  of mode
      choices and destination choices are desirable and EPA encourages their development.

      More realistic analysis will result from  application of separate models for each  primary trip
      purpose,  including home-based work, .home-based other,  and  non-home  based  trips.  Many
      agencies,  however, do not have mode choice models  for non-work travel.  Accuracy would be
      further enhanced by accounting for effects by income strata, as well  as trip purpose, in  order to
      reflect differences in cost sensitivity by income  level. This would also help analyze the equity-
      related issue of differential impact by income group.                                         ,

      Finally, model accuracy can increase if models can capture the attractiveness of an area to other
      travel  options, including  non-auto travel. Just as the  design of the road system  increases  or
      decreases the  " impedance" a driver experiences, so the walking and  cycling infrastructure helps
      determine whether, how often, and how far people walk or bicycle.  And just as  impedance for
      drivers can be quantified,  new work has significantly improved the quantification of determinants
      of walking  and bicycling. Recent work on Pedestrian Environment Factors and, Urban Density
      Factors has helped improve modeling in Portland, for example.13

      2.3.2.6 Time of Day


      Effects of Market-Based Measures on Time of Day
      Only three of the described market-based measures would be  expected to have any effect  at all on
     travel time of day; neither pump charges nor emissions fees  would influence time patterns since
      assessment is not linked to specific trips.

     Parking pricing could have a secondary effect on time of day if the differential between peak rates
      and off-peak rates were sufficient to shift users with flexibility to off-peak travel. In many, cases,
      however, impacts would be at a trace level.

     Modal subsidies that favor alternatives (mode and/or specific service within  that mode)  used
     primarily  in peak periods could shift users from unsubsidized modes/services used off-peak to the
     13
       See, for example, 1000 Friends of Oregon, Making the Land Use, Transportation, Air Quality Connection:
       Volume 4A, the Pedestrian Environment. 1993. Portland: 1000 Friends of Oregon.               -   '   -.
U. S. Environmental Protection Agency                           .  '  .               ,   '               2-39

-------
 Technical Methods for Analyzing Pricing Measures  •
 to Reduce Transportation Emissions

      subsidized service during peak periods.  Nonetheless, only trace effects would be expected for the
      level of subsidies generally proposed unless subsidies vary by time of day.

      Roadway pricing in the specific case of congestion pricing would be expected to have & primary
      effect on time of day choice.  Higher peak-period prices would be expected to shift lower-valued
      trips to off-peak travel and would draw some trips into the peak if a higher level of service becomes
      available as a result of the pricing program.
                                    'i                                   '                   •

      Analytical Techniques for Time of Day

      Basic or Advanced Analysis
      Time-of-day analysis  procedures are still uncommon  in transportation planning  practice. Some
      agencies have time-of-day models.  For those that do not, analysis could be based on procedures
      borrowed from another location or based on a special  analysis at the affected  site to  identify
      relationships and estimate behavioral factors.  More specifically, analysis of the impacts of market-
      based measures expected to have a measurable impact on time of day could be based on any of
     , three approaches:

               •   Use of a time of day choice model acquired from another location (see Section
                   3.3.7);

               •   Consultation of references on time of day sensitivity, which could include analyses
                   of other area programs, transit or parking peak pricing results, or other  special
                   analyses; or

               •   Performance of a  special stated preference survey analysis to develop a model for
                   the area under investigation.


      2.3.2.7 Traffic Assignment/Route Choice


      Effects of Market-Based Measures on Traffic Assignment/Route Choice
      While all of the pricing measures affect traffic assignment in terms of revised traffic patterns, only
      roadway pricing directly targets route choice.

      Roadway pricing forces travelers to make conscious decisions about spatial patterns and, as such,
      has a primary impact on traffic assignment.  Travelers may have a choice between routes that are
      tolled but that offer a higher level of service/travel time and another set of routes that are not tolled
      but that have a lower level of service/travel time. As with destinations, the choices ultimately made
      depend on the specific characteristics of the pricing scheme and the available travel alternatives.
2-40                                                               jj_ $. Environmental Protection Agency

-------
                                                          Technical Methods for Analysing Pricing Measures
                                          ,                           to Reduce Transportation Emissions


      Analytical Techniques for Traffic Assignment/Route Choice          -  .

      Basic or Advanced Analysis                         ,     ^
      Roadway pricing explicitly forces travelers to evaluate and respond to the trade-off between tolled
     . roadways that may have a higher level of service and toll-free routes'that may be more congested.
      They also may choose shorter trips or forego some trips that would previously have been taken.
      The anticipated close links between roadway pricing and traffic assignment (route choice) mandate
      analysis of these effects.  Yet, traveler route  choice, is not  directly handled by most traffic
      assignment models.  There are various options for addressing anticipated impacts, including:

               •   Use of procedures that represent toll and free path opportunities to travelers in an
                   assignment routine.  These are relatively advanced techniques, however (see Section
                   3.3.8.5); or

               •  Representation of the effect of tolls for a given facility as an increase in travel time
              ,    through use of value of time relationships to estimate the appropriate value. This
               •   raises the cost of a tolled facility to a traveler considering its use (also see Section
                 ,3.3.8.5).


     2.3.2.8 Emissions /Vehicle Mix
     Effects of Market-Based Measures on Emissions Vehicle Mix
     All market-based measures should ultimately affect the level of emissions generated. Two market-
     based measures—pump charges and emissions fees—could directly affect vehicle mix and, through
     that, emissions rates.                                                              .

     Pump charges could induce shifts in favor of vehicles with greater fuel economy. The net effect
     on emissions is the result of two factors: lower fuel consumption per mile traveled (and potentially
     overall) and possibly different emissions characteristics (rate of emissions per mile).  The overall
     relationship between pump charges and  vehicle mix/emissions  could be considered primary,
     though the indirect link of pump charges working through increased efficiency and lower total fuel
     consumption qould be considered secondary. Pump charges would induce travelers to shift to more
     efficient vehicles, partially offsetting the benefits of fu,el price increases.

     Emissions fees would  act  similarly to the pump charge, except that by inducing a shift in the
     vehicle fleet toward cleaner vehicles, it would constitute a: primary effect. Influence of a VMT
     component of emissions fees would also show up in lower aggregate mileage rates within the fleet.
U. S. Environmental Protection Agency                    '   •                                        2-4.1

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions


     Analytical Techniques for Emissions /Vehicle Mix

     Basic Analysis
     Agencies  should use an approved .emission factor model to calculate emissions,  with factors
     computed to reflect any program-induced adjustments in the regional vehicle mix. Areas without
     formal emissions factor models would be expected to explain  how the average emissions rates
     would change with the induced shifts in vehicle fleet composition and/or use.

     Basic or Advanced Analysis
     Analysis  of anticipated shifts in vehicle fleet composition or usage can utilize the following
     approaches:

               •   Use of hedonic models (models of consumer  satisfaction with  particular vehicle
                  attributes) to project changes in vehicle ownership in relation to cost differences;

               •   Consultation of research studies on shifts in vehicle ownership  in response to price,
                  and application of factor methods to estimate shifts; or

               •   Performance of  special analyses (using  revealed  or  stated  preference survey
                  methods) to forecast possible shifts in vehicle ownership and use.

     These methods are further detailed in Chapter 3, Section 3.3.3.


     2.3.3    Analysis Guidance


     This  section presents EPA's recommendations for evaluation of market-based mechanisms and
     describes the types of analyses that EPA expects to see accompany submissions of the respective
     measures for emissions credit.  The guidance descriptions in  this chapter are at a general level,
     designed to communicate to managers and decision-makers EPA's general thinking; specific step-
     by-step guidance suitable to direct technical staff in the actual evaluation practices is provided in
     Chapter 3, further supported by examples in Appendices A and B.

     The table on the  following page summarizes rules of thumb for submissions to  EPA; impacts are
     assessed for individual market-based measures at each stage  of the analysis hierarchy (refer to
     Table 2.3 for description of anticipated impacts). Clearly, the  more careful the analysis, the more
     credit EPA can be expected to approve.                      -
                                    •    '                          U. S. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                                                                      to Reduce Transportation Emissions

        ; fapectejnpact .1 * Acceptable Analysis
        ^Primary effect-^       Fom?a/ analysis expected for impact on that step of analysis hierarchy.
                             Must include some treatment oi impact on that stage of analysis hierarchy, though
                        *•  J  requirements are less formal.                      .
                        v ~   Agency expected to fate nofe of possible side effects, which could occur under special
                        --*•   applications of the measure.                              .
<-3/aceeffect   v
     In laying out its recommendations for effective analysis, EPA has taken into consideration the wide
  .   range of capabilities that are available to those considering market-based measures. Because of the
     strong  likelihood that there will  be sites  with minimal analysis  capabilities,  the guidance
     recommendations have been made flexible enough to allow these states to.respond.

     This flexibility does not take away from the position, however, that an. accurate, comprehensive
     analysis is extremely important.  Accuracy and completeness  are important to EPA's review  and
     decision to grant emissions credit. Therefore, agencies proposing pricing measures should consult
     with a wide range of parties to get as much assistance as possible in developing good analytical
     approaches.  Furthermore, an accurate analysis also is essential to the performing agency's ability
     to assess the impacts and tradeoffs that are critical to local acceptance and implementation.  EPA's
     position is to  support the best  possible  analysis, t6 make allowances  for  states  with limited
     capabilities, but to expect that states will take every reasonable step to perform the  best analysis
     within their ability.

     The remainder of this section presents  analysis recommendations for each of the five measure
     groups: parking pricing, modal subsidies, pump charges, emissions fees^ and road  pricing.  To
     avoid repetitive discussion of analytical  approaches, EPA submission components are often noted
     only briefly with reference to the more extended discussion of basic or advanced analysis organized
     by  impact category in Section 2.3.2. Recommendations for analysis are  separated  into basic
     analysis and advanced analysis and, within each lof these, according to expected level  of  impact
     from primary to less important effects.

     Advanced  analysis  is optional  from-the  standpoint  of  EPA  guidance.   This  reflects  an
     acknowledgment of current limitations, but also willingness on EPA's part to work with all areas in
     exploring advanced  strategies to  deal with  long-term air quality concerns.  Nonetheless^ EPA
     expects that those areas with advanced, planning tools will  use those tools to their best ability.
     Areas that do not have such tools or capabilities are encouraged to strive to acquire those advanced
     capabilities, both to  serve their own current needs and to support other air quality and planning
     requirements that are likely to be part.of future planning and funding tests.

     In addition to attention to analysis tools  and procedures, EPA places  importance on the  quality of
     data used to conduct the analyses, particularly with regard to data age, coverage, and accuracy.

     The level of scrutiny applied by EPA to analyses that accompany submissions of market-based
     measures will be tailored to each circumstance.  In particular, the level of credit claimed for  the
     measure and its importance in achieving a given area's attainment or maintenance goals will be a

U. S. Environmental Protection Agency                                            .                    2-43

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions

      factor considered by EPA in its review.  More rigorous scrutiny can be offset through a more
      accurate and defensible analysis.


      2.3.3.1 Parking Pricing

      Suggestions for basic analysis and advanced analysis pertaining to parking pricing are illustrated in
      Figure 2.2a.


      Basic Analysis for Parking Pricing Measures

      Mode Choice: Toe primary effect of parking pricing on travel and emissions would be through its
      impact on mode choice.  A submission that seeks emissions credit for parking pricing must  pay
      particular attention to the shift  of travelers  from  private vehicles,  especially single-occupant
      vehicles, to transit, ridesharing or non-motorized alternatives.

      Given the importance of parking pricing on mode choice, submitting agencies should use the most
      sophisticated procedures available to them, following the order of preference described in Section
      2.3.2.5 on analytical techniques for mode choice—basic or advanced analysis (i.e., mode choice
      models, "quick response" techniques, elasticity analysis).

      In performing appropriate analysis, the performing  agency must address, either explicitly in its
      analysis or through a  listing  of  assumptions, how  it  has  accounted for the  following  types of
      questions:

               •   What percentage of parking situations and consumers are affected by the proposed
                   policy, and how is this reflected in the tested rate?

               •   What travel groups are affected by the proposed price (e.g., commuters, shoppers,
                   etc.) and how is this reflected in the trip populations to which the analysis is applied?
2-44
                                                                   U. S. Environmental Protection Agency

-------
                                                                     Technical Methods for Analyzing Pricing Measures
              .••'•'..      .                                             to Reduce Transportation Emissions

 Figure 2.2a  Elements of Analysis for: Parking Pricing


  ELEMENTS of Basic Analysis    ANALYSIS STEP  RECOMMENDATIONS for Advanced Analysis

___•_,_	              (Type of Impact)
Formal analysis not required;
Statement of possible effects if rates are
applied with geographic differences.
 None
 None
 Formal analysis not required;
 Statement of possible effects if differential
 rates are applied.         '
Analyze through mode choice model, if
available;   •
^Elasticity factor" or "quick response" analysis
if no model;
Document sources and assumptions.


None           '        .    '
Standard assignment analysis, if capability
exists.
None.
Standard analysis, if emissions model
available;
If no model, apply emissions factors to the
change in vehicle trips and VMT.
                                            LAJiDUSE
                                                   ie i
                                               ^ #£'.•
       TRIP'
p MO|>E CHOICE,
    ^(Primary) •*, l
                                           :(Secoijdary)
                                       . ^ASSIGNMENT
                                       .-•;  •  (Trice)
                                         RATKAflEHICLE"'
                                          •  ""  .wlx    \v
                                        " "   (None)  \ -
£x*
T*f <.-V «/ " ,*„
« TRIP,,, -
GENERATION .
^^fr^o^ - >
, •£ * * ^
                                       ^-' (S^oondtary)  "-;
                       Develop alternative growth scenarios or apply land use models
                       capable of considering changes in transportation access and
                       cost;                           ...
                       Integrate alternative land use/urban design supportive of
                       alternative modes into land use models, or adjust subjectively to
                       evaluate:                   '     , ',   .  '

                       Re-run if shifts in land use (above).  •
                                                             Re-run if shifts in land use .or vehicle ownership;
                                                             Try to incorporate change in cost; '  ,
                                                             Run separately by purpose and income;
                                                             If using motorized vehicle trip tables only, adjust trip rates to
                                                             reflect shifts to non-motorized and non-automobile driver modes
                                                             using market share analysis or other model.

                                                             Run separately by time of day if time-variant rates;    ',
                                                             Run separately by purpose and income;            •
                                                             Explicitly model to reflect impact of change in parking cost.
                                                             Use separate models for each purpose;            -
                                                             Run separately by purpose and income;
                                                             Separate coefficient for parking cost;
                                                             Account for pass-through effect and HOV exemption/ discount;
                                                             Estimate shift to non-motorized/ non-travel modes.

                                                             Try to model if parking prices vary by time of day or short/ long
                                                             term;
                                                             Model by trip purpose and income.


                                                             Standard analysis
                                                              None
   .ANALYSIS^
                      Standard analysis.
U. S. Environmental Protection Agency
                                                                                                             '2-45

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions


               •   What proportion of the tested fee is assumed to be passed through to the traveler
                   (affects tested rate)?

               •   Does the policy provide for exclusion of or parking discounts to HOVs?
               •   Is the dollar value of the policy being tested consistent with the year in which the
                   elasticity was developed?

               •   What other factors are likely to be introduced or changed during the evaluation
                   period that could affect the implementation or traveler response to the proposed
                   application?

      Traffic Assignment and Emissions Analysis/Vehicle Mix:   Subsequent to the required  mode
      choice analysis, areas with full model systems are expected to perform traffic assignment analyses
      and apply an emissions  factor model  to determine  the resulting change in emissions. If these
      procedures are not available, emissions impacts may be estimated  by applying emissions factoring
      methods to estimated net changes in vehicle trips and VMT, taking cold start effects into account if
      they are considered significant.

      Land Use and Trip Distribution:  Parking  pricing that  is applied evenly in an area could have
      important secondary impacts  on land  use/activity allocations and trip distribution. Mode  shifts
      would tend to shift people to locations with more  available services.  Parking pricing that is
      applied unevenly in an area also could have  important secondary impacts on land use/activity
      allocations and trip distribution.  If the proposed parking pricing strategy will not be uniform.
      across different geographic and travel purpose markets, then the performing agency should address
      this concern in its submission.

      While a formal analysis  of these effects is not required, the submitting agency should provide at
      least a  statement of the potential impact from the  proposed strategy on  near-term  choice  of
      destination and longer-term land use trends,  and the anticipated implications of these shifts for air
      quality.  This statement  should include acknowledgment by decision-makers of the anticipated
      impact.


      Suggestions for Advanced Analysis of Parking Pricing Measures
      Use of the following  advanced analysis techniques  could produce a more comprehensive and
      accurate analysis of the impact of parking pricing; these are listed in approximate order  of
      presumed importance of effects (most important first).

     Mode  Choice:   More  advanced evaluation  of mode choice  impacts would utilize  more
      sophisticated models (described in Section 2.3.2.5 as advanced analysis for mode choice) such as
      " nested"  multinomial  logit models, incorporating parking cost coefficients (where possible), and
      differentiating among income strata and trip purpose.

      Trip Distribution: Enhanced analysis would be based on use of a trip distribution model sensitive
     to trip costs such as parking pricing, or on a model that uses time-equivalencies to account for such
      costs (see Section 2.3.2.4 on advanced analysis for trip distribution).

     Land Use/Activity Allocations:  Programs that introduce major changes in the level or pattern of
     parking prices could affect longer term land  use and activity patterns.  Advanced analysis may be
     warranted, using procedures such as a " land use model" sensitive to cost and/or growth scenarios


2-46                                                                U. S. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                                                 '-••:-               to Reduce Transportation Emissions

      (see Section 2.3.2.1 above on advanced techniques and, for a more technical discussion, Section
      3.3.1 of the following chapter).      ,

      Vehicle Ownership  and  Trip Generation:   If the trip  generation procedure deals only with
      motorized-vehicle trips, then an adjustment should be made to the vehicle trip generation estimates
      to  reflect  shifts to non-motorized or non-automobile-driver options;  the analytical approaches
      described in Section 2.3.2.3 ar,e appropriate for advanced analyses of parking pricing effects.

      Time of Day.  Unless the parking policy results in major  changes in cost structure between peak
      and off-peak periods, it is unlikely that a sufficient time-of-day travel shift would occur that would
      be discernible by analysis.  Advanced  time of day analysis, if  warranted,- would follow those
      approaches described in Section 2.3.2.6, above (see also the more technical discussion in Section
      3.3.7); these include choice models, other references, or special surveys.

      2.3.3.2  Modal Subsidies

      Suggestions for both the basic analysis and the advanced  analysis pertaining to Modal Subsidies
      are illustrated in Figure 2.2b and described below.                                      •


      Basic Analysis for Modal Subsidy Measures
      Mode Choice:  Because the only primary effect of modal subsidies on travel and emissions is
      through mode choice, to be acceptable, an analysis should demonstrate a convincing evaluation of
      the modal  split effect of the proposed subsidy. (Some effects on trip distribution may be likely
      with mode shifts).

      The options for completing the basic analysis include, in order of preference,  use of: accepted
      mode choice  model; accepted "quick  response" methods;  or elasticity methods (see basic  or
      advanced analysis, Section 2.3.2.5).

      The performing agency also, is expected  to relate how it has accounted for the following types of
      questions:

      What percentage of travelers will be affected by the proposed policy, in terms of receiving a
      subsidy, and how is this reflected in the tested rate?

               •   What travel groups are affected by the subsidy (e,g., commuters, shoppers, etc.) and
                  how is this reflected in the travel base which has been used for analysis?

               •   What is the actual subsidy that is expected to be passed through to the traveler?  '
               •  Does the policy reward different groups at different levels (e.g., ridesharers versus
                  transit users) and how does the analysis represent the competition among modes?
               •  Is the dollar value of the policy being tested consistent with the year in which the
                  elasticity was developed?             •  , -
U. S. Environmental Protection Agency                           .       ~'!  ~~'   ~   \   !      -2-47

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
 Figure 2.2b  Elements of Analysis for:  Modal Subsidies


 ELEMENTS of Basic Analysis      Jiype of Implct)'   RECOWIMENDATIONS for Advanced Analysis


                                                              If significant impact on inter-mode cost relationships, deve
                                                              alternative growth scenarios or run land use models with
                                                              consideration of cost;
Formal analysis not required;
Statement of possible effects if subsidy levels
are significant.
 None
 None
Formal analysts not required;
Statement of possible effects if subsidies are
at significant level.
Analyze through mode choice model, if
available;                 '  .
"Elasticity factor" or 'quick response" analysis
if no model;
Document sources and assumptions.
Formal analysis not required;
Statement of possible effects if subsidies are
significant.
Standard analysis if have assignment
procedure.
None
Standard analysis if have emissions model;
If no model, apply emissions factors to the
change in vehicle trips and VMT.
  LAND USE
ALLOCATION
                                              (Trace)
                                       .«    VEHICLE
                                       tj-" OWNERSHIP
                                              (Trace)
                                               TRIP
                                          GENERATION  j
                                              (Trace)
                                              TRIP
                                              i rvic
                                         DISTRIBUTION

                                           (Secondary)
                                         MODE CHOICE
                                            (Primary)
                                        •TIME OF DAY/
                                           PEAKING
                                           (Secondary)
                                            TRAFFIC
                                         ASSIGNMENT
                                             (Trace)  fe
                                           EMISSIONS
                                         RATE/ VEHICLE
                                               MIX
                                              (None)
                                          EMISSIONS
                                           ANALYSIS"
                                     ihips, develop
aiiGiiiauve yiuwin sueiidnus ui iuii tana use models with
consideration of cost;
Integrate alternate land use/ urban design supportive of
alternative modes.


Re-run if shifts in land use. (if significant mode shift, ownership
should drop).




Re-run if shifts in land use or vehicle ownership;
Try to incorporate change in subsidies;
Run separately by purpose and income;
If vehicle-based trip generation, account for non-motorized/ non-
automobile driver shifts.


Run separately by purpose and income strata;
Try to reflect impact of change in cost among alternatives and
mode shift.
                   Use separate models for each purpose;
                   Run separately by purpose and income;
                   Make certain models appropriately sensitive to the different
                   subsidized modes/ instruments;   ,
                   Perform supplemental analysis to assess effects on use of
                   subsidized or non-motorized modes.
                   Perform supplemental analysis to account for peak/ off-peak
                   effect of subsidies.
                   Standard analysis
                                                            None
                   Standard analysis
2-48
                                                                              U. 5. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                  •   •                                to Reduce Transportation Emissions

               •,  What other factors are  likely to be introduced or changed  during the evaluation
                   period that could affect the implementation or traveler response to  the  proposed
                   application?

      Traffic Assignment and Emissions Analysis/Vehicle  Mix: Subsequent to the require'd  mode
      choice analysis, areas with full model systems'are expected to perform traffic assignment arialysis
      and apply an emissions factor model to determine the resulting change in emissions. If these
      procedures are not available, emissions impacts may be estimated  by applying emissions factoring
      methods to estimated net changes in vehicle trips and VMT, taking cold start effects into account if
      they are considered significant.                      "'•'',.-.

      Land Use, Trip Distribution and Time of Day:  Modal subsidies also may have secondary impacts
      on land use/activity allocations, trip distribution, and time of day.  Impacts on land use trends and
      trip distribution would be anticipated only from significant levels of subsidy.  Time of day impacts
      might occur if subsidized rates drew travelers into or out of peak periods. In such cases, although
      a formal analysis would not be required, EPA expects a statement of anticipated impacts.' ^


      Suggestions for Advanced Analysis of Modal Subsidy Measures
      Expanding  the evaluation to  include  the following  components would generate  a  more
      comprehensive and accurate analysis of the impact of modal subsidies; these suggestions are listed
      in descending order of importance.

     Mode Choice:  More extended analysis would incorporate those approaches described as advanced
      analysis in Section  2.3.2.5 above (i.e., mode choice models that specifically break out and allow
      for interactions among each of the candidate travel, modes), may include separate coefficients for
      parking costs, transit fares, etc., and are  preferably run  for each primary trip purpose, including
      home-based  work,  home-based other, and  non-home  based trips.   Mode choice  should be
     performed separately for each purpose and income strata.

      Trip Distribution:   Extended  evaluation would incorporate those procedures described under
     advanced analysis in Section 2.3,2.4 on trip distribution (i.e. models  that are  sensitive to cost
     considerations and that distinguish among trip purposes and income strata).

      Time of Day: Only if the subsidy policy results in major changes in cost structure between peak
     and off-peak should a formal analysis be attempted of time of day travel shifts. Such analyses are
     described in Section 2.3.2.6 (and Section 3.3.7) and  include models borrowed from other locations,
     previous research, or conducting special surveys.

     Land Use/Activity Allocations: If significant levels of modal subsidy suggest a different cost
     structure for regional travel, estimating this effect may  be important.  Because modal subsidies .
     would result in across-the-board changes in the cost of travel, significant subsidies could lead to
     longer-term  shifts in residential and employment locations to areas that require less automobile
     dependence.  Approaches for extended evaluation  include those described  in Section  2.3.2.1  on
     advanced  analysis for land use/activity allocations, including land use models sensitive to such
     costs and growth scenarios.                                                          -  •   '-

      Vehicle Ownership and Trip Generation: On their own, modal subsidies would not be expected
     to  cause households to own fewer vehicles or make a different number of trips. If alternative land
U. S. Environmental Protection Agency                                            ,                ,   2.49

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions


      use patterns are determined in the previous step, then the vehicle ownership and trip generation
      steps should be re-run to reflect the relocation of activity and its association with the respective
      vehicle ownership and trip rates in the new location.

      If the trip generation procedure deals only with motorized-vehicle trips, then an adjustment should
      be made to the vehicle  trip generation  estimates  to  reflect shift to non-motorized or non-
      automobile-driver options.

      Wherever possible, trip generation analysis should be performed separately for each purpose and
      income strata, to support the subsequent analysis of trip distribution and mode choice.


      2.3.3.3  Pump Charges and VMT Fees

      Suggestions for both the basic analysis and the advanced analysis pertaining to pump charges are
      illustrated in Figure  2.2c.  The primary effects triggered by pump charges would be on mode
      choice and  vehicle mix. Higher fuel prices would be expected to reduce the demand for private
      vehicle travel and increase ownership and  use of higher-mpg vehicles or alternative-fuels/electric
      cars.


      Basic Analysis for Pump Charges
      Mode Choice: Basic analysis includes those techniques described as basic or advanced analysis in
      Section 2.3.2.5 on mode choice impacts, that  is (in  order of preference): mode choice models,
      "quick response" methods, or elasticity methods.

      The performing agency also is expected to  indicate how it has accounted for the following factors
      or concerns:

               •   What percentage of regional travelers will be affected by the increased pump price,
                   versus those who would be expected to evade the price increase by purchasing fuel
                   outside the area; how is this reflected in the assumptions regarding the tested price?

               •   What average cost per mile is being assumed, and upon what average mpg is it
                   based?

               •   Is the dollar value of the policy being tested consistent with the year in which the
                   elasticity was developed?
2-50                                                               • U. S. Environmental Protection Agency

-------
                                                                   Technical Methods for Analyzing Pricing Measures
                                                                                to Reduce Transportation Emissions
 Figure 2.2c Elements of Analysis for:  Pump Charges

 ELEMENTS of Basic Analysis     ANALYSIS STEP
               '  ':  •        • ,     •  '  . •  (Type of Impact)   REC°MMENDATIONS for Advanced Analysis
 Formal analysis not required;
 Statement of possible impacts.
. None
 If trip generation for vehicle trips only, account
 for non-motorized and non-automobile driver
 modes.              :               .
 Formal analysis not required;
 Statement of possible effects.
 Analyze through mode choice model, if
 available;             '
 "Elasticity factor" or'quick response" analysis
 if no model;
 Document sources and assumptions.
 None
Standard assignment analysis if have
assignment procedure.
Indicate assumptions regarding change in
vehicle mix to reflect shift in mpg anticipated.
Standard analysis if have emissions model; .--
If no model, apply emissions factors to change
in vehicle trips and VMT.
   LAND USE
 ALLOpAftpN,,
                                          '  VEHICLE
                                          OWNERSHIP
                     Develop alternative growth scenarios; or run land use models with
                     change in transportation access and cost considered;       .   '
                     Integrate alternative land use/urban design supportive of    , '.
                     alternative modes into land use models.
                                                             Re-run if shifts in land use.
 * -- - TRIR,  .
 GENERATION'
   * a J     ««•£  ^
*~   -  '* *~ ^V~
  '  '(Track)  ':-
                                          
                    Standard analysis
  EMISSIONS <"'
RATBP/EHICLE
   , - MIX ~
-^(brimaryyc  '.
                    Adjust vehicle fleet mix to reflect changes in ownership and use.
  EMISSIONS
  ANALYSIS.,
                    Adjust vehicle fleet mix to reflect changes in ownership, and use.
U. S. Environmental Protection Agency
                                                                                                         • . 2-51

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
      Trip Generation: If the trip generation process forecasts only vehicle trips, an attempt should be
      made to estimate the shift to non-motorized or non-automobile-driver modes as a result of higher
      fuel prices, following the recommendation for advanced analysis in Section 2.3.2.3 on analytical
      techniques for trip generation.

      Traffic Assignment:   Following mode  choice  analysis, areas with appropriate 'capability are
      expected to perform a revised traffic assignment for input to the emissions analysis.

      Emissions/Vehicle Mix:   The performing agency should  relate its assumptions on  changes in
      average mpg due to shifts in vehicle mix, the impact this would have on the effective cost per mile
      used in  each  of the analysis steps, and whether this would be expected to result in important
      differences in impacts for different market groups,  particularly  by  income level.  Analysis
      requirements  can be met  with  use. of an emission factor model that incorporates vehicle mix
      impacts or, if that is not available, a narrative description that details program-induced changes in
      emissions rates in response to increased fuel prices.


      Suggestions for Advanced Analysis of Pump Charges
      A more comprehensive and accurate analysis of the impact of pump charges could be achieved
      with  more  extended analysis, using the  following  techniques (listed in  descending order  of
      importance).

      Mode Choice: Appropriate evaluation is described under advanced analysis in Section 2.3.2.5 on
      analytical techniques for mode choice (i.e., mode choice models such as "nested" logit models,
      with attention to differentiation among trip purposes and income levels).

      Emissions/Vehicle Mix:   Extended evaluation of emissions/vehicle mix effects  is described  as
      basic or advanced analysis in Section 2.3.2.8, and  includes hedonic  models, research studies,
      and/or special surveys.

      Trip Distribution: Extended evaluation of trip distribution effects would incorporate the advanced
      analysis approach described in Section 2.3.2.4; that is, a trip generation model sensitive to cost,
      used  to perform  a revised trip  distribution analysis  (performed separately by trip purpose and
      income strata if possible), which also reflects  changes in  attractiveness of destinations  (and
      purposes) due to fuel-price based overall increases in travel costs.

     Land Use/Activity Allocations and Trip Distribution: Because fuel price increases would result in
      across-the-board  changes in the cost of private vehicle  travel, significant increases in price could
      lead  to near-term shifts to closer  trip destinations and  longer-term  shifts  in  residential and
      employment locations to  areas that require  less automobile dependence.   Because these are
     secondary impacts, no formal analysis is required, but the performing agency would be advised to
     perform such an  analysis if it has the capability,  or at a minimum, state its assumptions as to the
     anticipated  impacts  this price change would have  on trip distribution and land  use/activity
     allocations.                                  '                ;

     If significant levels of fuel prices  suggest a different cost structure for  regional travel, it may be
      important to estimate this  effect on long-term land use locational preferences. Two options are
     described under advanced analysis in Section 2.3.2.1 on land use/activity allocations, that is:  land
     use models sensitive to cost, and growth scenarios to depict the alternate ways in which the growth
2-52                                                                u 5; Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                                                                      to Reduce Transportation Emissions

      patterns might adjust to an overall change in pricing.  Scenarios might be developed in conjunction
      with application of the land use model, and involve development and integration of alternative land
      use/urban design configurations supportive of subsidized modes.

      Vehicle Ownership and Trip Generation-. If shifts in land use patterns are determined in the step
      above, then the vehicle ownership and trip  generation steps should  be  re-run to  reflect the
      relocation of activity and its association with potentially different rates of trip making and vehicle
      ownership.   Wherever possjble, trip generation should be performed separately for each purpose
      and income strata, to support the subsequent analysis of trip distribution and mode choice.

      In general, an increase in fuel price would not be expected to have a-maj.or, impact on the overall
      number of trips made by households. However,  if the local planning process considers vehicle-
      serviceable trips rather than person trips in trip  generation,  then  the evaluating agency should
      investigate any impact of higher fuel prices on shifts to non-motorized or non-automobile-driver
      modess and to closer destinations.


      2.3.3.4 Emissions Fees

      Suggestions for both the basic analysis and the advanced analysis pertaining to emissions fees are
      illustrated in Figure 2.2d and discussed below. (If there is a significant VMT component; then
      address also the Pump Charges guidance).


      Basic Analysis for Emissions Fees,

      The primary^ effects of emissions fees would be on  vehicle mix and emissions. If the proposed fees
      imposed a significant cost on vehicle ownership, then impacts on land use and vehicle ownership
      also could be important. Because emissions fees (as defined here) would be assessed on a periodic
      basis, they would not be expected to affect daily travel choices.

      Vehicle Mix/Emissions Analysis-. Fees levied on vehicles in relation to their age and/or emissions
      production (emissions rate,  perhaps cpupled  with VMT) would have the primary effect of
      encouraging replacement of higher emitting/older vehicles and/or encouraging higher use of lower-
      emitting vehicles in multi-vehicle households.                  ,

      Evaluation required in submissions to EPA includes the basic or advanced  analyses techniques
      described in Section 2.3.2.8,' above; that is,  hedonic models,  research studies, and/or  special
      surveys (these methods are further detailed in Chapter 3, Section 3.3.3).  The agency also should
      make an effort, to indicate emissions fee impacts on relative usage patterns of particular vehicles,
      accompanied by a statement on assumptions relating impacts to market segment (income).  For
      fees based on actual (measured) emissions rates, evaluation  should  include an estimate .of the
      effects on improved maintenance.                                         ,                    .
U. S. Environmental Protection Agency
            "

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
Figure 2.2d  Elements of Analysis for:  Emissions Fees

ELEMENTS of Basic Analysis      ANALYSIS STEP  „«-/»«»«.=•,•«.««««.«   --,
                                        (Type of Impact)  RECOWIMENDATI°NS for Advanced Analysis
If measure affects vehicle ownership/ use,
submit statement that assesses effects on
tocational patterns.
    LAND USE
   ALLOCATION
    (Secondary)
, If price significantly affects vehicle ownership or VMT, then
 consider development of alternate growth scenarios or application
 of land use models to reflect change in costs on number of
 vehicles.
Assess if measure affects number of vehicles
owned, following procedures for vehicle mix
analysis.
     VEHICLE
   OWNERSHIP
     (Primary)
 Re-run if shifts in land use;
 Possible supplemental analysis to estimate change in vehicle mix
 and possibly usage patterns favoring cleaner vehicles.
Investigate impacts if land use or vehicle
ownership affected.
       TRIP
'('  GENERATION
                                       if    '  (Trace)
 Re-run if shifts in land use or vehicle ownership;
 If trip generation does not include non-motorized, non-automobile
 driver modes, adjust to assess impact;
 Run separately by purpose and income.
Investigate impacts if land use or vehicle
ownership affected, or if VMT fee significant;
Otherwise, no analysis.
       TRIP
  DISTRIBUTION
      (Trace)
 Run separately by purpose and income;
 If significant VMT cost effect, incorporate cost modeling.
Investigate impacts if land use or vehicle
ownership affected, or if VMT fee significant;
Otherwise, no analysis.
  MODE CHOICE

      (Trace)
 Use separate models for each purpose;
 Run separately by purpose and income;
 Supplemental analysis to assess any effects of shift toward higher
 mpg vehicles;
 Assess demand for non-motorized modes, or impact of
 periodically-assessed charge on daily travel.
None
                                          TIME OF DAY/
                                            PEAKING
                                              (Ndne)
                                                             None
Investigate impacts if land use, vehicle
ownership, distribution or mode choice
changes;
Otherwise, no analysis.
   ASSIGNMENT
       (None)
                      Standard analysis
Indicate assumptions to account for shift in
vehicle mix to cleaner vehicles.
    EMISSIONS  '
  RATE/ VEHICLE
        MIX
     (Primary)
Adjust vehicle fleet mix to reflect change in ownership and use;
Adjust emissions rates for better vehicle maintenance, if a feature
of measure.
Standard analysis if have emissions model;
If no model, apply emissions factors to change
in vehicle trips and VMT.
    EMISSIONS
    ANALYSIS
Adjust vehicle fleet mix to reflect change in ownership and use;
Adjust emissions rates for better vehicle maintenance, if a feature
of measure.
2-54
                                                                               U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                   •                               . • .,;f           ;    to Reduce Transportation Emissions

      Areas with emissions factor models are expected to use these to assess the result of the proposed
      strategy on emissions  and to .incorporate the effect  of emission fees on vehicle mix and/or
      improvements in'maintenance.

      In the absence of an emissions factor model, agencies may apply emissions factor relationships to
      changes in vehicle trips and VMT, with appropriate supporting documentation of their analysis and
      assumptions  for emissions factors that reflect anticipated shifts in fleet mix and maintenance
      levels.                                       ,    .      .                         ,         ,   '

      Vehicle Ownership:  If proposed emissions fees lead to significant cost increases in automobile
      ownership, household ownership rates may be reduced, particularly ownership  of high-emitting
      vehicles.  Evaluation would follow that described under vehicle mix/emissions analysis above:  .

      Land Use/Activity Allocations:  If vehicle ownership is affected by a  sizable .emissions-based
      increase in registration fees, or if the VMT component of emissions fees is significant, households
      could shift to locations where fewer vehicles and/or  less dependence is required.  This effect
      should be investigated.          '


      Suggestions for Advanced Analysis of Emissions  Fees
      Listed in declining .order  of  importance,  the following analyses could help generate a more
      comprehensive and accurate evaluation of emissions fee impacts.
                                   >*                        .        ;""'-•'           '
      Vehicle Mix and Emissions Analysis:  Because the primary impact of emissions fees is expected
      to be on vehicle ownership  patterns, enhanced evaluation of emissions/vehicle mix effects could be
      achieved through more sophisticated application of the same  techniques described above under
      basic analysis (i.e., hedonic models and/or secondary analyses, such as stated preference surveys to
      explore vehicle ownership and usage).  In  addition, the agency could use  the above techniques to
      investigate whether the VMT aspect of the fee produces any change in daily travel  behavior, how
      large a fee would have to be applied, and how frequently it would have to be assessed.

      The most  accurate emissions analysis  will probably  use an  accepted emissions factor model
      adjusted to account for these changes, rather .than application of emissions factors to net trip and
      VMT changes.

      Vehicle Ownership and Land Use:  Vehicle ownership would  be affected by significantly higher
      emissions-based registration   fees.   Agencies proposing such  fees  should  estimate  vehicle
      ownership patterns, by income group if possible. In turn, revised ownership rates should then be
      incorporated in a revised land use analysis that reflects the impact of fewer vehicles and higher
      costs, using models, if available, or best professional judgment.

     Mode Choice, Trip Distribution,  Trip Generation:  Generally, emissions fees collected through
      infrequent periodic assessments would not be expected to lead to major, adjustments in daily travel
      behavior.   If the  VMT component of the fee is either significant or controlling,  however,
      investigate its effects on mode choice, trip  distribution,  and trip generation by consulting research
      studies or performing a supplemental analysis of relative impact on daily behavior of annual (or
     periodic) fees.        ."'.'••-
U. S. Environmental Protection Agency   ,                                                            ,2.55

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions     .  '


     Any changes in vehicle ownership and household location also would have an indirect effect on
     trip generation, distribution, and mode choice, and should be assessed.  Trip generation analysis
     should reflect total person trips, and allow for shifts to non automobile-driven and non-motorized
     modes.


     2.3.3,5 Roadway Pricing

     Suggestions for both  the  required.and  advanced analysis  pertaining  to roadway  pricing are
     illustrated in Figure 2.2e and discussed below.


     Basic Analysis for Roadway Pricing
     Because it is a fairly complex measure in terms of travel behavior impacts, there are more primary
     effects associated with roadway pricing than any of the other measures.  Roadway pricing would
     be expected to have a primary impact in relation to mode  choice, trip distribution, and traffic
     assignment, and, if peak period or congestion pricing is instituted, time of day choices as well.

     Trip Distribution: Travelers would be expected to respond to the pricing  of specific transportation
     facilities by favoring comparable  destinations which can be reached without payment of a toll
     (where such a substitution can occur). There are several options for approaching this analysis.
     The more rigorous approaches are described  in Section 2.3.2.4 as basic or advanced  analysis
     techniques  for  trip  generation,  which  include  cost-sensitive  trip  generation models  and
     "generalized cost" procedures.

     Sites that do not have formal trip distribution modeling capabilities should formulate  a statement
     that reflects a careful assessment of the anticipated impacts of the road pricing scheme on shifts in
     travel patterns.

     Traffic Assignment  (Route Choice):  Traffic assignment becomes particularly important in the
     case of road pricing since travelers can choose between tolled routes that may have a higher level
     of service, and toll-free routes that may be more congested.

     This choice is not directly handled by most traffic assignment models. The options for addressing
     anticipated impacts,  described in  Section 2.3,2.7 on basic or advanced analysis  techniques for
     traffic assignment, include assignment routines  for toll  and free  paths, or use  of  travel time
     equivalency estimates (see also Section 3.3!8.5).

     If an area cannot perform traffic assignment analysis, or if the tolled facility can be easily avoided,
     then it may not be possible to consider the emissions  impacts of roadway pricing because the
     impacts would be too uncertain.

     Mode Choice;  Mode choice is still an important factor in considering road pricing.  Analytical
     options include those detailed above in Section 2.3.2.5 as basic or advanced analysis techniques for
     mode choice (i.e., mode choice models, "quick response" methods, or elasticity methods).
2-56                                                           •    . U. S. Environmental Protection Agency

-------
                                                                    Technical Methods for Analyzing Pricing Measures
                                                                                 to Reduce Transportation Emissions
 Figure 2.2e Elements of Analysis for:  Roadway Pricing

 ELEMENTS of Basic Analysis      ANALYSIS STEP  RECOMMENDATIONS for Advanced Analysis
============^^^^^^^  (Type of Impact)
, Formal analysis not required;
 Statement of possible effects due to pricing of
 select facilities.
                                         ^ALLOCATION,
                       Develop alternative growth scenarios or run land use models, to
                       reflect cost related to priced network;
                      , Integrate alternative land use/urban design supportive of
                       alternative modes.     :                        -
 None
                                          OWNERSHIP,''
                                                             Re-run if shifts in. land use-
 None
                                                 Jp'
Analyze if have trip distribution model;
If no model, provide estimates of expected
impacts;
Account for non-motorized trips if vehicle trips
only.
  DISTRIBUTION;
 ;/~ (Primary)',
                       Re-run if shifts in land use;
                       Run separately by purpose and income;
                       Consider shifts to non-motorized alternatives, if area-wide pricing
                       and vehicle^only trip generation.
                                                             Run separately by purpose and income;
                                                             Explicitly model to reflect change in cost and travel time changes.
Analyze through mode choice model, if
available;
"Elasticity factor" or 'quick response" analysis
if no model;
Document sources and assumptions.
  MODE CHOICE
   ,  (Primary) •
  "*_ "'<13>v
                                                             Use separate models for each purpose;           .
                                                             Run separately by purpose and income;
                                                             Directly represent pjice for trip (rather than travel time equivalent);
                                                             Possible supplemental analysis to ascertain response to
                                                             electronic/periodic charge assessment.
Must analyze if apply time of day pricing;
Document sources and assumptions.
                                           PEAKING  ,T,.
                                          •• (Primary'lt '  r
                       Acquire/ develop model to simulate response to time of day
                       pricing;
                       Possible supplemental analysis to gauge time of day shifts.
Must analyze if pricing applies to selected
facilities;  •
Can use elasticity factor methods; Document
sources and assumptions.        r
                                        **  TRAFFIC
                       Develop and use toll diversion model to reflect toll/free path
                       options;                                    ,
                       Possible side analysis to gauge route choice shifts.
None
   .EMISSIONS
  WE/VEHICLE,,
:    :MIX,'(Norie)   :
                                                             None
Standard analysis if have emissions model;
If no model, apply emissions factors to ihe
change in vehicle trips and VMT.
    EMISSIONS^
    ANALYSIS  -
                                                            Standard analysis;
                                                            Possibly adjust emissions rates for major speed changes on
                                                            facilities.                  .
U. S. Environmental Protection Agency
                                                                                                            2-57

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions                      •


      Most likely, an area considering road pricing would have a reasonably sophisticated modeling
      system, complete with a mode choice model. Application of elasticity techniques would only be
      expected  to be valid where simple pricing-schemes are being considered (i.e., a bridge or tunnel
      with few/no parallel routes).

      Important issues that should be addressed in the analysis include:

               •   Will  the toll  be  paid to toll booth operators, to  exact change  machines, or
                   electronically   through  advanced  vehicle  identification   (AVI)   transponder
                   technology? What are the queuing and delay impacts of the chosen method?

               •   Are HOV users exempt from the toll, or charged discounted rates?'
               •   Is the dollar value of the policy being tested consistent with the year in which the
                 .  elasticity was developed?

               •   What other factors are likely to be introduced  or changed during" the evaluation
                   period that could affect the  implementation or  traveler response to the proposed
                   application?

      Time of Day:  Peak-period "congestion pricing"  versions of road pricing would be expected to
      affect time of day choices as well as route choice and, if such strategies are being considered, the
      agency should attempt to estimate the impact of the fee on peak versus off-peak travel demand by
      trip purpose and income group.

      Options for conducting this  analysis are described  above in  Section  2.3.2.6, basic or advanced
      analysis techniques for time of day, and include: time of day choice models, references to other
      research,  and survey analysis.  Under extenuating circumstances, the area may be allowed to use
      assumptions to suggest the degree of change, but it is expected to provide documentation as to the
      assumptions and their degree of impact on the result.
                                                X
      Land Use/Activity Allocations: Implementation of toll pricing on selective transportation facilities
      would be expected to cause a shift in locational trends, which could reduce use of these facilities.
      Because these are secondary impacts, no formal analysis is required.  Nonetheless, the performing
      agency is expected to state  its assumptions (and show the concurrence of decision-makers)
      regarding the anticipated impacts that the system of road pricing would have on land use/activity
      allocations. This is particularly important in the context of partial-network pricing.
2-58                                                                U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                                      to Reduce Transportation Emissions
      Suggestions for Advanced Analysis of Roadwiy Pricing
      In descending order  of importance, the  following  analyses  would  contribute  to  a more
      comprehensive arid accurate evaluation of roadway pricing impacts.

      Traffic Assignment (Route Choice): The best analysis would use a procedure including both toll
      and free paths as options to travelers, taking into consideration the fact that traveler choice will be
      sensitive to type of trip and  traveler income level (see Section 2.3.2.7  for a  more 'extended
      discussion, and Section 3.3.8 for a more technical presentation).

      Time of Day:  Time of day choice flexibility is particularly important in assessing the impacts of
      congestion pricing schemes. Shifts in time of day would be expected to vary significantly by trip
      purpose and  income  level.  Time of day  transportation modeling is a  relatively .uncommon
      practice; options described above in Section 2,3.2.6 include time of day choice models, references
      to other research, and surveys (Section 3.3.7 describes appropriate models).

      Trip Distribution: Shifts to alternative destinations induced by road pricing could be important,
      particularly with partial-network pricing  and numerous travel  or route alternatives;  impacts are
      likely to vary by trip type and traveler income levels.  Include distribution effects of reduced
      congestion.

      Application of a  trip  distribution model  which is sensitive to cost through a generalized cost
      relationship to individual  trip purposes and income strata  would  be essential to an effective
      analysis.                                           ,  .

      Mode Choice:  Mode choice  analysis would  be'most accurate if it were accompanied  by the
      described advanced analyses of trip distribution, route choice, and time of day alternatives.  It also
      would be most accurate if performed with separate models  of each trip purpose and  evaluated
      separately for each income strata.

      Land  Use/Activity Allocations:   Road prices' would  be expected to make  some  areas more
      attractive than others in terms of future locational preferences.  Options for approximating andv
      accounting  for these trends include use of land use models  that incorporate sensitivity to cost,
      and/or the development of alternative growth scenarios.  It would be particularly meaningful to
      develop land use  and urban design scenarios which would complement travel alternatives to those
      sites with priced facilities/access.                               *

      Vehicle Ownership  and Trip Generation:  If alternative land  use patterns are anticipated,  the
      vehicle ownership and trip generation steps should be re-run to reflect the relocation of activity and
      its  association with potentially different  rates of trip making and vehicle ownership.  When
      possible, trip generation analysis should be performed  separately for each purpose and  income
      strata, to support the subsequent analysis of trip distribution and mode choice.
U. S. Environmental Protection Agency                                                                2.59

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions


•   2.4  Packaging Measures for Best Effect


     The preceding discussion of market-based measures and guidelines for their evaluation necessarily
     provide a somewhat one-dimensional view of the nature and impact of these strategies.  In reality,
     such measures would  seldom be deployed in isolation, as described here.  Rather, they would be
     deployed as part of a carefully considered "package" of different market measures and other
     actions to complement and support the market-based measure(s).  Market-based measures serve as
     signals and incentives to change;  other  actions provide the appropriate setting and options
     necessary to complete the process.

     There are two major benefits from packaging market-based measures into balanced programs:

              •  Balanced programs offer synergy that will generally produce a. greater impact than a
                 market-based measure acting alone; and

              •  Balanced programs that offer tangible and attractive choices are more likely to foster
                 public acceptance of the proposed pricing measure component of the program.

     This  guidance "manual cannot offer hard and fast rules detailing the most effective packaging
     strategies to accompany pricing initiatives, given the wide variation in settings, needs, and
     circumstances. In general, however, pricing measures are expected to work best if the following
     general principles are applied:

              1. Match actions to consumer responses— Anticipate the types of responses consumers
                 are likely to make in response to a particular pricing  measure. Analysts can use the
                 guidance presented in the preceding sections, coupled with the performing agency's
                 knowledge and  experience.  Analysts  also should evaluate reciprocal actions that
                 could be mated with these responses.

              2. Identify realistic alternatives—  Where  a measure is likely to raise interest  in an
                 alternative mode of travel, identify alternatives that will be attractive and appropriate
                 in that setting.  For example, consider transit service improvements if a high density
                 corridor or downtown area is involved. If the impact area is more suburban, transit
                 may still have a role, but actions to stimulate carpooling or vanpooling may be more
                 suitable.

              3. Consider non-motorized modes— Broaden the analysis to include non-motorized
                 alternatives such as walking and biking, which are often overlooked when options
                 are considered.  A fair portion of shorter trips may be  amenable to walking or biking
                 under the appropriate circumstances; shifting these trips from private vehicles to
                 non-motorized modes could eliminate substantial emissions associated with vehicle
                 " cold-starts."  More specifically, consider enhancing circumstances associated with
                 bike and pedestrian modes such as: improving sidewalks; and pedestrian access to
                 buildings;   providing    frequent,   safe   pedestrian   crossings;    alleviating
                 vehicle/pedestrian  design conflicts;  separating vehicles from  pedestrians  and
                 reducing  speeds;  and   providing  bicycle lanes,  paths,  suitable  lockup/storage
                 facilities, and lockers or change facilities at destinations. Also consider approaches
                 that encourage mixed use land development,  such as reduced taxes for development
                 that reduces transportation costs.
                                                                 U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                    "..•-"       "                to Reduce Transportation Emissions

               4. Consider positive  incentives for  desirable  alternatives—Identify  and  evaluate
                  incentives for users of the more environmentally efficient modes. Incentives such as
                  fare subsidies to transit or vanpool users, parking subsidies for HOVs, and equipment
                  rebates can provide a powerful economic complement to market-based measures that
                  primarily increase the cost of travel for driving or use of high-emitting vehicles.

               5. Consider barriers to  counter-productive traveler choices—Articulate strategies to
                  limit 'or  direct the choices of travelers  who  might  select "counter-productive"
                  alternatives in response to the market-based measure.  Depending on who they are
                  and the purpose of their travel, many travelers have flexibility in their choice of
                  route, destination, and time of day.  When faced with market-based incentives, these
                  travelers  would be  expected to consider and use that  flexibility  in their own  best
                  interest.  Analysts  should anticipate these responses and, unless these secondary
                  responses support emissions  reduction  objectives, should propose complementary
                  actions that reinforce desired behavior patterns.  Mechanisms include:  limitations on
                  off-street parking  coupled with  site parking charges; and limitations on  use of
                  secondary/neighborhood roads as alternatives to a priced facility.

               6. Explore  outreach  and indirect support—Consider  complementary  actions in
                  addition to financial incentives and physical alternatives that support travelers faced
                  with changes in their travel context.  For example, information outreach, education,
                  and marketing can ease barriers to use of alternative travel modes   They can also
                  explain and demonstrate the problems caused by driving and the virtues  of walking,
                  bicycling, and telecommunications.  Effective  matching programs can put people
                  into acceptable ridesharing arrangements.  Employers can play a very valuable role
                  in supporting employee travel choices, including: guaranteed rides home in event of
                  late work nights or family emergencies;  assistance with dependent care;  preferential
                  or covered parking for ridesharing employees; and assistance with vanpool formation
                  arid operation.  Employers also can offer opportunities  to work at .home through
                  telecommuting arrangements, or introduce modified work schedules to reduce the
                  number of physical trips to worksites.                         ,

               7. Promote  land  use planning  that supports  alternative  travel—Recognize  the
                  importance of an environment that supports the  use of alternative travel patterns. If
                  persons are expected to travel to targeted destinations without their personal vehicles,
                  they must be confident that they can achieve their objectives at that location without
                  a car at their disposal  This calls for innovation in urban design that allows people to
                  circulate  and accomplish  activities pleasantly through walking, biking, transit or
                  through shuttle services. This also means designing those centers so that the mix .of
                  uses is varied, complementary and attractive.   Street space  should be managed in
                " these areas to provide pedestrians  with safety from motor vehicles, .freedom of
                  movement, attractive visual surroundings, and comfortable settings for conversation,
                  rest arid shopping.      , '   •

               8. Use program revenues to support transportation and emissions goals— Clearly and
                  explicitly link the revenues that are collected through application of a pricing action
                  with the Use of those revenues. If the public perceives that the revenues extracted
                  from them are used for the direct  enrichment of their environments,  rather than
               -   diverted to other uses, they should show stronger support for those programs since
                  they are getting something back  from their contribution.  Revenues  from pricing
U.S. Environmental Protection Agency                                     •                          2-61

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
                   programs are often  a convenient and adequate  source to finance a  variety  of
                   supporting strategies.

      For more  information on techniques  for  effective packaging, the reader is referred  to  other
      information sources, such as Implementing Effective Travel Demand Management Actions, which
      is recommended as a  comprehensive source for packaging and  impact information and contains
      considerable information on a wide variety of strategies and packaging tips, as well as a significant
      number of references to other sources on specific related topics.14
       COMSIS Corporation et al., Implementing Effective Travel Demand Management Measures:  Inventory of
       Measures and Synthesis of Experience, prepared for the Federal Highway Administration and the Federal
       Transit Administration, U. S. Department of Transportation, Washington, D.C.,1 September 1993 (DOT-T-
       J/*T~l/^ ),
2-62
                                                                     U. S. Environmental Protection Agency

-------
                                                     Technical Methods for Analyzing Pricing Measures
                                                                to Reduce Transportation Emissions,
3.0    Guidelines for Analysis  of
           Market-Based Measures
     3.1  Introduction
     This chapter provides guidelines and suggestions for those responsible for performing the actual
     evaluation of market-based measures for emissions credit. Whereas the preceding chapter was
     designed for managers and those without a technical background,, this chapter was developed for
     those versed in the basic principles of transportation modeling and analysis. Expanding on topics
     discussed in Chapter 2, this  chapter employs terminology and concepts that'assume  greater
     famijiafity with transportation and emissions modeling tools and procedures. The intent is to
     provide fairly specific language as to EPA's expectations when reviewing credit submissions.

     The chapter is divided into two main sections. The first section, Section 3.2, presents guidance on
     each of the five basic groups of market-based measures (parking pricing, modal subsidies, pump
     charges, emissions fees,  and road pricing). The second section of the chapter, Section 3.3, is  a
     reference compendium on methods and procedures suitable for the types of analyses suggested in
     Section 3.2. The  reference compendium is organized according to steps in the analysis process.
     The methodological guidelines in Section 3.2 refer extensively to the methodological discussions
     of Section 3.3, many of which also list external references for further details.
                         I               ,                            '

     This document has been prepared more as guidance than traditional requirements. EPA recognizes
     that that no single approach can work for all sites and that areas vary widely with respect to
     analytic tools, data, and staff capabilities. It is EPA's intention to suggest analytic approaches for
     best practicable evaluation of the proposed market-based measures.

     EPA will use this guidance when evaluating  an  area's  submission of a measure for  credit.
     However, EPA also desires to work cooperatively with any  area that  is seriously considering
     market-based measures and will support reasonable .best efforts to introduce such measures. EPA
     expects those areas with more sophisticated analytical tools to use them to the best possible
     advantage in performing an accurate assessment.  At the same time, areas with limited analytic
     capability should  not  be  discouraged  from  investigating market-based measures,  and  are
     encouraged  to  use this interest  as a  basis for upgrading  their analytical tools for  future
     investigations.

     This chapter was  designed  to give direction.to  technical staff and, as such, goes beyond basic
     description  of analytical approaches. Nonetheless, some practitioners may want  additional
     information, such as clear-cut examples or instructions on how to make particular changes to their
U. S. Environmental Protection Agency                                                           3-1

-------
Technical Methods for Analyzing"Phcing Measures
to Reduce Transportation Emissions
     models or how to apply their models in specific circumstances. Appendices A and B address these
     needs by providing detailed technical procedures on model applications.

     Appendix  A  describes a "Case  Study Model," which  involves  practical  modification of a
     conventional  four-step  model  system to improve its sensitivity to pricing.  In addition to a
     description of the modification itself, the appendix demonstrates model applications to test market-
     based measures.  Appendix B describes a somewhat different approach that uses household data
     and sample enumeration methods to obtain increased flexibility and potentially greater accuracy in
     evaluating pricing measures.
                  ,!'                                   '                 '            '
     Performing sites  are encouraged to contact their respective EPA regional offices for more
     information.  In the course of undertaking an evaluation of market-based measures, agencies also
     are advised to:

           •   Establish and maintain  a dialogue with EPA and other  organizations with approval
               authority;

           •   Confer with appropriate agencies to coordinate assumptions and data;

           •   Consult with state and local air agencies and EPA on planned analytic approaches to
               obtain agreement on the proposed approach; and

           •   Consult with transportation  agencies, environmental  and community  groups,  and
               possibly consultants or academics to  make sure that  analyses  and assumptions  are
               reasonable and accepted.
3-2                    ,                                             U. S. Environmental Protection Agency

-------
                                                        Technical Methocfs for Analyzing Pricing Measures
                                                                   to Reduce Transportation Emissions
     3.2  Evaluation Guidance for Individual Pricing Measures


     This section provides step-by-step, guidance for evaluation of each of the following five types of
     market-based measures:

              •  Parking Pricing (Section 3.2.1);

              •  Modal Subsidies (Section 3.2.2);

              •  Pump Charges (Section 3.2.3);                                 ,   .   '

              •  Emissions Fees (Section 3.2.4); and
                                                          '  .                     t
              •  Road Pricing (Section 3.2.5).

     The discussion of each follows a common four-part format. Each of the five segments begins with
     an introductory table that summarizes the relationship between the given measure and each of the
     nine steps in the analysis hierarchy with regard to the following:              ,              •

              •  Form of impact expected at that step— that is, how the market-based measure will
                 work to affect transportation factors analyzed in that step of analysis;
                       of impact expected— -primary, secondary, trace, none  and derivative (i.e., it
                 must be addressed as a result of impacts in other steps); and                  '
             . •  Impact indicator— that is, measures of performance that should be employed in the
                 evaluation of each step.

     Discussion begins with a statement of Anticipated Impacts, or those effects that are expected to be
     important in evaluating the specified measure.  This is followed by a short summary statement of
     the Basic Analysis. The third component is Guidelines for Analysis, which presents step-by-step
     procedures  or considerations  that EPA  would expect to see applied in evaluating the given
     measure.  Analysis recommendations progress from "basic analysis5' to "advanced analysis,"
     which is more likely to result in the granting of maximum emissions credit.
U. S. Environmental Protection Agency
                                                                                           3-3

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
     3.2.1    Parking Pricing

          3.2.1.1  Anticipated Impacts
      The primary impact of parking pricing would be on Mode Choice.  Parking prices raise the cost of
      vehicle trips and thus increase the attractiveness of alternative.modes, including transit, fidesharing
      and non-motorized options. They may also affect trip distribution, shifting destination to  those
      more accessible  without private vehicles.  The degree to which other  options become  more
      attractive  depends   very  much  on-
                                               Table 3.1 Impacts of Parking Pricing
 attendant factors, such as:
 •   Importance of reaching the
    particular destination;
 •   Amenity/design characteristics of
    that location; and

 •   Service/quality of the options that
    serve that location.

 Parking pricing collected as a daily fee
 may have a greater effect than other
 pricing measures, since it is directly
 associated  with  the   cost  of  the
 particular  trip.   Parking  pricing  is
 expected  to  have a greater effect on
 shorter trips, for which it represents  a
 higher proportion  of overall trip costs.
 This  has important  implications  for
 affecting  cold starts  and  micro-level
traffic problems.
Parking pricing  policies  also  raise
practical concerns about distributional
impacts. It often is difficult in practice
to:
Analysis
Step
Land Use/
Activity
Allocation
Vehicle
Ownership
Vehicle Mix
Trip
Generation
Trip
Distribution
Mode Choica
Time of Day
Traffic
Assignment
Emissions
" Impact Form
Greater attractiveness
of areas with lower
prices (uneven
application) or with
better alternatives &
supportive land
use/density
None ,
None
Possible effect on
attractions
Diversion to
destinations with
lower prices or better
alternatives
Decreases private
vehicle use
Shift time of day
Shift in trips and VMT
as result of mode
choice & distribution
Reduced vehicle trips
and VMT
lm£act>v '
TsrwT
Secondary
None
None
Primary on
work-trips,
trace overall
Secondary
Primary
1 */
Secondary
(if time of
day differs)
Derivative
effect
Derivative
effect
impact indicator
<;•- '=-
Population and/or
employment, by
• zone
Vehicles per
household per zone
Percent of vehicles
by VMT
Number of person
trips to/from each
zone
0/D patterns
(average trip length)
Parson trips by ^ ,
mode; % transit; ''•*
% walk/bike;
# vehicle trips;
average vehicle L; „
occupancy-
Trips by time period
Link volumes; VMT, '
Tons or kg. of VOC,
NOX, CO, PM
      •  Apply parking  pricing  uniformly
         with regard to geography (e.g., suburb versus city), type of site (site size and accessibility, on-
         street versus lots or spillover options, etc.), and

      •  Eliminate employer or merchant discounting.

Travelers with flexibility in choice of destination may shift away from locations with increased parking
costs. This could apply to many non-work trips, and also to work and other fixed- location destinations
in  the  longer  run.  In  turn,  this  could  have  an  effect  on  land  use  as  the   comparative
advantages/disadvantages of various locations shift in response to pricing levels and uniformity, quality
of alternatives, amenities, etc.                      ,

If parking pricing is applied at different rates to different travel markets (e.g., commuters versus non-
commuters, or peak versus off-peak), then travelers with time flexibility, can be expected to shift their
chosen time of travel.
3-4
                                                                    U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing, Pricing Measures
                                                                  -   to Reduce Transportation Emissions
                                                              \

     3.2.1.2   Basic Analysis for Parking Pricing

     Aside from the standard analysis of emissions, parking pricing analysis should include a careful
     Assessment of the impacts on mode choice (person trips by mode) and the resultant effects of mode
     choice changes on  overall vehicle trips, .VMT, and  emissions.   If parking rates are to be
     implemented non-uniformly across the region, the effect on-shifts in trip distribution (particularly
     for non-work travel), land use, and time  of day (if applicable) should be assessed, either through
     formal analysis/assumptions or through  a discussion that describes  the anticipated impacts on
     travel and emissions.

                                              •  .                / •    •        .•'"..'
     3.2.1.3  Guidelines for Analysis


     1. Parking Pricing and Land Use/Activity Allocations

     Parking pricing can affect long-run locational patterns by decreasing vehicle trips, particularly to
     locations affected by the policy.  EPA does not require a formal analysis of parking pricing's
     effects on land use, although it would be more important to consider land use impacts if pricing is
     not uniform.  If parking prices are not applied uniformly  across the region, businesses  and
     employers might shift to areas with lower charges or areas with  better travel alternatives (e.g.,
     transit) and supporting density. Households are expected to make long-term shifts in residence to
     areas that support lower cost travel patterns—that is, access to work/non-work destinations  that
     have parking with low or no fees or that offer effective travel alternatives.

     If an area's planning process includes a land use allocation model that incorporates transportation
     cost through a measure such as composite impedance* the area may  wish to consider using this
     model to  investigate  parking price impacts on long-term development trends.  However, these
     results should be interpreted with caution given the limited accuracy of such models and the
     existence  of other factors  that  can  affect  location trends.   Section 3.3.1.2 provides more
     information on the nature and use of land use models.

     Agencies  may  wish  to consider formulating  alternative growth Scenarios  based  on various
     anticipated effects that development shifts might have on regional travel and emissions.  This
     approach is particularly relevant for agencies that do not use models for land use analysis.   This
     option, detailed in Section 3.3.1.1., involves  defining alternative land use/urban design schemes
     that support the alternative modes and/or travel patterns  encouraged by  the proposed parking
     pricing policy.                           '

     If the  proposed  parking policy entails major disparities in regional  coverage or level, and the
     agency cannot perform a formal analysis of its impacts, the agency should provide a statement that
     reflects local decisionmakers' appraisal of (1) the long-term consequences such a policy would be
     expected to have on regional development, travel trends, and  emissions; and" (2) what impact, if
     any, this will have on other land use and transportation investment analyses.                  ' "
      Composite impedance is a measure that combines the time and cost of highway and transit (and perhaps
      other) modes into one statistic; see Section 3.3.5.1.   •     ••'•..
U. S. Environmental Protection Agency

-------
 Technical Methods for Analyzing'Pricing'Measures'
 to Reduce Transportation Emissions


      2. Parking Pricing and Vehicle Ownership                                 '

      Because parking pricing  is expected to have little or no direct impact on vehicle ownership, no
      formal analysis is required.  Indirect  impacts on vehicle ownership could occur as a result of
      household location shifts described in (1) above; in this case, advanced analysis would include re-
      running the vehicle ownership analysis.


      3. Parking Pricing and Vehicle Mix

      Parking pricing is not expected to have a direct impact on ownership vehicle mix, thus no formal
      analysis is required.  Given that parking pricing is likely to affect short trips more than long trips,
      however,  parking pricing could  affect  vehicle usage  patterns; this should  be  taken  into
      consideration in determining vehicle distribution used for the emissions estimation  step.

      4. Parking Pricing and Trip Generation

      It is unlikely that parking pricing will have a measurable impact on the overall number of person
      trips made by households  as forecast in trip generation analyses, thus formal  analysis  is not
      required.

      Household vehicle trip rates could be affected by parking pricing in the following ways:

               •  Higher numbers of workers involved in telecommuting or compressed work weeks;

               •  Increased errand chaining to reduce the number  of stops at priced destinations;
                  and/or

               •  More use of telecommunications for non-work activities.


      While these responses may affect the pattern of household tripmaking, it is unclear whether they
      affect the overall number of trips or simply trip characteristics, such as destination, trip length,
      mode, or time of day.   Because there are no formal existing  methods for dealing with these
      responses, however, no formal analysis  is required (see Section 3.3.4.5).

      There is some concern regarding what types of trips are depicted  by the trip  generation process
      used for analysis—total person trips or only those trips that are vehicle serviceable.  Because the
      latter  approach excludes walking  and bicycle trips, important shifts in household trips from
      vehicular to non-motorize^ modes are  likely to be underestimated;  Section 3.3.4.4 discusses this
      aspect in greater detail.
                               ••••''     •                 t
      Areas that  do not currently  forecast total person trips are encouraged to upgrade  to such  a
      framework as a future goal; for the present, these areas should try to consider these effects in their
      analysis as best they can.  Section 3.3.6.6 suggests ways in which areas currently unable to project
      non-motorized modes can account for these effects later in the mode choice step.

      Smaller urban areas without mode choice model.s may have  an even more restricted definition of
      trip generation, considering only, vehicle trips  (i.e., only those person trips that are  made in private
      vehicles, excluding  transit).   Trip tables limited to  private vehicles  inhibit analysis of pricing

3-6                    .                                             U.. S. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                                    '.     .                             to Reduce Transportation Emissions

      strategies, although  options  for overcoming  this constraint  are  discussed  in  the  mode choice
      segment below.

      Parking pricing may have an effect on trip attraction rates, although there are no known models
      that are directly sensitive to  parking pricing in this way.  For example, suburban shopping sites
      tend to have higher vehicle trip attraction rates than downtown areas,  although this may simply
      derive from a failure to account for transit and  non-motorized trip's.   Areas are advised to be
      sensitive to these differences in their trip generation analyses.
                                                  \ .     '     -     -          - ,        '          '    .

      Parking pricing  is not expected to have a major (or predictable) impact on freight/commercial
      traffic, except in the event that parking.supply itself also..becomes more restricted, for example,
      due to limitations on curb parking. It also is unlikely that parking pricing would have a tangible
      effect on through (or "external-external") traffic—that is, trips originating from and, destined for
      locations outside the area of the pricing scheme.


      5. Parking Pricing and Trip Distribution

      Parking pricing is expected to  have only a secondary effect on trip distribution; thus, no formal
      analysis is required.  If parking pricing is applied differentially, however, a statement of possible
      effects should be included.                                          •.   :                  ,

      Trip distribution analysis can be a particularly important element in assessing the travel/ emissions
      response to parking pricing in the following ways:

               •  Trip purpose:  Travelers who have  the flexibility to  shift  to alternative destinations
                  for some trip purposes, such as shopping and social/recreation, would be expected to
                  consider  and perhaps shift to destinations which  serve more needs and/or are
                  accessible without a private vehicle and if parking  charges increase in only some
                  locations  to those with free or lower  parking charges  (unless there are offsetting
                  amenity or utility circumstances).  Travelers are less likely to  shift destinations on
                  trips with more rigidly determined  destinations, "such as  work, school o'r personal
                  business,  at least in the short run.

               •  Trip length:  Parking charges will have a higher proportionate effect on shorter trips,
                  which also are strong candidates for transit or non-motorized alternatives (see mode
                  choice discussion below).

      Advanced analysis for areas with appropriate modeling capability includes evaluation of the effect
      of .parking pricing on trip  distribution,  both for  non-uniform and uniform pricing applications.
      This is best done with a gravity model that incorporates cost in its measure of separation through a
      feature such as composite impedance (see Section 3.3.5.1). If parking price is added to overall cost
      of travel to the respective destination, a different pattern of destination choices in the trip table can
      be expected to result. If the separation measure in the gravity model does not directly incorporate
      cost, then it may be  reasonable to transform the parking cost into a travel time equivalent (see
      Section 3.3.5.3).   '                    .              ,

      This analysis should be done  independently for each trip purpose and income strata^ or done with
      an equivalent income-sensitivity procedure (see Sections 3.3.6.3 and  3.3.6.4). If parking pricing
      varies by time of day, trip distribution analysis should be performed separately for each period.

U. S. Environmental Protection Agency                                                      •           3.7.

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions


     Users are cautioned that in typical gravity models, which are doubly constrained, the model tries to
     allocate  (through  iterations) the same number  of trip attractions  to each zone that  the  trip
     generation model has estimated would be attracted to that zone.  If the relative attractiveness  of a
     zone decreases because of an increased parking price, the  gravity model will still try to force a
     certain number of attractions to go to that zone.  This could cause unintended distortions in trip
     patterns. That problem could be minimized by including some type  of price-based variable (like
     composite  impedance) in the attraction model.  Another option is to  switch to a logit destination
     choice model to allocate attractions, as is done in Portland,  Oregon or San Francisco, for  example
     (see Section 3.3.5.4).                      »

     Areas without trip distribution modeling procedures will have difficulty accounting for the impacts
     that would result  from differential application of parking pricing.  In this case, there  are  two
     options:

               1.   Assume only applications of parking pricing  that would affect all areas at the same
                   level, and avoid applications to non-work travel; or

               2.   If the above option  is  not realistic, then clearly state assumptions  on how the
                   proposed implementation is expected to affect the pattern of regional trip origins
                   and destinations and how this is reflected in the estimated emissions reductions.

     6. Parking Pricing and Mode Choice                                                •

     Changes in choice of mode is the principal impact expected from  parking pricing.  All areas
     considering parking pricing are expected to furnish estimates of:

               •  Probable impact on number of person trips by mode; and
               •  Change in number of vehicle trips and average vehicle occupancy.

     Areas -with Mode Choice Models

     Areas with mode choice models in their planning systems are expected to (1) use them for  this
     analysis, or (2) explain to EPA their reasons for not doing so. In the absence of appropriate models,
     agencies should use " elasticity factor" or " quick response" analysis. Advanced analysis  includes
     disaggregating analysis by  purpose and income, applying separate coefficients for parking costs,
     and estimating shifts to non-motorized modes and non-travel.

     EPA offers the following guidelines for the best possible mode choice analysis results:

               •  Type of Model:   Multinomial  logit (or similar probability-based)  models  are
                  expected to provide the  most realistic estimates of impact; models with a " nested"
                  choice structure are among the most advanced of these (see Section 3.3.6.5 for more
                  information).   An ability to represent different automobile  occupancy levels as
                  "modes"  increases  accuracy in  estimating the  response  to  parking  charges,
                  particularly if special  exceptions in  parking fees are made for HOVs. An ability to
                  account for non-motorized modes and multi-modal (e.g. park and ride), options  will
                  increase realism of modeling, subject to the capabilities and information  furnished by
                  the overall model system.
3-8                                                                 U. S. Environmental Protection Agency

-------
                                                         Technical Methods for Analyzing Pricing Measures
                                                   .   •              -to Reduce Transportation Emissions'


               •  Parking Cost Coefficient:   Some models  carry a  separate coefficient value for
               -   parking cost, Which suggests that parking cost carries more weight than routine
                  operating cost (typically by a factor of 1.5 to 2.0). Agencies may wish to incorporate
*    •             such a. coefficient in their models (see Section 3.3.6.1),                        .
               •  Non-Work Mode Choice Models:  There should be  a mode choice model for each
                  separate trip purpose.  Some areas model only work trip mode choices and estimate
                  non-work mode choice through a "factoring"  process; this is not likely to yield a
                  credible result for non-work trips (see Section 3.3.6.4).
               • '  Income Stratification: Mode choice analysis should be performed separately for
                  different income levels (strata), both to allow for different cost sensitivity and to later
                  account for distributional impacts.  An acceptable alternative approach would be to
              :    base the analysis on  mode cost prices  divided by  traveler  income (see  Section
                  3.3.6.3):

               •  Program versus Effective Rates:   Agencies should not take it for granted that
                  travelers will bear the full weight of program-based increases  in parking prices.
                  Effective rates may be altered by .employer subsidies  or other discounts. Unless the
                  policy explicitly assumes full rate pass-through, some assumption should be  made
                  regarding effective rates paid per zone.  It also is desirable'to develop  models that
           ,'     .  consider the distribution of parking costs by site, rather than using zonal averages.
                  The STEP approach discussed in Appendix B offers such a capability.     •
               •  Cost Basis:  Mode choice modeling is typically based on the comparison of costs for
                  a one-way trip.  Thus,  it is generally appropriate to apply half of the daily parking to
                  each one-way work trip. For non-work trips, where rates  may be hourly, or for trip
                  chains with multiple destinations, the relevant price per trip may be less  clear and
                  analysis will require explicit assumptions.

               •  Adjustment for Inflation: The,dollar value implied  in the suggested parking price
                  must be consistent with the base year for which the^ model coefficient was developed.
                  Generally, this requires deflating the tested parking price with a time-inflation factor
                  that makes parking costs compatible with'the coefficient  in real terms (see Section
                  3.3.6.2).               .    :

   "           •   HOV Special Rates:  Few, if any, mode choice models permit testing of different
                  parking  prices for  different automobile occupancy  levels.  Such .distinctions  are
                  important if the proposed pricing  strategy provides incentives^ to carpools  or
                  vanpools. The TDM Evaluation Model (discussed in Section 3.3.6.9) may be useful
     .             to areas that wish to perform such an analysis'.
                                                         r .     '  •   • • -       .
     Areas without Mode Choice Models                              •  •         '

     Areas that do not have mode choice models will  have a more difficult time  assessing parking
     pricing impacts, but may still do so.  Suggestions for evaluations by these areas include:           ;

 1              •   Areas with Person Trip Tables: Areas with person-trip tables (by individual trip
                  purpose), can obtain an estimate of modal shifts by applying  one of the following
                  methods:

 -.',••''            ~  Quick Response Methods:  Mode choice analysis can be achieved with a
                         number  of techniques that use  sketch planning or pivot-point  methods.

U. S. Environmental Protection Agency                                ,  :                              o_o

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions


                          These include the FHWA's TDM Evaluation Model and the QRS software
                          system (see Section 3.3.6.9 for more information).

                       - "Elasticities:  Mode choice analysis can be approached through application of
                          pricing elasticity estimates from empirical studies or  from mode  choice
                          models of similar areas  (see  Section  3.3.6.9 for more information and
                          references).  It must be stressed that elasticities will vary widely from region .
                          to region, depending on base costs, travel times, urban-design  factors, and
                        .  mode shares as well as socio-economic factors.
               •  Areas with Vehicle Trip Tables (not Person Trip Tables): Some areas perform
                  analysis using only vehicle trip tables.  It is still possible,to evaluate mode  choice
                  responses to parking pricing if the agency first generates an estimate.of person trips
                  and then uses the derived estimates  with either of the methods identified  above.
                  Person  trips can  be derived  by (1) dividing vehicle trips by average  vehicle
                  occupancy,  or (2) using a technique such as the TDM Model to produce a starting
                  person trip table  (note that the TDM model does not include non-motorized trips,
                  however).

               •  Areas with No Trip Tables: Areas without trip tables have fewer analysis options,
                  but can develop at least a rough estimate of mode split response to parking pricing if
                  they  can  estimate  total person  trips, total  vehicle  trips  (or average  vehicle
                  occupancy), and VMT through HPMS  or another source.  Elasticity methods can
                  then be used to factor the base trip rates to reflect the change in parking.  Segmenting
                  the base population by average trip length, if possible, may increase accuracy.

     Generally, if an area's planning tools  and data are limited to trip tables  or gross trip and VMT
     measures, the  range of policy  options that can be  evaluated is also limited."  Only the most
     simplistic definitions can be considered, such as a uniform charge on all parking, which also raises
     questions as to realism of the assumptions and assessment of the impact (both locally and to  EPA).

     7. Parking Pricing and Time of Day
                                          '          '"••••"
     Parking pricing effects on time of day of travel are likely to be minimal in most cases; thus, no
     formal analysis is required.  Shifts in travel time of day are anticipated only if parking rates varied
  t   substantially by time of day, in which case the analysis should be treated more like congestion
     pricing (Section 3.2.5-7).  Presumably, most parking schemes directed at work travel would be in
     effect all-day, and thus traveling off-peak would offer no advantage.  Restructuring daily rates vis-
     a-vis hourly rates away from all-day discounts for commuters and toward more favorable short-
     term rates,  however, could induce non-work travelers to shift time patterns. Analysis of such
     effects is quite complicated and requires either fairly specialized tools or off-line methods (see
     Section 3.3.7).


     8. Parking Pricing and Traffic Assignment/Route Choice

     Parking pricing is likely to have limited or no direct effect on traffic assignment. Parking pricing
     applied in a special geographic pattern (e.g., downtown peripheral parking zones)-may, in some
     cases, generate micro-level  traffic  concerns.   If these threaten  to  be. significant,  then  traffic
     microsimulation procedures should be considered (see Section 3.3.8.3 for more  information).

3-10                                       '        '                JJ. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                      :'              to Reduce Transportation Emissions

      Areas with traffic assignment as part of their transportation  analysis  process are expected to
      perform a new assignment to reflect the impacts of mode choice and trip distribution analyses on
      vehicle trip patterns and their corresponding VMT levels in the travel network. These inputs are
      necessary for application of the emissions model in the next step. Areas lacking traffic assignment
      capability must rely on,net changes in vehicle trips and VMT to perform emissions analysis.

      9. Parking Pricing and Emissions

      All analyses of parking pricing should conclude with an estimate of the net changes in VOC, NOX,
      and CO. Areas with PMio attainment problems also should evaluate changes in anticipated PMio'
      emissions.  Parking pricing is not'expected to alter emissions rates or emissions analysis factors
      related to baseline vehicle mix.  However, because shorter trips are likely to be most affected by
      parking pricing, emissions analysis cold start factors should be adjusted or evaluated separately to
      account for changes.

      Areas with traffic assignment capability are advised  to use  results  from the  traffic assignment
      process in an emissions factor model, such as MOBILE or EMFAC (PARTS where appropriate), to
      generate estimates of emissions (see Section 3.3.9).  Areas without traffic assignment capability
      are limited to application of emissions  factors to net changes in vehicle  trips and VMT (also see
      Section 3.3.9), again accounting for possibly disproportionate changes  in cold starts.
U. S. Environmental Protection Agency
                                                                                             3-11

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
    3.2.2   Modal Subsidies
         3.2.2.1   Anticipated Impacts

     Modal subsidies are expected to have their
     primary impact on Mode Choice.   The
     reduced cost  of  the  subsidized  mode
     increases its attractiveness relative to all
     other modes, particularly driving  alone.
     Subsidies could include reductions in cost
     for  transit  users, rideshare  travelers,
     walkers, and cyclists.   Modal subsidies
     also   might   include   incentives   for
     alternative   work   arrangements  (e.g.,
     telecommuting, compressed work weeks).

     Impacts depend on the subsidy level (and
     its  net  value to the recipient), mode or
     modes  subsidized  (modal  competition),
     and   quality   of   the    alternatives
     (service/coverage).       Subsidies   can
     increase the attractiveness of destinations
     where the  alternatives are best deployed,
     which could  affect trip distribution  and
     long-term land use decisions.

     Subsidies   that    affect  peak/off-peak
     differences could  trigger time  of  day
     shifts.     Significant    subsidies  could
     possibly   influence vehicle   ownership
     decisions,  but  probably  more  as  a
     secondary result of household relocation.
Table 3.2 Impacts of Modal Subsidies
Analysis
, Step - „
Land Use/
Activity
Allocation
Vehicle
Ownership
Vehicle Mix
Trip
Generation
Trip
Distribution
Mode
Choice
Time of Day
Traffic
Assignment
Emissions
Impact Form
Areas where subsidy
applied can be more
attractive
May eliminate need
for multiple vehicles
None
Could induce non-
motorized trips or
alternate work
schedules
Destinations
reachable by
subsidized mode
more attractive
Encourages use of
subsidized modes
i ~ f*'
Shift trips to periods
with most subsidy
benefit
Shift in trips and
VMT as result of
mode choice &
distribution
Reduce vehicle trips
and VMT
Impact
Type
Trace
(Secondary
level if
significant)
Trace
None
Trace
Secondary
Primary -
^ "*
Secondary
(if time of
day differs)
Derivative
effect
Derivative
effect
Impact Indicator
Population and/or
employment by zone
Vehicles per
household per zone
Percent of vehicles by
VMT
Number of person
trips to/from each
zone
O/D patterns (average
trip length)
Person trips by
mode; Sfrtransjt;^ ^
no vehicle trip*;*
Average vehicle
occupancy
Trips by time period
Link volumes; VMT
Tons or kg. of VOC,
NOX, CO, PM
     3.2.2.2     Basic Analysis for Modal Subsidies

     In addition to the standard analysis of emissions, agencies should carefully assess the impacts on
     mode choice (person trips by mode) and, in turn, mode choice effects on overall vehicle trips,
     VMT, and emissions.  If subsidies are mode-oriented, submitting agencies should  investigate the
     effects on other modes.  If subsidies favor particular locations, in terms of modal access or other
     amenities, agencies also should address the indirect impacts on trip distribution and land use:
3-12
                                                                   U. S. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                                                                       to Reduce Transportation Emissions
       3.2.2.3  Guidelines for Analysis
        1. Modal Subsidies and Land Use/Activity Allocations

        Modal subsidies  are unlikely to  affect long-run  locational  patterns unless they are substantial,
        considered permanent,  and linked to concurrent investment  in alternative mode  and/or land
        use/urban design strategies that complement the subsidy; these latter also would add to the sense of
        "permanence." EPA does not require a formal analysis of subsidy impacts on land use unless the
        nature of the subsidy suggests an obvious impact on long-term trends; in this  latter case, EPA
        requires a statement of possible effects.                                               ''

        2. Modal Subsidies and Vehicle Ownership        ;

        Modal subsidies are.expected to have little or no direct impact on vehicle ownership and, thus, no
        formal analysis is required.  If an agency feels its subsidy program will directly affect vehicle
        ownership levels, the agency should present  appropriate assumptions or analysis to  support this
        claim.                          ,

        There may be an indirect effect on vehicle ownership, however, if modal subsidies are expected to
        affect land use through household shifts to areas with lower  rates. Advanced analysis could then
        include a re-run of vehicle ownership evaluation to reflect anticipated land use impacts.

        3. Modal Subsidies and Vehicle Mix
                                     "                             '            _'-("•,*
        Modal subsidies are expected to have little or no impact on vehicle mix and,  thus, no formal
        analysis is required.  Since mpdal subsidies can. have differential effects on  different trip types and
        lengths, however, shifts in usage patterns may be important in determining the regional fleet mix
        used in the emissions estimation step                                     '


        4. Modal Subsidies and Trip Generation

       It is unlikely that  modal subsides  will have a measurable impact on the overall number of person
        trips generated by households.  However, an analytical problem arises if  trip generation is only
       performed for vehicular trips since the overall  vehicle trip reduction effects  will be understated if a,
        subsidy draws travelers from private vehicle to transit, non-motorized or telecommuting modes.

       Modal subsidies might. influence household-trip generation, and estimates of trip generation,
       through:                           .                     •  •  .                      •

                 •  , Shifts to or from non-motorized modes, if those modes are not reflected in trip
                   generation estimates; and/or                          -            .

                •  Attraction  of non-work trips to  subsidized transit that may have not been  made
                   previously,  or  were made by npn-motorized mode (and not included in the trip
                   generation rate).                                 -."-_.
'  U. S. Environmental Protection Agency
                                                                                              3-13

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
    *             *               '                   ' '

      While a  formal  analysis  is not required, the performing  agency may wish to estimate modal
      subsidy impact on trip generation if the proposed subsidy:

               •   Stimulates alternative work arrangements;
               •   Involves transit or non-motorized modes; and
               •   Is  set at a significant level (e.g., free off-peak transit).

      While there are no known formal methods (i.e., methods typically included in a complete regional
      travel demand modeling analysis) for dealing with these impacts, the user may consider off-line or
      sketch planning techniques such as described in Section 3.3.6.6 (Special Modes) and 3.3.6.9 (Off-
      line Methods).  If the agency does not perform an analysis, it should offer an assessment of the
      anticipated impact of the subsidy on vehicle trip generation.

                                    /
      5. Modal Subsidies and Trip Distribution

      Subsidies that favor travel to particular destinations can affect destination choices. The effects on
      destination chqice for work-related trips would only be realized in the longer term; the effects for
      non-work trips, however, could well be realized in the shorter term.  Trip lengths also  may be
      affected, depending on the nature of the subsidy (whether it is fixed or distance-based).

      As these effects are likely to be secondary, a formal  analysis is not required.  If subsidies are
      significant, however, agencies should provide a statement of possible effects on trip distribution.
      Agencies with trip distribution models can  perform  ah analysis of modal subsidy  effects on
      destination choices.

      Models that incorporate a composite impedance measure (see Section 3.3.5.1) can estimate the
      effect of most subsidies  directly,  by altering  the respective generalized travel costs for each
      affected origin-destination pair.   This  approach  is generally  effective for motorized  modes.
      Current techniques,  however, do not deal well with subsidies that greatly influence the choice
      between non-motorized and motorized mode destination choices.2

      If subsidies vary by destination, a different pattern of destination choices in the trip table would be
      expected to  result.  Analysis should be  disaggregated by trip purpose  and should account for
      differences in income though income stratification or other techniques (see Section 3.3.5.2).

      Areas without trip distribution models of the type described here will not be able to perform this
      analysis.  These areas are advised to exercise caution when interpreting the modal choice effects of
      subsidy policies  and should state  how  failure to account for changes in trip distribution might
      affect their final analysis.
     1 It would be necessary to incorporate measures of bike/pedestrian generalized cost measures into the models.
       Initially, such models would need to be estimated where there is a large difference in non-motorized mode
       utility  and  use, with a transfer of coefficients to regions  lacking such diversity, using  choice model
       coefficient scaling (see section 3.3.6.8).
3-14                                                  •               Ui S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                                     to Reduce Transportation Emissions
      6. Modal Subsidies and Mode Choice
     Because modal subsidies  are expected  to  have their greatest impact on mode choice,  formal
     analysis should be used. All areas considering modal subsidies should furnish an estimate of the
     change in the number of person trips by  mode, the change in number of vehicle trips and average
     vehicle occupancy, and the change in the  percent of travelers using transit as a result of the
     particular policy.

     Modal  subsidies  alter the  relative  cost and attractiveness between subsidized and unsubsidized
     modes, thus altering competition among the  modes for person trips.  Modal subsidies are expected
 v   to have limited or no impacts on commercial travel or through travel.  Because the relationships
     are complex, areas with mode choice models in their planning systems are encouraged to use them
     for this analysis.                                                              -

     EPA offers the following guidelines to help agencies develop the best possible mode  choice
     analysis:

     Areas-with Mode Choice Models

               •  Type of Model:  Multinomial logit (or similar probability-based) choice models will
                  provide the best estimates of impact, provided the relevant modes are included in the
                  modelstructure.  Nested models are expected to provide the  most accurate estimates
                  (see Section 3.3.6.5).                     .

               •  Suitable Modes/Coefficients:   Since  most  mode  choice models include  cost
                  relationships for transit fare and automobile operating cost, they can accommodate
                  many subsidy policies.  Coefficients' suitability is addressed in Section 3.3.6.1.  If
                  subsidies for HOV travel involve parking discounts, then use of a parking coefficient
                  may be appropriate (see Sections 3.3.6.9 and 3.2.1.6).

               •  Accommodation of Modes: In order to test the full range of modal pricing options,
                  the  mode choice model should incorporate different costs for each sub-mode (e.g.,
                  walk to  transit, drive to transit, drive alone, 2-person carpool, 3-person carpool, 4-
                  persori carpool,  and  perhaps  even vanpool,  walk and  bike).  Not  all models
                  accommodate  such refinements.   Alternatives  include use of the TDM Evaluation
                  Model (Section 3.3.6.9), which allows subsidies to be targeted individually to various
                  carpool levels, vanpools, and transit, and distinguishes between direct out-of-pocket
                  and less direct per-mile costs.  Most model systems (including the TDM Model) do
                  not accommodate non-motorized modes.  Areas considering subsidies to these modes
                  will need to consider off-line methods for their evaluation (see Sections 3 366 and
                  3.3.6.9).

               •  Non-Work Mode Choice Models: Mode choice analysis of modal subsidy impacts
                  should be performed  separately for each separate trip purpose;  use of factoring
                  methods is discouraged,(see Section 3.3.6.4).

            '  •  Income  Stratification:  The performing  agency should  perform  mode choice
                  analysis separately for each  income stratum, or use an alternative  approach to
                  address income group differences in sensitivity and impact (see Section 3.3.6.3).
               •  Cost Basis: Costs should'be specified on the basis of a (daily) one-way trip.

U. S. Environmental Protection Agency                '                                               3_I5

-------
Technical Methods for Analyzing Pricing i
to Reduce Transportation Emissions  "


               •  Adjustment for Inflation:  The dollar value implied by the proposed subsidy should
                  be deflated for  consistency to model  coefficient's  year of calibration (see Section
                  3.3.6.2).

               •  Special Policy Definitions:  Certain subsidies may be difficult to test with existing
                  models because they may require specialized information; for example, parking cash
                  out requires knowing whether the commuter currently has free parking, whether the
                  employer leased the parking, and what the value would be to the employee back as a
                  subsidy.   Policies of this  nature  can  sometimes  be addressed through sample
                  enumeration methods (see Section 3.3.6.7) or use of stated preference methods (see
                  Section 3.3.6.8).

        Areas -without Mode Choice Models

     Areas that do not have mode choice models will have a more  difficult time assessing subsidy
     policies, but may still do so. Suggestions for these areas are as follows:

               • Areas with Person Trip Tables:  If agencies have access to trip tables based on
                 person trips (by individual  trip purpose),  they can estimate the impact of modal
                 subsidies on modal shifts through application of one  of the following methods:

                      -  Quick Response Methods:  Mode  choice analysis can  be achieved with a
                         number  of techniques that use  sketch  planning or pivot-point  methods.
                         These include the FHWA's  TDM Evaluation Model and the QRS  software
                         system (see Section 3.3.6.9 for more information).

                      -  Elasticities: Mode choice analysis can be approached through application of
                         pricing elasticity  estimates  from empirical  studies or from mode choice
                         models  of similar areas (see  Section  3.3.6.9 for  more information  and
                         references). It must  be stressed that elasticities will vary widely from region
                         to region, depending on base costs, travel times, and mode shares as well as
                         socio-economic factors.
                                             '             '"         '.-                  '
              •  Areas with Vehicle Trip Tables (not Person Trip Tables):  Some areas  perform
                 analysis using vehicle trip tables only. It is still possible to evaluate mode choice
                 responses to parking pricing if the agency first generates an estimate  of person trips
                 and then uses the derived estimates with  either of the methods identified above.
                 Person  trips  can  be  derived by (1)  dividing vehicle trips by average vehicle
                 occupancy, or (2)  using a technique such as the TDM Model to produce a starting
                 person trip table (note that the TDM model does not include non-motorized trips,
                 however).

              •  Areas with No Trip Tables: Areas without trip tables have fewer analysis  options,
                 but can develop  at least a rough estimate of mode split response to parking if they
                 can estimate total person trips, total vehicle trips (or average vehicle occupancy), and
                 VMT through HPMS or  another source.   Elasticity methods can then  be  used to
                 factor the base  trip rates  to reflect the  change in  parking.  Segmenting the base
                 population by average trip  length, if possible, may increase accuracy.

    Generally, areas that are limited to trip tables or gross trip and VMT measures can evaluate only a
    limited range of policy options and consider only the most simplistic policy definitions.


                                                                  U. S. Environmental Protection Agency

-------
                                                         Technical Methods for Analyzing Pricing Measures
                                                                    to Reduce Transportation Emissions

     In all cases, more advanced analysis includes the  use of separate models for each purpose,
     disaggregated by income levels, and an accompanying supplemental analysis to assess the effects
     on the use of subsidized or non-motorized modes.


     7. Modal Subsidies and Time of Day

     Modal subsidies can have secondary effects on time of day of travel, but, in most cases, the effects
     are expected to be minimal and formal analysis is not required.

     Subsidies that entail major peak/off-peak differentials may trigger migration into or out of peak
     periods, particularly for non-Work trips. Since ridesharing is almost exclusively a commuter mode,
     ridesharing subsidies are not expected to produce much of a shift in time of day unless time of day
     differentials include off-peak periods close to conventional work day beginning and end.

     Subsidy-induced differentials in transit pricing by time of day, however, could produce travel time
     shifts—lower off-peak fares would be expected to attract non-work riders and a small number of
     workers, while relatively lower peak fares might draw some non-work trips into the peak, as well
     as some work-based transit trips.                    .

     If time of day impacts are considered possible,  methods such as those described in Section 3.3.7
     may be of use.  In the particular case of transit, time of day  demand information may be available
     through studies (possibly local) of transit peak/off-peak fare adjustments.


     8. Modal Subsidies and Traffic Assignment

     Modal subsidies are not expected to exert any special influence on traffic assignment. Areas that
     have traffic assignment as part of their transportation analysis process, however, should perform a
     new assignment analysis to reflect the impacts of mode shifts on vehicle trip and VMT levels, and
     to accommodate these changes in the travel network pattern.  These inputs are necessary for
     application of the emissions model in the next step.

     Areas which do not have traffic assignment capability must rely on net changes in vehicle trips and
     VMT to perform emissions analysis.

     If HOV  subsidies  are  high enough,  demand  may  justify new  or expanded  HQV facilities.
     Regarding analytical procedures, this would require some additional effort in network coding and
     assignment procedures;  however, this  would be a categorically different policy from a. straight
     HOV subsidy and would argue for a formal HOV analysis.

     9. Modal Subsidies and Emissions
                                            : -         ••         /
     All analyses of modal  subsidies should conclude with an estimate of the net changes in the major
     pollutants for which they are claiming credit.  Modal subsidies are not expected to affect emissions
     through emissions rates or through changes in regional vehicle fleet distribution; thus, no special
     attention need be paid to these impacts in the emissions analysis process.
U. S. Environmental Protection Agency              :    .,                                            3-17

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions


      Areas with  traffic  assignment capabilities are strongly encouraged to use an emissions factor
      model, such as MOBILE or EMFAC or PARTS, to furnish estimates of emissions, using the results
      from the traffic assignment step (see Section 3.3.9).

      Areas without traffic assignment capability are limited to application of emissions factors to net
      changes in vehicle trips and VMT (see Section 3.3.9), accounting for the possibly disproportionate
      changes in cold starts.
3-18
                                                                     U. S. Environmental Protection Agency

-------
     3.2.3   Pump Charges
                                                         Technical Methods for Analyzing Pricing Measures
                                                                    to Reduce Transportation Emissions
         3.2.3.1   Anticipated Impacts

     Pump charges (or similar VMT-based fees)
     increase the per-mile cost of driving. This
     is expected  to  reduce the  demand for
     .private  vehicle   travel.    The  expected
     primary impact is on mode choice, though
     significant price increases also could affect
     trip  distribution,,  long-term  locational
     decisions  (land  use),  and possibly  trip
     generation.  Households also  are expected
     to'react to higher fuel prices by shifting to
     ownership   and/or  use  of  higher-mpg
     vehicles, thus affecting vehicle mix.
Table 3.3 Impacts of Pump Charges
', Analysis -
'*"-Step
Land Use/
Activity
Allocation
Vehicle
Ownership
Vehicle Mix
Trip
Generation
Trip
Distribution
,«ode'r; --
Choice
,, ,^.
Time of Day
Traffic
Assignment
Emission
Analysis
-• Impact Form "
Causes location in
less auto-dependent
areas
If shifts in location
encouraged
Shift to- higher mpg -
vehicles;:' "
Overall higher cost of
travel
Causes shorter
trips/closer
destinations
Decreases private '
vehicle use ~ *'/ *
*> V"
. J^j? ^ -
None
Change in trips and
VMT; may affect
speeds
Reduces vehicle
trips and VMT; may •
affect'speeds
'jJmpsicfl'l
- Ty|ie\-
Secondary
Trace
^Primary -
1/1 >,
Secondary
Secondary
Primary ' ,
None
Derivative
effect
Derivative
effect
(possible
change in
vehicle mix)
Impact Indicator
V" , ~
Population and/or
employment by zone
Vehicles per
household per zone
Percent of Vehicles
JSyVMT " ';,<•,,
Number of person
trips to/from each
zone
O/D patterns (average
trip length)
Person'trips hy*~ ,
mode; % transit} •
no vehicle trips;
Average vehicle '
occupancy^ ^ ^"f
Trips by time period
Link volumes; VMT
Tons or kg. of VOC,
NOX, CO, PM
     3.2.3.2  Basic Analysis for Pump Charges

     Aside from the standard analysis of emissions, analysis of pump charges should include mode
     choice (person trips by mode) and the effect of mode choice shifts on overall vehicle trips, VMT,
     and emissions.  Agencies also should address the implications of a potential shift to higher-mpg
     vehicles  on effective operating cost and vehicle mix (which,  in  turn, affects factors used  in
     emissions analysis).


     3.2.3.3  Guidelines for Analysis


     1. Pump Charges and Land Use/Activity Allocations

     Pump charges-(or VMT fees) influence long-run  locational  patterns by favoring locations that
     support reduced dependence on private vehicles (e.g.,  better walk access, availability of transit,
     etc.). Because this is designated as a secondary impact,  a formal analysis is not required; however^
     EPA expects some acknowledgment of the long-range impacts. •         •
U. \ Environmental Protection Agency
                                                                                           3-19

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions


     If pump charges (or VMT fees) are significant, they may influence land use/locational effects.
     Although a formal analysis is not required, any agency anticipating major changes in pump prices
     should submit at least a statement of likely long-term  impacts on  regional development trends,
     travel, and emissions.  The agency may or may not support this statement with a more rigorous
     analysis of anticipated impacts.

     Areas with land use allocation  models (see  Section 3.3.1.2) that incorporate transportation cost
     through some measure  like composite impedance (see  Section 3.3.5.1) may wish  to use  these
     models to investigate potential  land  use impacts as part of their  advanced  analysis.  Another
     approach would be to consider one or more alternative  growth scenarios that: (1) are based on a
     system  of higher travel costs,  (2) incorporate alternative  transportation investments,  and (3)
     include various land use/urban design elements (Section 3.3.1.1).


     2. Pump Charges and Vehicle Ownership

     Pump charges (or VMT fees) are unlikely to directly affect  overall levels of vehicle ownership
     and, thus, no formal analysis is required.  If shifts in land use/locational patterns are projected in
     Step 1, however, then these changes should be carried forward into a revised assessment,of vehicle
     ownership.                   •


     3. Pump Charges and Vehicle Mix

     Given  the primary impacts anticipated, submitting agencies should indicate their assumptions
     regarding the effects of pump charges on vehicle mix.

     Assuming that program-based pump charges are proportional to gallons of fuel purchased (rather
     than set as fixed fees, per transaction or based on recorded VMT), households can be expected to
     respond by gradually shifting ownership and/or use to higher-mpg vehicles.3  Shifts to higher-mpg
     vehicles could have two effects:

               1.  Offset some of the gains from pump-based charges  by reducing cost-per-mile of
                  surcharges; and

               2.  Shift  the  mix of vehicles  reflected in the  emissions model,  thus  the  effective
                  emissions  rate.

     If the increase in fuel price will be significant, the performing agency should assess the shift in
     ownership and usage to higher-mpg vehicles.  This assessment could employ:

              . •  Forecasting tools, such  as  hedonic or other vehicle share models (see Section
                  3.3.3.1);
               •  Scrappage models and studies (see Section 3.3.3.2);
               •  Empirical/research studies or trend data (see Section 3.3.3.3);
     3 If ongoing emissions charges are pure VMT-based charges (using smart-card or other mileage tracking
      technology) where fuel consumption is not used as a proxy or intermediate variable, there is no incentive to
      shift to higher-mpg vehicles.
3-20       .                                                         U. S. Environmental Protection Agency

-------
                                                           . Technical Methods for Analyzing Pricing Measures
                                                                       'to Reduce Transportation Emissions


                •  Stated preference analyses (see Section 3.3.3.4); and

                •  Elasticity methods (see Appendix A discussion related to Vehicle Registration Fees).

      Clearly, this assessment has applicability beyond those agencies that employ vehicle ownership as
      a formal part of their planning system, since this assessment gauges the changes in average mpg,
      cost per mile of driving, and vehicle distribution variables used for emissions analysis.


      4. Pump Charges and Trip Generation

     .Because pump charges increase the overall cost of vehicle usage for all trips regardless of purpose,
      they.could have an  impact on the number of activities that households support with trips, i.e., trip
      generation.  More likely,  however, households would adjust trip characteristics without altering
      the number of trips, for example, by favoring other modes, shorter  trips,  and increased trip
      chaining.  Thus, formal analysis of the impact of pump charges (or VMT fees) on trip generation is
      not needed.
      In general, EPA encourages agencies to do the following:

               •   Re-assess trip generation 'if land use or vehicle ownership shifts are projected  in
                   earlier steps and the agencies have trip generation models based on these inputs;
                   and/or                            ,

               •   Perform an assessment of the impact of increased pump charges (or VMT fees) on
                   trip  generation  if the agencies  have trip generation models able to reflect cost
                   through a measure such as composite  /mpedowce,(see Section 3.3.5.1).4

      Agencies basing trip generation on vehicle serviceable or vehicle only person trips rather than all
      person trips may face concerns regarding accuracy in this step since higher fuel prices can shift
      trips to non-motorized modes.  These agencies may use off-line methods to approximate mode-
      choice shifts (see discussion below in sub-section 6). EPA encourages these agencies to upgrade
      their models to deal specifically with non-motorized trips in the future.

      5. Pump Charges and Trip Distribution

      Vehicle per mile charges will have a secondary effect on the  distribution of trips.  By  increasing
      the effective separation between zones, longer-distance trips will have a higher price and thus be
      discouraged. The impact on work trips is expected to be long term, while the impact on non-work
      trips could be realized in the near term.
      A difficulty in modeling the influence of an increase in the cost per mile of automobile travel is that at the
      trip generation stage of a four-step model, there is no specific trip length from which to gauge the actual
      increase in the user cost of travel.  Also, for transit users and bicyclists, there would be no increased  user.
      cost. One way around this is to use the mode choice model to calculate a composite accessibility variable
      which reflects changes in the generalized price of reaching all destinations via all modes.  There are few
      current trip generation models that are.sensitive to user cost (see.Section 3.3.4.1.).
U. S. Environmental Protection Agency

-------
Technical ffethods for Analyzing'Pricing Measures
to Reduce Transportation Emissions


     Although formal analysis is not needed, submitting agencies should submit a statement of possible
     effects.  Agencies that anticipate significant increases in pump prices as a result of the proposed
     policy should perform a trip distribution analysis in relation to the higher fuel prices if they have
     trip distribution models that:

               •  Accommodate costs through  measures  such  as composite impedance  (Section
                  3.3.5.1); or

               •  Incorporate travel time approximation techniques (Section 3.2.5.3.).

     Where possible, this analysis should be performed separately by trip purpose and income level.  In
     addition, if the  price per gallon is likely to trigger a shift to higher-mpg vehicles, agencies should
     consider the effect of mis shift on cost per mile.

     Areas that do not have trip distribution model capability, or that anticipate only insignificant changes
     in prices, should document their assumptions and effects on final results.

                                                                                         .r
     6. Pump Charges and Mode Choice

     Because pump  charges (and VMT fees) are expected to have a primary effect on mode choice, a
     formal assessment of anticipated effects should be conducted.

     Pump charges (or VMT fees) increase the per-mile cost of private vehicle travel both absolutely
     and relative to  other modes; this should increase the demand for alternative modes.  Transit and
     non-motorized  modes  would be exempt from price increases, thus improving their  comparative
     cost position. Although rideshare modes would be assessed as private vehicles, this  mode would
     still enjoy a cost advantage over solo driving since the increased price would be distributed among
     several travelers.   Although the comparative  cost positions  of bike  and walk modes also are
     improved,  these modes may be  less  affected because of the short trip length (small  total cost
     impact for competing automobile trips). However, significantly higher fuel prices could generate
     more non-motorized  trips to closer  destinations. Commercial vehicle trips could be affected
     depending on the nature of the business activity.

     Areas -with Mode Choice Models

     Areas with mode choice models should use them to evaluate mode choice effects of pump prices
     (see Section 3.3.6  for model  characteristics).  In  using these  models, agencies should increase
     operating costs for all private vehicle modes by the  increase in cost  per mile times the origin-
     destination trip length as determined from the highway distance matrix. The following guidelines
     should be applied to this analysis:

               •  Calculation of Cost Impact:  For a VMT fee, the increase in cost per mile is direct.
                  For pump charges, the incremental cost per mile must be determined by first dividing
                  the  change in price per gallon by the average mpg of the regional vehicle fleet.  In
                  the  event of significant price increases yielding a shift  in the vehicle mix toward
                . higher mpg vehicles, the revised average mpg should be  used to estimate the tested
                  per-mile cost impact.
3-22                             .                                  U. S. Environmental Protection Agency

-------
                                                         Technical Methods for Analyzing Pricing Measures
      •                                                       •        to Reduce Transportation Emissions

               • . Control for Inflation:  Agencies should ensure that the cost increase input to the
                .  model has been inflation-adjusted to match the year of calibration of the  model
                  coefficient (see Section 3.3.6.2).

               •  Income Stratification:  The mode choice analysis should be performed separately
                ' byincome strata, or with an acceptable alternative approach (see Section 3.3.6.3).
               •  Trip Purpose Distinctions:  Separate mode choice models should be applied for
                  each trip purpose (see Section 3.3.6.4).               ,

     Areas without Mode Choice Models            ,

     Areas without mode choice models may estimate mode choice impacts'from pump charge or VMT
     fee policies using less formal methods.  Suggestions for these areas include:

               •  Quick Response Methods, which include computer software  methods  like  the
               .   FHWA  TDM Model and the QRS system (particularly useful for areas with trip
                  tables); and/or

               •  Elasticity and Sketch Planning Methods,  which  include  various pivot-point
                  methods and elasticity estimates that may be applied to either a trip table base or
               ' ' more aggregated travel estimates.

     These techniques and their application are detailed in Section  3.3,6.9. Areas that do not formally
     account for non-motorized modes in mode choice may wish to do so by considering one of the off-
     line methods listed in  Section 3.3..6.9.    .    /               •

     Areas without formal mode choice models or trip tables are cautioned that their analyses may have
     questionable  accuracy with respect to differences among travelers regarding  trip purpose, trip
     length, modal alternatives, and income level.                      ,                     .


     7. Pump Charges and Time of Day

     There could be a second^order effect on peak/off-peak travel distribution to the extent that total
     vehicle trips, VMT, and congestion are reduced by pump charges (or VMT fees).  This effect,
   .  however, is not expected to be measurable; and no analysis is needed.                           '


     8. Pump Charges and Traffic Assignment

     Most of the impact from pump charges (or VMT fees), on traffic assignment is likely to come from
     reductions in length and number of vehicle trips.-  However, increases  in  the per-mile cost of
     driving also may affect route selection for given origin-destination pairs.  If the additional cost is
     high enough, it will encourage drivers to select paths that minimize cost as well as time (e.g.,
     selecting shorter but less speedy routes).  In general, these effects are expected to be at the trace
     level, and standard assignment  analysis is expected only of those agencies with  assignment
     procedures.

     Agencies with appropriate capabilities can estimate traffic assignment impacts of pump charges by
     building highway paths  based on a combination of travel time and user cost, using the per-mile


U. S. Environmental Protection Agency   '        '    ,~!     '    ~'.    2-23

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
      user cost of vehicle operation and an assumed average value of travel time to motorists. All major
      software packages are believed to provide this capability.5

      Areas without traffic assignment capability must rely on net changes in vehicle trips and VMT to
      perform emissions analysis.


      9. Pump Charges and Emissions
                                               jN-

      All analyses of pump charges (or VMT fees) should conclude with an estimate of net changes in
      VOC, NOX, and CO (and PMjo for areas with PMjo problems).  The impacts on emissions  are
      expected to derive  from projected changes in vehicle trips and VMT (without shifts in emissions
      rates), and shifts in mpg and possibly speeds from the traffic assignment step (with concomitant
      changes in emissions rates).

      Areas with traffic assignment capability and with emissions  factor models, such as MOBILE or
      EMFAC, are expected to use these procedures to estimate the emissions changes.  PART 5 should
      be used if particulates are a  problem.   If analysis  in Step 3 indicates a change in the  regional
      vehicle fleet mix, then  the agency should use the new vehicle distribution results to alter the
      distribution in the MOBILE or EMFAC model and resultant emissions factors (see Section 3.3.9).

      Areas without emissions model, capability, that  is,  areas that rely strictly on emissions  factor
      estimates, should detail their assumptions and the procedures they used if they have modified their
      base emissions rates to reflect the shift in vehicle fleet mix.
       It may be possible to make this analysis sensitive to income groups by first building (skim and assign)
      separate paths for each income group, so that low-income drivers are assumed to use routes that minimize
      vehicle operating cost and high-income drivers to use routes that minimize travel time.
3-24                                                               .  U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                                     to Reduce Transportation Emissions
      3.2.4   Emissions Fees
         3.2.4.1   Anticipated Impacts

      Emissions fees, as defined for this guidance, consist of increases in vehicle registration fees based
      on some combination of vehicle emissions characteristics (age and/or emissions rate) and usage
      (odometer reading). The emissions element rather than VMT is presumed to be dominant; that is
      emissions would account for the greatest variation in the level of the fee: (-VMT-based fees are
      addressed as a variation of pump charges in Section 2.2.3.-)
      The  primary   impact  anticipated  is
      increasing ownership and greater usage
      of newer,  low-emitting  vehicles.   If
      emissions fees are significant, they also
      could  affect  the number  of vehicles
      owned.  In turn, this could influence
      land use/locational patterns, which then
    " affects trip generation.

     If  emissions    fees    are   assessed
     infrequently, it is  unlikely that  they
     would  have direct  effects  on  travel
     unless education campaigns make people
     aware of the charges.
Table 3.4 Impacts of Emission Fees
^Analysis *
- '$t»p :
Land Use/
Activity
Allocation
Vehicle '71
* **a £, 4? <
Ownership «
Vehicle Mix
V !?»•;, "
Trip
Generation
Trip
Distribution
Mode
Choice
Time of Day
Traffic
Assignment

, Impact Form
r* f, '"£:•* '!**
Lowers demand for
auto-dependent
' environments
Reduces demand
for vehicle " <'-c """
ownerslftp' , ~i'
^Increases ** !
ownership, & use
of newer/ tow^ >
Remitting vehicles •*/
If vehicle ownership
affected
If vehicle ownership
affected
If vehicle ownership
affected '
None
None
'.:ChaijjeSi®t6le'-
SrroffemaSWancC..
impact
., ... TSP*"
Secondary
(if vehicle
ownership
is affected.)
, Primary
s Primary
•V ^
Trace
Trace
Trace -
None
None
Primary "•> -~
^ Impact Indicator „
^ -> ~ *=v®^
Population and/or
employment by zone
Vehicles per household ,f
jparzone J " - -
^ 'V ^^
Percent of vehicles by
«MT- , '
" <* ". ™Z-
Number of person trips
to/from each zone
O/D patterns (average
trip length)
Person trips by mode;
% transit; no vehicle
trips; average vehicle
occupancy
Trips by time period
Link volumes; VMT
Tons or, kg. of VOC,
NOXt COJPM - - - „
     3.2.4.2  Basic Analysis

     Agencies   proposing   to   introduce
     emissions  fees  should   analyze  the
     effects  of this  measure  on regional vehicle mix and  the  subsequent effects  on emissions
     attributable to a change in vehicle mix.  Additional assessments should be used if significant
     impacts are anticipated for vehicle ownership and, from that, for land use and other secondary
     effects.
U. S. Environmental Protection Agency
                                                                                            3-25

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
      3.2.4.3  Guidelines for Analysis


      1.  Emissions Fees and Land Use/Activity Allocations

      Two instances in which emissions fees could trigger land use/locational impacts, are:

                1. Major VMT component in fee: If the emissions fee has a major VMT component, the
                  cost per mile for vehicle use could deter vehicle use.  In this case, analysis should
                  follow the guidelines for pump charges as well (Section 3.2.3).

                2. Significant increase in automobile ownership costs: If the emissions fee significantly
                  increases the cost of vehicle ownership, it may deter vehicle ownership.  A reduction
                  in household vehicle  ownership could increase  the demand  for  location-efficient
                  areas.
      If the nature or magnitude of emissions fees are considered sufficient to affect vehicle ownership
      rates, EPA suggests that the performing agency address the potential impact on land use/locational
      patterns with a land  use  allocation model, if one is available and suitable, or  through scenario
      testing.   Agencies should  prepare a statement that reflects the agency's assumptions of the
      potential impacts on long term  location patterns and the overall impact on emissions if the agency
      (1) does not have appropriate analysis tools, or (2) elects to declare this impact to  be insignificant.


      2. Emissions Fees and Vehicle Ownership

   .   Emissions fees can trigger a reduction in vehicle ownership if the fees cause a significant increase
      in costs of  vehicle  ownership.  The  performing agency should account  for  shifts  in  vehicle
      ownership if the agency (1) thinks that its proposed emission fee may have an impact on vehicle
      ownership levels and (2)  uses  vehicle ownership in its modeling process.  The agency may use
      analytic methods or the best possible assumptions. Techniques suitable for this analysis include:

               •  Forecasting  tools, such  as hedonic or other  vehicle share  models (see  Section
                  3.3.3.1);

               •  Scrappage models and studies (see Section 3.3.3.2);

               •  Empirical/research studies or trend data (see Section 3.3.3.3);
               •  Stated preference analyses (see Section 3.3.3.4); and

               •  Elasticity methods (see Appendix A discussion related to vehicle registration fees).

      3. Emissions Fees and Vehicle Mix

      Emissions fees are expected to have their primary effect  on vehicle mix.  Households would be
      expected to either replace high-emitting vehicles with cleaner ones, and/or shift  use within multi-
      vehicle households in favor of low-emitting vehicles.

      In some circumstances, such as in  a revenue-neutral feebate approach, an emissions fee program
      could actually increase the  number of motor vehicles by  encouraging diversification in favor of
      low-emissions vehicles.   This  would  be  in contrast to recent trends toward vehicles with  high

3-26                                                     ;           U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                          •                                           to Reduce Transportation Emissions


     weight/power/performance characteristics. If the agency's proposed strategy is likely to encourage
     net expansion in vehicle fleet size, the agency should address the effect of its proposed emissions
     .fee program on overall vehicle ownership levels as well as on distribution.

     'Emissions fees also may affect the composition or.use of commercial vehicle fleets registered in
     the affected jurisdiction.  Emissions fees would not, however, affect through traffic or external
     travel vehicles.

     Section 3.3.3 provides options that may be used to estimate shifts in vehicle mix.

     4. Emissions Fees and Trip Generation

     In general, emissions fees are expected to have only trace effects on trip generation. (Recall that
     infrequent  assessment  of  emissions  fees would mean  limited  direct impact  on  daily travel
     decisions.).  Nonetheless, emission fee'effects on the costs of vehicle ownership could lower the
     number of vehicles owned by households.  In turn, this  could affect trip generation analysis in
     jurisdictions that use  models based on vehicle ownership. These agencies may wish to estimate
     the impacts on trip generation of changes in vehicle ownership.

     Higher annual costs of ownership and operation of high-emitting vehicles also could discourage
     their use in favor of low-emitting vehicles in multi-vehicle households, thereby altering household
     interactions and travel decisionmaking. This could generate more errand chaining, and/or use of
     non-motorized modes, thus affecting trip generation for motorized trips and trip characteristics. It
     also could  induce various  other  effects that  influence mode choice, destination, or trip timing.
     Formal  procedures to evaluate these impacts are not  in common use; off-line techniques are the
     only realistic approach for addressing them (see recommendation under emissions fees and vehicle
     ownership above).


     5. Emissions Fees and Trip Distribution

     Emission fees  are expected to have only trace effects on trip distribution.  Although formal
     analysis is therefore not needed,  some investigation of effects should be undertaken by agencies
     anticipating impacts on land use and vehicle ownership and by agencies contemplating fees with a
     significant VMT component.

     Reduced vehicle ownership could have an effect on trip distribution results in models that employ
     vehicle  ownership to  estimate trip generation. If earlier steps in the analysis indicate changes in
     automobile ownership, areas with such models are advised to repeat the trip distribution step to
     account for ownership changes.

     There are no  known techniques, however, that permit direct assessment of trip distribution effects
     in response to changes in vehicle  "availability" (i.e., vehicle ownership).  One possible technique
     is to use a composite impedance approach (as defined in Section 3.3.5,1),  stratified by vehicle
     availability.-  If the coefficients are arranged so that households with fewer vehicles are more
     sensitive to trip cost (as a surrogate for distance), then such households' .trips would be more likely
     to be distributed to cjqser destinations, yielding a revised pattern of trips by origin-destination.
U. S. Environmental Protection Agency                                                             .   3-27

-------
 Technical Methods for Analyzir^PncingMeasures
 to Reduce Transportation Emissions


      Trip distribution also might be affected by the  VMT component of an emissions fee if that
      component is  significant enough.   It is difficult, however, to analyze the VMT-based impact
      because the relationship between aperiodic—typically annual—fee and daily travel is uncertain.

      In the absence of documented relationships, areas that believe the  VMT  component of their
      emissions fee  to be significant are advised to  use the same methods as they would for pump
      charges (Section 3.2.3.5), which are VMT-sensitive. Alternatively, such areas might consider one
      of the off-line techniques described in Section 3.3.6.
                                                                  '"*'." i              '
                  ,i'                   .       •             •               ;
      6. Emissions Fees and Mode Choice

      Emissions fees are not expected to produce a major impact on mode choice.  If the proposing
      agency believes the VMT component of a  proposed  fee  will be a substantial  element in mode
      choice, however, the agency should undertake a  mode choice analysis similar to that specified for
      pump charges, incorporating the cost from the emissions fee as a change in operating cost.

      Caution should be taken in treating the emissions fee as a cost element if the fee  is administered
      infrequently  (i.e., annually), since there  is little  information  on travelers' translation of periodic
      fees into daily/per mile costs. If the performing agency chooses to evaluate a mode choice effect
      in this instance, its assumptions regarding periodic  fee  assessment  and its effect on behavior
      should be clearly documented.

      Emissions fees can affect mode choice indirectly through changes in vehicle ownership.  If prior
      analysis indicates  an impact on vehicle ownership,  agencies  with a mode choice model that
      includes vehicle ownership  in its specification may revise their estimates of mode distribution of
      person trips.


      7. Emissions Fees and Time of Day

      Emissions fees are not expected to have any measurable effect on  the time of day of travel.
      Therefore, no analysis is needed.


      8. Emissions Fees and Traffic Assignment

      In general, emissions  fee  measures are not expected  to  exert any special  effect on  traffic
      assignment and, thus, no formal analysis is needed.

      If travel changes are calculated in any  of  the  preceding steps, for example, land use, vehicle
      ownership, trip distribution, or mode choice, then agencies with  assignment procedures should use
      them to determine changes in vehicle trips and VMT.  Areas without traffic assignment capability
      must rely on net changes in vehicle trips and VMT to perform the subsequent emissions analysis.


      9. Emissions Fees and Emissions Calculations

      Given the anticipated impact of emissions fees on emissions, EPA requires proposing agencies to
      (1) indicate how the vehicle fleet mix and VMT  distribution will be changed by the particular fee
      structure,  and then  (2)  revise the appropriate parameters in  the emissions model (MOBILE or

•*"2°                                                               U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                     '    .. '      '                    to Reduce Transportation Emissions

     EMFAC; PARTS for PM areas) to reflect the impact on average emissions factors (see Section
     3.3.9 for information on procedures to effect this change).

     Emissions-fees based on direct measurement of vehicle emissions output by inspection may result
     in  improved maintenance and reduced emissions rates.  EPA's Office of Mobile Sources  has
     developed a methodology for estimating emission reductions from inspection/maintenance (I/M)
     programs. The methodology is based upon the observed effectiveness of different I/M short tests
     for identifying high-emitting  vehicles and the  observed emission  reduction  associated with
     commercial repairs resulting from I/M programs.                    •      '             :   ''

     Areas without emissions model capability, which rely strictly on emissions factor estimates, must
     detail their assumptions and the procedures they use to modify existing emissions rates to reflect
     the shift in vehicle fleet mix.                                '    -. •    •

     If the proposed emissions fee policy is expected to generate changes in travel, resulting in a new
     travel assignment in Step 8 (or  revised vehicle  trips and VMT for areas without models),  the
     submitting agency should calculate the revised levels of VOC, NOX, CO, and PMio (where it is an
     issue) due to these changes in travel and in fleet mix/emissions factors.
. S. Environmental Protection Agency
                                                                                            3-29

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
     3.2.5   Roadway Pricing
         3.2.5.1   Anticipated Impacts

     The impact  of roadway pricing on travel
     and   emissions   is  complex.   Traveler
     responses   to   roadway   pricing   are
     conditioned  by roadway coverage (i.e., the
     extent and  number of facilities that  are
     priced), availability  of alternatives, and
     trip  purpose. Traveler responses  include
     changes in destination, route, mode, and,
     under congestion pricing, time  of day of
     travel. Activity planning may be effected
     to  schedule trips  more  efficiently  to
     minimize costs.

     The  impact on  emissions depends  on
     reduction in VMT and changes  in flow
     conditions     (reduced     congestion).
     Location and land use can be affected in
     the long term by the pricing scheme.
Table 3.5  impacts of Roadway Pricing
Analysis
Step
Land Use/
Activity
Allocation
Vehicle
Ownership
Vehicle Mix
Trip
Generation
Trip
Distribution
Mode
Choice
Time of Day
Traffic
Assignment
Emissions
Impact Form
** * j^j
Changes relative
attractiveness based
on pricing and
alternatives to driving
Depends on type of
location shift
None
May reduce vehicle .
trips if location shift
or "area" fees
Pricing affects
access (cost/
time) of particular
destinations
Use alternatives to
avoid road price
Congestion pricing
cause* time shifts
Compound:
reduces trips & „
VMT; cauuos rout* > x
shift&v '
Reduces vehicle
trips and VMT; also
congestion & speeds
/Impact
Type
Secondary
Trace
None
Primary if
area-wide —
otherwise,
trace
Primary
•/
Primary
•f ^ /
Primary
(» , C.
connection
pricing)
Primary
/ /
Derivative
effect
, Impact Indicator
Population and/or
employment by zone
Vehicles per household
per zone
Percent of vehicles by
VMT
Number of person trips
to/from each zone
0/D pattermt (average
trip length)
Person trips, by mode;
% transit; %wsMbikes
no vehicle trips;
average vehicle
occupancy
xTHp* by time period
/
•>> \ *
Link volume*; VMT
„ f
/
Tons or kg. of VOC,
NOX, CO, PM
     3.2.5.2 Basic Analysis

     Submitting  agencies  should  analyze the impacts of roadway pricing on  mode choice,  trip
     distribution and route choice through traffic assignment. Agencies considering congestion pricing
     forms of roadway pricing also should evaluate the anticipated impact of the measure on time of
     day of travel. Emissions analysis should reflect changes in vehicle trips, VMT, and speeds.

     Roadway pricing is expected to have a secondary, rather than primary, impact on land use; thus,
     formal analysis  is not- needed.  However, submitting agencies  should  include a statement  that
     describes their assumptions and the measure's anticipated impacts on long-term land use.


     3.2.5.3  Guidelines for Analysis


     1. Roadway Pricing and Land Use/Activity Allocations

     In the long  term, roadway pricing is likely to affect locational patterns, particularly if roadway
     pricing is applied only to  selected facilities.  However, if a large number of facilities are priced
3-30
               U. S. Environmental Protection Agency

-------
                                                         Technical Methods for Analyzing Pricing Measures
                       ,                                        . -   . • to Reduce Transportation Emissions


     then it becomes very difficult to make longer trips without paying. In that case there is a strong
     incentive to avoid vehicle trips, particularly longer ones.  In turn, businesses and households have
     an incentive to shift to locations that either (1) are not dependent on priced roadways or (2) will
     benefit from higher levels of service on the priced roads. "Areaywide"  pricing reduces or even
     eliminates incentives to re-locate.  The degree and pattern of relocation depend on roadway pricing
     rates and scope of the coverage. Substantial improvement of alternatives and supportive land use
     planning together with roadway pricing, would positively reinforce these trends.

     Although a formal analysis of roadway pricing effects on land use is not needed, submitting
     agencies should include  either (1) an assessment of the implications of the proposed policy on
     travel and emissions or (2) a statement that describes the area's assumptions and interpretation of
     potential impacts.  This  statement should reflect the appraisal of area decisionmakers as to the
     impact on long-term land use and transportation plans and programs.

     Performing  agencies with access to land use allocation models (Section  3.3.1.2.) that incorporate
     transportation cost through a measure like composite impedance (see Section 3;3.5.1) may use such
     models to investigate roadway pricing impacts on long-term development trends.  Results should
     be treated with caution, however, in view of the analytical limitations of these procedures.

     Submitting  agencies  (particularly those that do not use models  for land use analysis) might
     consider formulating alternative growth scenarios (Section 3.3.1.1) to account for the effects on
     regional travel and emissions from roadway-pricing induced development shifts.  These scenarios
     also should reflect any complementary actions with regard to land use alternatives or urban design.
                *                             -    "   -  • '                             -

     2. Roadway Pricing and Vehicle Ownership

     No direct impact is anticipated on vehicle ownership as a result of roadway pricing and, thus, no
     formal analysis  is needed.  If land  use and household location shifts are projected in Step 1,
     however, these shifts may translate to revised vehicle ownership levels;  agencies  should account
     for these effects.
                   ,                                 -                           \       -     -  '

     3. Roadway Pricing and Vehicle Mix    '

     Roadway pricing  is expected  to  have only limited  impacts on the types of vehicles used by
     households and, therefore, regional vehicle fleet distribution; no formal analysis is suggested.

     4. Roadway Pricing and Trip Generation                               '

     Roadway pricing can have a measurable direct impact on household person trip generation. This is
     particularly  true when an  area-wide pricing policy is pursued.  EPA suggests, when an area-wide
     pricing policy is pursued, that analysis of trip generation should be conducted. Areas with trip-
     generation models that incorporate a measure of generalized transportation cost (i.e., composite
     impedance)  are encouraged to use those methods to assess the impact  on trip generation (see
     Section.3.3.5.1.).  Use of these techniques  is  likely to  show significant impacts on total trip
     generation, and it would be beneficial for areas to develop this capability if they are considering
     area-wide pricing. Shifts  in land use and vehicle ownership should be carried through to the trip
     generation stage when area-pricing is being modeled.
U. S. Environmental Protection Agency                      '                                         3.31

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
      Roadway pricing that focuses on a single corridor or facility is less likely to have major impacts on
      trip generation. In these cases, most of the impact will be from mode and route shifting, changes in
      destinations, and time-of-day (if congestion pricing is used). At a minimum, areas modeling single
      corridor pricing  effects  should document  assumptions used for any anticipated changes in trip
      generation.
      5. Roadway Pricing and Trip Distribution

      Roadway pricing is likely to have a primary impact on trip distribution; thus, a formal analysis of
      anticipated effects is suggested.  In the short term, roadway pricing is likely to affect only non-
      work trips through shifts in destinations selected in response to pricing patterns. In the long term,
      work as well as non-work trips could be affected  as  workers change their work and/or home
      locations to  avoid the additional road  price (These  latter  effects may be  more appropriately
      modeled through the land use allocation step.).

      Areas with trip distribution models that can account for the effect of cost through a measure such
      as composite impedance (Section 3.3.5.1) are encouraged to use those models to assess the impact
      of roadway pricing on the  origin-destination patterns of non-work trips (both short-run and long-
      run) and work trips  (long-run).   Agencies should perform independent analyses for each trip
      purpose by income strata or use some other approach to account for income differences (Section
      3.3.5.2).

      Areas with  trip distribution models  that do  not incorporate  cost directly  in their  measure  of
      separation may consider using a travel time approximation for the increased link cost (see Section
      3.3.5.3).

      Areas without trip distribution models will have difficulty accounting for roadway pricing impacts;
      these agencies may not be able to perform  a realistic analysis given the complexity  of roadway
      pricing programs. Pricing applications considered by these  agencies would likely be limited to
      situations where chances for toll-avoidance are  minimal (e.g., pricing of a strategic bridge  or
      tunnel).

      In contrast to some of the other market-based measures considered, road pricing can have a fairly
      important effect on commercial and through traffic. For commercial users, the existence of a road
      user charge may actually prove to be  an  advantage, given anticipated impacts on road congestion
      on priced roadways and the importance of travel time to commercial users. The ultimate net
      benefits depend on the nature of the goods or services being transported.

      When travel time is of higher value than the anticipated increase in cost, local commercial  traffic
      (as well as personal traffic) will be drawn to the priced facility if pricing results in higher levels of
      service.  If pricing-induced increases  in travel cost are  considered important, however, trips will
      shift to unpriced facilities; this, in  turn, may generate  local traffic  congestion  and emissions
      problems.

      Like local commercial traffic, through traffic will also weigh the toll costs  against the  value  of
      time saved on priced facilities. Price sensitive through-travelers will shift to non-priced alternative
      routes or be inclined to pick an alternative route around the given region unless they simply know

3-32                                                                U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing'Measures
                                            .-.:.-.                ,. to Reduce Transportation Emissions

     of no alternatives.  Traffic response to trade-offs between time and money are difficult to assess
     with currently available methods, although new methods are under development.6  Agencies
     wishing to assess these impacts are encouraged to do so within the limits of their analytic tools and
     data.  All assumptions should be clearly documented.

     6. Roadway Pricing and Mode Choice

     Roadway pricing is likely to have a primary impact on mode choice. Agencies should submit a
     formal assessment of the impact of roadway pricing on person trips by mode, change  in vehicle
     trips, and average vehicle occupancy and change in transit trips.

     Areas with mode choice models in their planning systems are encouraged to use them for this
     analysis.  However,  it should'be noted that the analysis  of roadway pricing presents  several
     important modeling issues and special requirements:

               •   Network Representation of Tolls:   Modeling the  pricing of specific  rpadways
                  requires a concerted effort to represent the price in the highway network.  This is
                  equivalent to properly coding a toll value on all affected, individual links.  When
                  coding is completed, the path-building and skimming software must be alerted to the
                  presence of the toll and must be capable of creating a matrix that reflects the total toll
                  cost of traveling on all origin-destination pairs.  It is, somewhat simpler to model
                  area-wide pricing since the matrix manipulation program in most modeling software
                  packages can build a cost matrix representing the additional cost of traveling into the
                  affected area from any origin zone.

               •   Mode Cost Sensitivity to Tolls:  Few mode choice models are set up to accept toll'.
                  costs explicitly; most deal only with automobile operating cost and possibly parking
                  cost.  This leaves open the-issue of what coefficient value to assign to tolls.  If the
                  toll is paid physically at the point of charge, it may be reasonable to treat it  as an out-
                  of-pocket cost comparable to parking and apply a  parking  cost coefficient  (see
              ;    Section 3.3.6.1).'  If the toll  cost is determined electronically, with billing issued
                  periodically, then it may be more appropriate  to treat it as an operating cost.   The
                  agency should state the assumptions underlying its choice of coefficient.  The agency,
                  also may .wish to  consider stated preference methods to ascertain  the relevant
                  sensitivity (see Section 3.3.6.8).                                         ' '  .  •

               •   Queuing and Delay:  If tolls are collected through automatic identification methods
                  (i.e., no toll booths), delays related to toll payment would be minimal.  When tolls
     6 Three recent studies Tiave directly addressed air quality and travel impacts of freight transportation.
      Cambridge Systematics,  Sierra Research,  and Jack Faucett Associates, "Air Quality Issues in Intercity
      Freight," prepared for the FRA, the FHWA, and U.S. EPA, draft final report, December 1996
       Cambridge Systematics et al., "A  Guidebook for Forecasting Freight Transportation Characteristics
      Demand," NCHRP Report 388, National Academy Press, Washington, DC, 1997.
      Cambridge Systematics, COMSIS Corporation, and Alan Horowitz, "Freight Planning Manual,"  developed
      for me Federal Highway Administration.                .                                       .
U. S. Environmental Protection Agency                                      •     -                    3-33

-------
Technical Methods for Analyzing'Pricing j
to Reduce Transportation Emissions


                  are collected manually,  however, toll booth  operations may well produce queue
                  delays that affect (1) travel time on the facility and (2) micro-level traffic effects that
                  contribute to  emissions.   Agencies  may consider using traffic microsimulation
                  methods to assess these effects (see Section 3.3.8.3).

               •  Type of Model: Multinomial logit, especially nested logit models, provide the most
                  realistic estimates  of mode choice impacts, particularly when toll treatment differs
                  across modes (see Section 3,3.6.5).

             ,  •  Non-Work Mode  Choice Models:  There should be a mode choice model for each
                  separate trip purpose (see Section 3.3.6.4).

               •  Income Stratification: The mode  choice analysis  should be performed separately
                  by  income  strata,  or  through an  equivalent method that  accounts for income
                  differences (see Section 3.3.6.3).

               •  Adjustment for Inflation:  It is important that the  dollar value of the toll price be
                 • adjusted for compatibility with the  base  year  for which the model coefficient was
                  developed. This generally means deflating the tested price by a time-inflation factor
                 that makes the costs compatible with the coefficient in real  terms (see Section
                  3.3.6.2).                    .

               • HOV Special Rates:  If the policy incorporates toll exemptions or  discounts for
                 HOVs, few mode choice models will provide sufficient analytical support. Off-line
                 methods, such as the TDM Evaluation Model, are suggested for assessing this effect
                 (see Section 3.3.6.9).

     Areas -without mode choice models will have a difficult time adequately assessing  a measure as
     complex as roadway pricing.  Off-line methods, which include  quick-response techniques and
     sketch-planning/elasticity approaches (see Section 3.3.6.9),  may be used with  caution; agencies
     using such  methods must document their assumptions regarding  potential  limitations  of the
     analysis.


     7.  Roadway Pricing and Time of Day

     If roadway pricing is applied evenly  throughout the day, then only limited impacts on travel time
     of day would be expected.  However, if the implementation involves peak/off-peak congestion
     pricing differentials, time of day shifts are anticipated; and these should be analyzed.

     Current time of day analysis methods that employ "peak factors" are probably not sufficient to
     capture time of day effects. Modified procedures, such as those described in Section 3.3.7, may be
     considered.  Analysts should estimate the specific sensitivity of travelers' start times to price
     differentials at  different times during the day.  Non-work trips are likely to exhibit greater
     sensitivity than  work trips because  of their greater flexibility (and perhaps  cost sensitivity),
     although some workers may be able  to shift hours outside the peak.  The sensitivity of travelers'
     trip times to price differentials is uncertain in current practice, but it may be possible to make an
     assessment that can be refined subsequently using observed  data. Such initial estimates may be
     based on empirical sources or on stated preference surveys (see Section 3.3.7.4).

     All applications of roadway pricing that involve time of day price  differentials should make an
     estimate of the time of day shift in travel, using approximation methods such as those described

'^                                                                U. S. Environmental Protection Agency •

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                                     to Reduce Transportation Emissions

      above for broad applications of the  policy.  However,  agencies  introducing  more complex
      applications are expected to use more sophisticated analysis methods.
                           ^                .                  I                "
      8. Roadway Pricing and Traffic Assignment

      Roadway pricing has a special effect on traffic assignment because pricing of specific links in the
      highway  network is likely to induce  some shifting to. alternative routes,  particularly by  cost
      sensitive  travelers with  some flexibility (lower income  and non-work travelers).  This jvould
      probably  constitute  the  policy's  primary effect and should be evaluated.   Evaluating agencies
      should be careful to  ensure that all alternative paths to the tolled network are represented in order
      to address the diversion issue.

      Areas with traffic assignment models may perform an analysis of roadway pricing impacts on
      revised vehicle trips, VMT, and speeds using one of the following methods:

               •  Toll Diversion Model:  Some areas have developed models that specifically offer
                  travelers a choice  of  a toll or free path during assignment.   This approach  is
                  described in Section 3.3.8.5.

               •  Travel Time Equivalent: A  less accurate,  but acceptable, approach is to translate
                  the impact of the toll into a travel time equivalent (also described in Section 3.3.8.5).
                  This approach, however, disregards the likelihood that travelers with a higher value
         ;         of time will opt to use a tolled facility even at higher cost simply to take advantage of
                  the improved travel  time.  That is, this method assumes away traveler willingness to
                  pay for time saved.

      Areas without traffic assignment capabilities will probably not be able to evaluate roadway pricing
      options unless alternative route choice options are greatly limited (e.g., isolated bridge or tunnel).

      9. Roadway Pricing and Emissions
         " '           i     l        '       '              "   <              ~
      All analyses of roadway pricing should conclude with an estimate of the net changes in the major
      pollutants for which credit is being claimed.                                        '

      If roadway pricing has an effect  on  emissions rates,  it would be through changes in congestion
      levels, rather than through  impacts  on vehicle mix.  That  is,  roadway  pricing implemented as
      congestion pricing can affect emissions rates if it significantly affects level  of service  and travel
      speeds.

      Areas able to perform traffic  assignment are advised to use an emissions factor model, such as
     MOBILE or EMFAC (PARTS for particulate emissions), to furnish emissions estimates, using the
      results from the traffic assignment process (note that MOBILE produces emission factor estimates
      rather than estimates  of emissions  amounts).                     '  •           :

     Areas without emissions models may apply emissions factor estimating methods to the net changes
      in travel projected from traffic assignment.                                         .

     Areas without traffic assignment capability are not expected to be able to perform this assessment,
     except in limited circumstances as earlier described.

U. S. Environmental Protection Agency                                   '..'-.'          3.35

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
     3.3   Review of Methods and Procedures
     Clearly, the most reliable basis for projecting the impacts of changes in pricing levels—especially
     major changes—would be to statistically track and measure consumer response to such actions
     under (and in relation to) actual conditions.  Before-and-after studies, possibly involving consumer
     panels, can  provide  great  insight  into  the  expected and unexpected nature  of response.
     Unfortunately, there are relatively few studies (in the United States) in which such measurements
     have been made. It will be important to institute such monitoring systems in coming years and as
     new policies are tested to obtain a fuller understanding of consumer response.

     In the interim, we must rely on existing planning and emissions analytic procedures and develop
     new ones. As indicated in Chapter 2, current capabilities vary widely across the U.S. Virtually all
     major urbanized  areas  and many smaller areas use some variation of the "four-step" planning
     process for all transportation-related analyses. This process portrays transportation decision-making
     as a series of sequential steps. The four core steps are trip generation, trip distribution., modal choice,
     and traffic assignment. These systems work from trip tables that describe travel movements between
     origins and destinations defined by traffic analysis zones.  These trips are "assigned"  to a coded
     transportation network to simulate actual traffic flows and conditions on actual facilities.  Emissions
     analysis is then performed as a separate step, utilizing basic travel outputs (vehicle trips, VMT) from
     the four-step process.

     In practice, there are many variations within each analytical stratum with respect to:

               •   Level of detail used to characterize travel by purpose, time of day, income and other
                   socioeconomic differences;

               •   Detail in the travel  networks; and

               •   Sophistication of individual models.

     Smaller urbanized areas, where transit service is minimal or non-existent, may not have a formal
     mode choice capability (these systems are characterized as "three-step" models).  In some areas,
     capabilities are even more limited. With this wide range of capabilities among areas that might
     consider market-based measures,  it is difficult to prescribe use of particular methods or " standard"
     techniques.   Rather, the  objective of this  guidance  is  to encourage  areas to  perform the best
     possible analysis using their existing tools  and data, and to recommend special procedures that
     may help them to improve their analysis.   Those with more sophisticated analytical capabilities
     will be rewarded with greater opportunity to receive emission credits.

     It is commonly acknowledged that current-generation travel  forecasting tools are not sensitive to
     pricing; rather, price sensitivity has had to be back-fit into their structure.  These forecasting tools
     were developed  primarily to guide the location and design of highway facilities; cost was not
     included  in their structure because of an underlying assumption that the price of transportation
3-36                                                     .         •   U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                     ,                                to Reduce Transportation Emissions

      would change at the same rate as inflation, thus would not be an important future criterion.7 The
      gasoline crises of the 1970s and 1980s; together with federal requirements related to transit system
      funding, caused many areas to initiate upgrades to their models to formally include pricing.

      Pricing sensitivity, however, is limited tp mode choice in most models. It is reasonable to believe
      that the effects of many pricing measures will extend beyond mode choice to include destination
      choice, route choice, and even time of day; some of these may have even more significant .impacts
      than mode choice.  It also is reasonable to expect that longer-term effects  on household or business
      locational decisions can affect land use, vehicle ownership, and trip generation
                                       /     '     • '                  '    "
      Current models  may be unable to deal with these broader  impacts.  In many areas,  model
      capabilities are in a state of flux as they (1) undergo change and enhancement in response to the
      requirements of the CAAA and ISTEA and  (2) respond to new data available  from the 1990
      Census and numerous urban area regional travel surveys.

      Until agencies have greater experience with actual implementation of major pricing applications,
      there will be limited  empirical evidence upon which to base or validate projected impacts from
      these measures.  Hence, it is necessary to  extract the best possible analysis from existing travel
      forecasting tools.  There are numerous  ways that  these tools can be adapted to improve their
      sensitivity and accuracy in the evaluation of pricing measures. Adaptive procedures include:

               •  Accounting for cost  throughout the  travel  demand choice  hierarchy through a
                  generalized cost measure such as composite impedance or accessibility;

             . •  Sharpening estimates  of price coefficients through special surveys, or  transferring
                  coefficients or models from other sites;

               •  Improving  sensitivity to differences in  response by income level through income
          •'.,       stratification or similar techniques;

               •  Qaining initial insight into relationships between special  pricing instruments and
                  travel behavior response through  stated preference surveys;

               •  Using sample enumeration methods  to achieve greater specificity for particular
                  measures as they are  experienced by  households and possibly greater  accuracy in
                '  estimating response; and

               •  Use of sketch planning or off-line methods to support formal methods where tools or
                  capabilities do not currently exist.

     This section provides a  brief review  of the steps in travel  forecasting that may be  involved in
     assessing pricing actions and evaluating their impacts on travel and emissions. The intent is not to
     provide a full review  of the transportation planning process, but rather to draw attention to those
     aspects  of the process that are  important in relation to pricing.  This  section describes modeling
     concepts that offer improved accuracy for analyzing pricing and presents  references for additional
     information on their development and use.
     7 Transportation cost also was a difficult element to include in model structures because of the.models' heavy
      reliance on travel time and the close correlation between [per-mile] cost and travel time for most trips in
      most U.S. urban areas.
U. S. Environmental Protection Agency    *-.  .                   .                                      3-37

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions

                             '••'"''             '       ''             '
      This section is designed to address the range of needs and capabilities that may exist among areas
      desiring to perform pricing analysis. Larger urbanized areas may include many of these concepts
      in their modeling systems, while other areas seek to develop such capabilities as part of a longer-
      term  model enhancement process.   Other areas with more  limited capabilities may  need help
      identifying more basic techniques, as well as help prioritizing future model development activities.

      The reader is directed to two  primary resources that serve as important guides for agencies
      involved in performing transportation air quality analysis:

               1. Manual of Regional  Transportation Modeling Practice for Air Quality Analysis*
                  (hereafter referred to as the "NARC Manual"), developed in 1993 for the National
                  Association of Regional Councils, is a major resource on current model capabilities
                  and deficiencies in relation to air quality applications.  It provides a thorough review
                  of the requirements of the Clean Air Act and their impact on travel forecasting tools,
                  the nature  and  capability of current tools, and suggestions  toward most effective
                  methods. Numerous references to example sites and literature sources are provided.

               2. Travel Model Improvement Program (TMIP)\ which is being jointly sponsored by
                  the U.S. Department of Transportation, the U.S. Department of Energy, and the U.S.
                  Environmental Protection Agency to foster both near and long-term  improvements in
                  travel  forecasting  tools.  An  October  1994  report  on  Short  Term Model
                  Improvements10 provides  a  valuable  update on  model enhancements  that  are
                  underway or in place  for addressing air quality and other closely related modeling
                  issues.
     Both of these sources were used extensively in preparing the synthesis of analytic procedures in
     this section.

     In addition to these two primary references, the reader should  be aware of several key research
     projects now underway  or  recently  completed  which are  charged  with  developing major
     enhancements and new approaches  to the  analysis of transportation-related  emissions.  These
     include:

               •   National Cooperative  Highway Research Program (NCHRP) Project No. 8-33:
                  Quantifying Air Quality and  Other  Benefits  and  Costs of Transportation
                  Control  Measures:  This study is undertaking  a comprehensive review  of the
                  analytic techniques used for evaluating TCMs. It is also charged with designing a
                  new  framework  that  addresses  the overall linkages  between transportation,
                  emissions, and atmospheric models. Computational limitations in each of the steps
     8 Greig Harvey et al., "A Manual of Regional Transportation Modeling Practice for Air Quality Analysis,"
      National Association of Regional Councils, 1993.
     9 Cambridge Systematics, Inc., "Travel Model Improvement Program:  Short Term Model Improvements,"
      prepared for the Federal Highway Administration, U.S. Department of Transportation, 1994.
     10 B. Spear, "New Approaches to  Travel Forecasting Models: A Synthesis of Four Research Proposals,"
      Travel  Model Improvement  Program,  Volpe  National Research Center for the U.S.  Department of
      Transportation and the U.S. Environmental Protection Agency, January 1994.
3-38                                                                U. S. Environmental Protection Agency

-------
                                                         Technical Methods for Analyzing^ Pricing Measures
                          .,         '                                to Reduce Transportation Emissions

                  are to be addressed individually, as are those features that affect accuracy in the step-
                  to-step linkages.

                  Phase 1 of NCHRP Project Number 8-33 is complete, and copies of Phase 1 reports
                  are available on loan from the NCHRP. Research papers based on Phase 1 work are
                  expected to, be available for review in Fall 1997. Working titles for these papers are
                  " Relationships  between  Implemented  Transportation  Control  Measures  and
                  Measured Pollutant Levels," and "Development of an Improved Framework for the
                  Analysis of Air Quality and Other Benefits  and Costs of Transportation Control
                  Measures."

                  NCHRP Project No. 25-11: Modal Emissions Factor Development:  Undertaken
                  by researchers at the University of California at Riverside, this study is charged with
                  developing a new emissions-factor model that addresses the deficiencies of the
                  current-generation  MOBILE and EMFAC models with regard  to  sensitivity  to
                  vehicle and transportation inputs.   Specifically, current models are  insensitive to
                  variations that occur in vehicle emissions under  normal driving conditions; where
                  acceleration, idling, steady state, etc., produce very different emissions rates. Current
                  models smooth over these variations by using emissions factors based on an average
                  speed.  This project is obtaining substantial new vehicle activity and emissions data.
                 This  data will be used  to develop a modal emissions modelthat (1)  is compatible
                  with the types of traffic simulation models used to design and evaluate transportation
                  operational improvements, and (2) accurately reflects the impacts of speed, engine
                  load and start conditions on emission.

                  Research has  been completed for Phase 1 of NCHRP Project Number 25-11.- This
                  phase called for 1) review of the existing literature on factors in the vehicle operating
                  environment; 2) definition of the domain and distribution of the modal parameters of
                  each  vehicle operating mode; 3) evaluation of extent to which  existing models meet
                  analytical needs; 4) design of testing protocol to measure vehicle modal emissions;
                  5) preliminary testing of a sample of vehicles currently in use; and, 6) development of
                  a working model for analysis.

                  NCHRP Project No, 25-7:  Improving Transportation Data for Mobile Source
                  Emissions Estimate: This study by the University of Tennessee and SAIC,. et al., is
                  specifically directed at appraising the data requirements needed for transportation/air
                  quality modeling. Key elements of the research include:  1) identifying transportation
                  variables that are available  or  heeded  for  preparing  air  quality, forecasts;
                  2) identifying techniques for quantifying these variables; and 3) describing the inter-
                  relationships between transportation and emissions  rates.

                  Project number 25-7 has been completed, and the final report will be available on a
                  loan basis pending a publication decision by the NCHRP.
                  Air Quality Issues in Intercity Freight: Sponsored by the U. S. DOT. and EPA, this
                  project has developed  methods for evaluating  the emissions impacts of intercity
                  freight operations,  especially truck'and intermodal rail. Extensive work  has been
                  done to develop practical  planning  procedures  to (1) realistically assess  intercity
                  freight travel characteristics,and reactions to policy actions (TCMs), and (2) refine
                  emissions factor estimates for these modes.
U. S. Environmental Protection Agency                                 .    '   '     .                  3-39

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions


     Finally, it  also  should be noted  that this guidance document  is followed by two  technical
     appendices that may be of great relevance to reviewers:                                      '

              • Appendix A:  A Case Study Adaptation of a Regional Model to Increased Price
              .   Sensitivity.  Appendix  A provides an illustrative, step-by-step example of how a
                 typical  four-step modeling process was  strategically upgraded  and' enhanced to
                 improve its ability to analyze pricing measures, and then applied to test a range of
                 market-based  measures.  The base  model that was transformed is  the Greater
                 Metropolitan  Washington  Council of  Governments (MWCOG)  model.   The
                 improvements   are   hypothetical;  because  the   adaptation  was  done   as   a
                 "demonstration.," typical steps to thoroughly  validate the revised model were not
                 performed, and the revised model is not in use by MWCOG.  However, the "case
                 study" model is useful in demonstrating how enhancements such as those suggested
                 in this chapter could be made.

              • Appendix B: An Example of the Application  of the STEP Process to the Analysis
                 of Pricing Strategies. Appendix  B illustrates  how sample enumeration techniques
                 can be used to link travel survey data and conventional modeling methods, which
                 would provide  additional  flexibility and  even accuracy to analysis  of pricing
                 measures' impacts on air quality.  These methods have, been used to analyze pricing
                 TCMs in Seattle, Washington and several California cities.

     Neither of the approaches set forth  in these appendices represent an analytic procedure prescribed
     by EPA.  They are provided for illustration only, to show analysts how specific adjustments and
     innovative applications are being made to address the complex range of impacts that may derive
     from market-based measures.  The procedures described also are useful in demonstrating two very
     different ways of approaching analysis of market-based  measures — through  concerted direct
     modifications to  the  existing  four-step process, and through  a "parallel"  approach that moves
     outside the conventional  process for key analytic  assessments.   Both approaches have distinct
     advantages  and disadvantages, which must be evaluated by the performing agency.


     3.3.1  Land Use/Activity Allocation Methods


     Most regional planning processes start with a procedure that  allocates household and employment
     activity to  specific geographic locations across the region.   The pattern and level  of  these
     allocations then becomes the basis for all subsequent estimates of trip levels and flow patterns.
                               ''                          „',.,!
     A key question for analysis  of pricing measures is whether these locatipnal decisions might be or
     should be sensitive to levels and variations of transportation pricing effects. This question has two
     levels: (1) "real world" sensitivity  of household and business  locational decisions to pricing
     measures and (2) analytical sensitivity of the transportation planning process to pricing measures.

     This is perhaps one of the least precise steps in  the current transportation analysis process, though
     it represents what can be a very significant factor in long-term travel trends and effectiveness of air
     quality and  other management strategies.  Currently, most land use allocation  decisions are made
     through a process of negotiation, with considerable political involvement and concern.  Market
     realism may or may not be prominent in this process.
                                                                  £/.. S. Environmental Protection Agency

-------
                                                           Technical Methods:for Analyzing Pricing Measures
                                ,                          -           to Reduce Transportation Emissions

      There are two basic ways in which areas attempt to account more  explicitly for the effects'of
      market forces or transportation policy on growth allocations:  scenario approaches and land use
      allocation models.                          .


      3.3.1.1  Scenario Approaches

      Generally, population and employment allocations based upon hard data (including population and
      employment trends, land availability and price, land use occupancy and rent, zoning and land use
      decisions, economic  development plans  and market assessments,  measures  of current and
      anticipated transportation accessibility, etc.) are more realistic and sensitive to market forces than
      those based purely on "fair share" allocations or political aspirations. Some areas address their
      uncertainty in both level  and distribution of growth by formulating one or more scenarios that are
      then carried forward into the analysis.  These scenarios can be. used to reflect major differences in
      expectations  linked to pricing,  urban design and zoning/growth management changes, or other
      policies.  Analysts should take care that eventual  emissions results that stem from these scenarios
      reflect realistic and implementable land use assumptions."
     11 Greig Harvey et al., "A Manual of Regional Transportation Modeling Practice for Air Quality Analysis "
   '   1993, Sec. 3.3.2, pp. 3-13 to 3-18.
U. S. Environmental Protection Agency
                                                                                              3-41

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
     3.3.1.2  Land Use Allocation Models

     Formal mathematical models for land use allocation exist and are in use in a small  number of
     regions.  Options  include DRAM/EMPAL12 (most commonly used  land use model in U.S.),
     POLIS13 (currently used in the San Francisco Bay Area), the Herbert-Stevens Model14 (used in the
     Penn-Jersey Study), MEPLAN15, METROSIM16, TRANUS'7, and TOPAZ18. A recent draft report
     prepared for EPA's Office of Mobile Sources provides a comprehensive review of the various land
     use modeling tools currently available.19

     The  models  mentioned  above  generally  use  zone-level time  series  data  on  population,
     employment, land availability and accessibility to allocate population and employment forecasts to
     subareas. Use of these models has been limited, however, both by the  effort required to run them
     and by questions about their predictive accuracy.

     Factors that appear to limit the accuracy of current generation land use models include:  a tendency
     to extrapolate past trends, rather than incorporate behavioral decision-making; a tendency to base
     household location decisions substantially  on  the work trip;  limited  accounting for market
     responses (real estate, speculative investments,  ethnicity) and time-lags; and weak representation
     of transportation accessibility and cost.
     12   S.H. Putman, "DRAM/EMPAL  - ITLUP:  Integrated Transportation Land-Use Activity Allocation
      Models: General Description,"  S.H. Putman Associates, Philadelphia, Pennsylvania, 1991.
     13 P. Prastacos, "Urban Development Models for the San Francisco Region:  From PLUM  to POLIS,"
      Transportation Research Record 1046, 1985.

     14 J.P. Herbert and B.H. Stevens,  "A Model for the Distribution of Residential Activity in Urban Areas,"
      Journal of Regional Science 2(2), 1960, pp. 21-36.

     15 MEPLAN was developed by staff at the Center for Land Use and Built Form Studies at Cambridge
      University,  at the firm of Applied  Research of Cambridge, and at the firm of Marcial  Echenique and
      Partners. MEPLAN allocates exogenous forecasts of basic employment to analysis zones and then allocates
      non-basic employment and population to zones based on costs of travel, floor space, and the availability of
      other goods and services. Demand for accessibility generates rent levels,  and land uses are segregated from
      one another based on capacity to pay rent.

     16 METROSIM models regional impacts on  land use patterns and transportation systems that may result from
      project development, and was developed by Alex Anas of the University of Buffalo. The model can analyze
      impacts for up to 3,500  analysis" zones.  For each  zone,. METROSIM  models employment, real estate
      demand, household demographic characteristics, and travel patterns.
     17 TRANUS shares many characteristics with MEPLAN. Development of the TRANUS model is outlined in
      T. de la Berra," Integrated Land Use and Transport Modeling," Cambridge University Press, 1989.
     18 J.F. Brotchie et al,  "Alternative  Approaches to Land Use Modeling," Commonwealth Scientific and
      Industrial Research, United Kingdom, 1981.

     19 Systems Applications International, Inc.,  "Evaluation of Modeling Tools for Assessing  Land Use Policies
      and Strategies," draft report prepared for Office of Mobile Sources, April 1997.
3-42                                                                 JJ. S. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                                                      .                to Reduce Transportation Emissions

      These  models  can aid land use  impact evaluations, provided they  are  applied  with careful
      judgment.  Similar to typical trip  distribution mbdels, land  use  models generally employ travel
      time m, their allocation process.  One way to enhance their utility for pricing analysis is to
      introduce pricing directly as a variable through a measure such  as composite impedance (see
      Section 3.3.5.1).  The Bay Area model system utilizes such a formulation, and would be worth
      further study by areas considering such enhancements and/or where the rate or pattern of growth is
      a major issue.

      Readers are encouraged to consult the NARC Manual (Section 3.3,2) and the TMIP's review of
      Short-term Travel Model Improvements (Section 3.0) for rnqre information on land use models
      and potential enhancements.20                                            ,    .


      3-3.2  Vehicle Ownership


      Vehicle ownership is often used as an input for U.S. urban area model systems. Agencies in some
      areas  use vehicle  ownership  rates  (or a measure of vehicle  availability)  as a measure of
      households' propensity to travel in trip generation and other modeling steps.  Very few areas
      explicitly estimate a policy-sensitive  model of vehicle ownership  rates as part of their overall
      transportation modeling system. Most planning professionals believe that explicit representation of
      vehicle ownership results in more accurate and realistic transportation analysis.  Evidence suggests
      that vehicle ownership influences person trip generation in ways that extend beyond the effect of
      income.  Vehicle availability may have a key effect on  discretionary trips and on the  level of
      errand chaining.  Households with fewer vehicles per licensed driver are more.pressed to work out .
      vehicle use priorities, which affects vehicle sharing decisions and use of non-motorized  modes.
      Where  residential  densities and levels of transit service  are high enough  to affect vehicle
      ownership levels, explicit prediction of vehicle ownership would clearly be expected to increase
      forecasting accuracy.

      Model systems that rely exclusively on household income and  not vehicle ownership can still offer
      good accuracy, depending on how well income is specified as  an explanatory variable.  Generally,
      this means that other modeling steps must exhibit strong relationships between travel and income. '
     20
       Throughout the remainder of this section, the Greig Harvey et al., publication, entitled "A Manual of
      Regional Transportation Modeling Practice for Air Quality Analysis," National Association of Regional
      Councils, December 1993, will be referred to as " the NARC Manual."
      The report by Cambridge Systematics, Inc., entitled " Travel Model Improvement Program:  Short Term Model
      Improvements," (prepared for the Federal Highway Administration, U.S. Department of Transportation  1994)
      will be referred to throughout the remainder of this report as " the Short-term Model Improvements Report."
U. S. Environmental Protection Agency           '       >                  '                             3 43

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
     There are several types of vehicle ownership predictive models.  These include:
                  ••       '         *      •   -       '             •
               •  Cross-Classification Models:   tables  of factors derived from  Census  or  travel
                  survey data (generally have no explicit linkage to transportation conditions).

               •  Nomographs or Curves:  empirical curves that  reflect the fraction of households
                  owning a particular number of vehicles, based on household size and income level.
                                                               ''•'!"
               •  Mathematical  Models:   actual  equations that calculate the  number  of vehicles
                  owned, based on the value of specific equation variables.

     Generally, the  form of model is  less important than the  sensitivity with which  it reflects  and
     responds to key socio-demographic and policy variables, especially pricing.  Aggregation of key
     variables may affect accuracy levels.  For example, some models produce an estimate of average
     vehicles per household rather than preserve the distribution of number of households by number of
     vehicles.  The reader is encouraged to consult the NA&.C Manual21 for a more detailed discussion'
     of the attributes of vehicle ownership models.

     In order to analyze the effect of pricing measures such as registration fees, these models must be
     somewhat sensitive to the price of purchasing a vehicle and the  annual cost of owning one.  In the
     context of travel modeling, increases in ownership cost should  be reflected in decreased numbers
     of multiple-vehicle households and increased numbers of one-vehicle households  (possibly even
     an increase in zero-vehicle households). The inclusion of ownership cost .is not a common feature
     in current vehicle ownership models, presenting an area for research, along with composition  and
     use of vehicles (discussed below under vehicle mix).  At least  one vehicle ownership model  has
     been documented that is sensitive to user cost.22
     3.3.3 Vehicle Mix


     Vehicle mix, which is a special area of concern in modeling pricing impacts on vehicle ownership,
     is not accounted for in conventional vehicle ownership analyses.  Vehicle ownership models
     account exclusively for the number of vehicles owned by a household, and do not deal in any way
     with the characteristics or use patterns of those vehicles.  Yet, vehicle mix is-an important factor
     for policies such as emissions-based registration fees or even fuel prices, where the respective
     economic forces may be more likely to influence the type of vehicle owned or its relative use,
     rather than the number of vehicles owned.
     21 Greig Harvey et al., "A Manual of Regional Transportation Modeling Practice for Air Quality Analysis,"
      National Association of Regional Councils, December 1993
       K. Train, Qualitative Choice Analysis, Theory, Econometrics, and an Application to Automobile Demand,
      The MIT Press, Cambridge, Massachusetts, 1986.
3-44                                                                U. S. Environmental Protection Agency

-------
                                                    ,      . Technical Methods for Analy2ing Pricing Measures
                                                                      to Reduce Transportation Emissions

      The accuracy of emissions  analyses based  on an  emissions factor  model depends on the
      distribution of the regional vehicle fleet and its average emissions rates, which derive from vehicle
      distribution. Shifts in the age or type of vehicles owned would obviously affect this relationship,
      a,s would different patterns of use, since the distribution is weighted by VMT for analysis.

      Typically, vehicle distribution is not directly considered  by conventional  travel analysis  tools.
      Hedonic models, scrappage studies,  empirical studies, special  studies, and sample enumeration
      offer approaches to considering fleet vehicle mixes. It should be noted that there may be overlap.
      between these categories of approaches.  For example,  empirical  studies  may include  some
      examples of special studies. The bullets below provide brief descriptions of these analytical tools
      and some illustrative examples   .--'".                                         .

                    Hedonic Models—These models  analyze  the  relationship  between  consumer
                    demand for characteristics of goods rather than for goods themselves. These models
                    can be used,  for example, to project price-induced shifts in ownership of vehicles
                    with different characteristics.  Hedonic models are complex and may only be an
                    option  where staff have a strong economics background.  Works by Train and by
                    Golub and Kitamura offer two examples of hedonic vehicle choice models.23*24
                    Sample Enumeration—Sample  enumeration involves, in the  vehicle mix  context,
                    taking  a sample of households within a region arid extrapolating to the regional
                    population  base from the sample's vehicle  ownership and/or  usage responses to
                    pricing measures. Survey samples have the advantage that they can be re-surveyed
                    in follow-up  analyses to study actual responses  to changes  in price signals (see
                    Section 3.3.6.7 below for more information).

                    Scrappage  Studies—Scrappage  studies have been performed in conjunction with
                    programs to buy back or otherwise accelerate the retirement of older, high-emitting
                    vehicles from the regional fleet.  These studies offer estimating methods that may be
                    of value to many areas.2526

                    Empirical Studies—Periods of major change  in fuel prices relative to income during
                    the 1970s and early  1980s, and major changes in the  price  of vehicles and their
                    characteristics relative to real incomes, have produced trends  in ownership  that are
                    reflected in various  research studies. These  studies could  be  used  to develop
                    estimates or assumptions regarding  shifts in ownership or use  patterns that might
                    occur in response to price signals.27'28
      23 K. Train, Qualitative Choice Analysis, Theory, Econometrics, and an Application to Automobile Demand
       The MIT Press, Cambridge, Massachusetts, 1986.
      24 T. Golob and R, Kitamura, for the California Energy Commission.  It should be noted that the analysis
       conducted by Golob and Kitamura is not publicly available as of this writing.                ,'    '  ' -
      25 U.S. Environmental Protection Agency, "Accelerated Retirement of Vehicles," in Transportation Control
       Measure Information Documents, March 1992.
      26
      27
  Beth Deysher  and  Don  Pickrell, Emissions  Reductions from  Vehicle Retirement  Programs,  U.S.
Department of Transportation, John A. Volpe National Transportation Systems Center, 1996.
 David L. Greene, "Vehicle Use and Fuel Economy:  How Big is.the Rebound Effect?"  The Enerzv
Journal 13, No. 1,1992, pp. 117-143.                                             ''           ^
,U.S. Environmental Protection Agency                         '                                       3.45

-------
Technical Methods for Analyzing Pricing Measures
to_ Reduce Transportation Emissions
                                                          /     '     i     f

        •           Special Studies—Locally-performed studies involving stated preference techniques
                   could produce  estimates of consumer response to particular types of vehicle-based
                   pricing policies (see Section 3.3.6.8.below for more information). Recent work
                   sponsored by the California Energy Commission has used such methods to ascertain
                   the demand for electric vehicles.  The study used travel  data from panels of
                   households to establish current travel and vehicle use patterns, and investigated how
                   these  patterns  might  change  under  different  pricing  and  vehicle   attribute
                   conditions.29

     3.3.4    Trip Generation

     This step of the planning process estimates the number of daily trips that are produced by
     households  {productions]  or that originate  in  a zone  [attractions].   These trips  usually are
     estimated separately by trip purpose. The  simplest classification is home-based work, home-based
     other, and non-home based trips; more complex systems split home-based non-work trips  into
     shop, school, and "other" and disaggregate non-home-based trips into work-related and other.30

     Trip productions and attractions are estimated through separate procedures.  These may be either
     cross-classification relationships,  matching a particular trip " rate"  with a particular combination
     of determining characteristics, or regression models, which use a functional relationship expressed
     as an equation.  The cross-classification models are the most common approach,  and can be
     reasonably accurate depending on the variables used for stratification, the degree of stratification,
     and  the reliability of the trip  rates themselves.  Regression models are generally regarded as less
     accurate (unless they are of a fairly sophisticated non-linear form).

     Significant references exist on trip generation practices.  The reader is advised to consult Section
     3.3.5 of the NARC Manual or the TMIP Short-term Travel Model.Improvement report for more
     information and references on the  general subject.31

     There are a number of  issues  related to trip generation methods that are  important when
     considering the suitability of models to evaluate pricing actions; these include:

               •   Person trips versus vehicle trips;

               •   Inclusion of transportation and cost variables;
     28 Don Pickrell, "Automobile and Gasoline Demand Revisited," Volpe National Transportation Systems
      Center, U. S. Department of Transportation, 1992.
     39 C. Kavalec, "CALCARS: The California Conventional and Alternative Fuel Response Simulator,"
      California Energy Commission, April 1996.
     30 Some models, such as those of the Maryland National Capital Parks  & Planning Commission, have
      evaluated  trips generated by  time  of day  and  purpose, (Montgomery  County, Maryland) and applied
      distribution, mode choice and parking factors.
     31  Cambridge Systematics, Inc., "Travel Model Improvement Program:  Short Term Model Improvements,"
      prepared for the Federal Highway Administration, U.S. Department of Transportation, 1994.
      Greig Harvey et al,  "A Manual of Regional Transportation Modeling Practice for Air Quality Analysis,"
      National Association of Regional Councils, December 1993
3-46                                                                  U. S. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                                                                      to Reduce Transportation Emissions


               •   Disaggregating household by vehicles (income) and size and workers; and.
               .•   Trip chaining and other trip adjustments.

     Procedures  exist within current practice or are being looked at in research  efforts  that offer to
     enhance trip generation capabilities in dealing with pricing.

     3.3.4.1  Person Trips versus Vehicle Trips

     Most trip generation models furnish estimates of daily person  trips;  however, these trips  are
     commonly limited to vehicle trips and exclude non-vehicle trips such as walking and bicycle
     trips.32  Smaller planning agencies may even restrict trip production estimates to person trips made
     in private vehicles (eliminating transit entirely).            ,.

     Exclusion of non-motorized or non-automobile trips from trip generation analysis generally stems
     from capabilities elsewhere  in the model system, especially mode  choice and trip  distribution
     analyses. Including non-automobile trips in vehicle trip-only systems implies a separate need for a
     mode choice model,  which  greatly increases the system's  complexity.  Yet,  agencies that use
     vehicle trip-only approaches seldom  deal  with non-automobile travel.  Consideration  of non-
     motorized modes requires an approach that:

               •  Looks at a finer-grained pattern and mix of land uses, street space allocation and
                  design; and

               •  Considers proximity, as well as mobility.
                                      -           .              ' .           •                     «  i
     The availability and consideration of these modes  as travel options in response to pricing (or land
     use) actions is clearly important; their inclusion  in the analytic process may represent a  future
     standard for progressive model systems and planning agencies.  Areas  that have developed true
     person trip generation capabilities include Portland, Oregon  and Montgomery County, Maryland.
     The EPA and the FHWA are currently developing a manual  for MPOs to use to upgrade their
     micro-scale design features modeling practice. This manual  should be available in 1998 thrpugh
     the Travel Model Improvement Program.

     As a near-term goal, agencies that limit trip generation to Vehicle trips should endeavor to expand
     their systems to full person trips.


     3.3.4.2  Inclusion of Transportation and Cost Variables

     Few trip generation models relate the effect of transportation level of service or cost to household
     trip productions.  More research is needed to determine whether a major change in transportation
     cost or  level of service would affect the overall  number of person  trips made by a household,
     32
       A vehicle-trip is the one-way movement of a vehicle between two points. A passenger- or person-trip is
      defined as one passenger making a one-way trip from origin to destination.  Person-trips are measured as the
      sum of the number of passengers, added across all vehicle-trips. Thus, person-trips are equal to vehicle-trips
      only if the average vehicle occupancy is one passenger per vehicle.
U. S. Environmental Protection Agency                                                                 3.47

-------
 Technical Methods for Analyzing Pricing Measures"
 to Reduce Transportation Emissions
      particularly  in  light  of  non-motorized  travel  options   and  growing   opportunities   for
      telecommunications or delivery as substitutes for travel.

      Transportation characteristics and  costs might  be incorporated through the  development of a
      measure of composite impedance (see Sec. 3.3.5.1 below).  Composite impedance is a measure
      which combines the time and cost of highway  and transit (and perhaps other) modes  into one
      statistic.  In the trip generation  context, composite impedance is the unit of measure used to
      determine how far travelers are willing to go from a defined origin. More specifically, the question
      is phrased as "How many jobs (or  other activity variable) in  each zone can be reached within X
      units of composite impedance from  that zone?"  The value of X, the tolerance  limit for composite
      impedance, would be determined in the calibration phase.

      Analytically, an increased road price (or reduced level of service) would increase the composite
      impedance associated  with  trips  along that  road, thus reducing  the  destinations that could be
      reached within a fixed impedance limit from zones in the area.  As composite impedance increases
      and accessibility decreases, trip frequency should also decrease  because in general  it becomes
      more difficult Jo go anywhere, particularly by vehicle.

      The Bay Area model system includes composite impedance (termed accessibility)  in its  trip
      generation model for home-based shopping, and is one example of an attempt to include such a
      relationship in  trip generation (see the NARC Manual,  Sec.  3.3.5, pp.  3-36-37 for more
      information).

      Trip generation also can respond  indirectly to pricing.  If pricing applications are evaluated with
      models or procedures that affect an  earlier-step change in land-use/activity location, or in vehicle
      ownership levels through  some type of composite impedance relationship, then trip generation
      estimates would be subsequently affected.  The Portland, Oregon vehicle ownership model, for
      example, is sensitive to transit service accessibility (and pedestrian environmental quality) and, in
      turn, trip generation is sensitive to vehicle ownership.


      3.3.4.3  Disaggregating Households by Vehicles (Income) and Size

      Most trip generation and distribution models are handicapped because they must use zonal-average
      data.  One of the more important recent modeling improvements is disaggregation of certain zonal
      data by characteristics that influence travel. A common practice is to split households jointly by:

               •   Household  size  and  vehicles available,  used by  Washington, DC,  Detroit,
                  Philadelphia, San Francisco, and Portland, Oregon; or

               •   Household size and household  income,  used by New Orleans,  Atlanta,  Northern
                  New Jersey, Dallas, Denver, Phoenix and Minneapolis-St. Paul.

      Such disaggregation of household  data  produces cells that  are more homogeneous  in nature,
      reducing the variance in trip rates, Using income or vehicles as one of the dimensions also allows
     the explicit identification of trips by  some measure of wealth, which can be useful for modeling the
      differential effect of pricing on various groups.

     Typically in these submodels, analysts use Census data to establish the relationship between zonal
      averages and the distribution of households by number of persons  (I,12, 3, 4 and 5+) and number


3 "48                                                               U. S.  Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                  1                                                     to Reduce Transportation Emissions

      of vehicles (0,  1,  2+).   From these individual distributions,  the  region's joint  distribution  of
      households by size and vehicles is used to estimate the equivalent joint distribution for each-zone
      The reader is encouraged to consult Appendix A for a discussion of this procedure, demonstrated
      by the case study model.


      3.3.4.4  Trip Chaining and Other Trip Adjustments

      Trip chaining is one likely response to an increase in the cost of travel (or reductions in the number
      of vehicles available to a household); that is, individuals group trips to make more efficient use  of
      vehicles o^ reduce the cost associated with a given trip.  This could have an effect on total VMT.
      and potentially an impact on cold starts.                                                -

      No conventional models deal explicitly with trip chaining33, although researches ongoing under the
      Travel Model Improvement Program, and improved operational models that deal with this activity
      should be available in the next few years.34, 3S, 3V37  Data from the 1974 and  1980 gasoline
      shortages might be investigated to gauge the impact of chaining responses to travel constraints.

      Current efforts to develop activity-based models represents  one  important approach to addressing
      methodological difficulties posed by trip chaining and other trip adjustments. In this approach, the
      number of trips  generated  by  a household is. not considered to be a function of household
      characteristics directly, but of activities that generate travel. This approach is expected to be useful
      because respondents  typically  can provide  better  information  about the  number of  travel-
      generating activities they are  involved in, and can provide better information about the number of
      trips associated with eaph activity.

      Households  can affect overall  trip rates through  other trip generation  adjustments,  such  as '
      telecommuting and other telecommunications applications and alternative schedules (compressed
      work weeks), which reduce the physical number of home-based work trips.  The TDM Evaluation
      Model allows for some estimation of the change  in person trips due to these strategies (see Section
      3.3.6.9. below).   However, there is some  question as to whether  households that engage in
      telecommuting or alternate schedules actually, reduce overall  household trip rates or simply -
     33 Montgomery County, Maryland: TRAVEL2 Model deals with chaining in a form that is sensitive to land
      use (Maryland National Capital Parks and Planning Commission, 1991).
     34 Erniq Cascetta (University of Naples - Italy), Agostinq Nuzzolo (University of Tor Vergata - Rome - Italy)
      Vito  Velardi (ELASIS - Salerno'- Italy),  "A System of Mathematical Models for the  Evaluation of
      Integrated Traffic  Planning and Control  Policies.^ Unpublished .research paper, Laboratorio Richerche
      Gestione e Controllo Traffico, Salerno, Italy, 1993.              .                             '•''',
     3? Moshe Ben-Akiva, John L.  Bowman, and Dinesh  Gopinath, "Travel Demand Model System for the
      Information Era"-Transportation, vol. 23, pp. 241-266, 1996.   ,     .                '
       Yoram Shiftan, "A Practical Approach  to Incorporate Trip Chaining in Urban Travel  Models,"  Fifth
      National Conference on Transportation Planning Methods Application, a Compendium of Papers, 1995.
     37 Konstadinos Goulias, Ram Pendyala, and Ryuichi Kitamura, "Practical Method for the Estimation of Trip
      Generation and Trip Chaining,"  University of California Transportation Center, University of California
      Berkeley, 1991.  ,                  •  "                                                .        '
U. S. Environmental Protection Agency                •:    !         '       3*49"

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions


     substitute other travel.  Research is still necessary on this issue, before these pattern shifts can be
     assumed to lower household trip rates.38


     3.3.5 Trip Distribution
                 ',"''•                       *'    ,           '              *-'
     The trip distribution analytic step links together the "trip ends"  (i.e., productions and attractions)
     that are  estimated in trip generation, into specific origin-destination  trip movements, or  "trip
     tables."                                                 ,              •

     In most instances, the distribution of trips from an origin  zone to a potential destination zone is
     performed using a gravity model.  This model distributes trips as a function of the number of trip
     attractions relative to  the degree  of separation  among the zones. Almost  all areas measure
     separation by highway travel time.

     Researchers have  long  known, however, that factors other than highway time play a role in the
     allocation  of trips among destinations.  For example, there  is considerable evidence that the
     presence of good transit service between two zones will increase the number of person  trips
     between  those zones. A logical extension of this concept  is that other dimensions of separation,
     such as prices that users pay for transportation, also influence destination choice. The significance
     of these factors, which  is supported by intuitive logic and  empirical observation, should be taken
     into account.

     This  section discusses  several topics  that affect analytic  accuracy  of the trip distribution  step:
     composite impedance, income stratification, travel, time equivalency,  over-allocation of attractions,
     trip distribution as "destination choice," and importance of short trips.

     3.3.5.1  Composite Impedance

     Gravity modeling that uses composite impedance as its measure of separation is one approach that
     is being tested and used as a way of accounting for pricing effects in trip distribution.  Composite
     impedance, a measure  that combines the time and cost of  travel modes  into one  statistic,  is
     generally computed as  the natural  logarithm of the sum of the exponentiated disutilities for all
     available modes (including non-motorized), i.e., the natural log of the denominator of a logit mode
     choice model, also called the " log sum."

     This approach has been used in the model systems of New Orleans, San Francisco, Boston, Atlanta
     and Denver.39  Most areas that use this formulation use it only for work trips, but theoretically it is
     applicable to all trip purposes.  The Case  Study Model in Appendix A also  illustrates such an
     application.
     38 P. Mokhatarian et al., "Effectiveness of Telecommuting as a Transportation Control Measure," University
      of California, Davis, 1994.
     39 Greig Harvey et al., "A Manual of Regional Transportation Modeling Practice for Air Quality Analysis,"
      National Association of Regional Councils, December 1993, Sec. 3.3.6, pp. 3-47,48.
3-50                                •                                 U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                         ,                             to Reduce Transportation Emissions

      3.3.5.2 Income Stratification

      The New  Orleans model  adds another dimension to trip distribution  that merits consideration:
      income, stratification.  The composite impedance calculation accounts for income  level of the
      traveler, as well as time and cost, of all modes.  This is practical because all elements of the New
      Orleans model (from trip generation to mode choice) are income-stratified.

      The benefit of income stratification in this context is that the effect of price on destination choice
      is.differentiated  according to traveler  [household]  income level.   Since  pricing changes will
      generally influence low income travelers more  strongly than high income,  stratifying by income
      allows the trip destination  choice to vary realistically in discontinuous fashion relative to income.
      Other models provide similar sensitivity by dividing price variables by average household income
      for travelers expressed as cents per minute.

     "     '               .                -'                 *  ',      '                        •     '•
      3.3.5.3  Travel Time Equivalency

      Many agencies include travel time only in their measure of separation for trip distribution and
      mode choice  analyses.'  These agencies may approximate the  effect of an increase in cost for
      certain measures by translating cost into an "equivalent"  value of travel  time.  Such agencies
      incorporate a  cost increase from market-based measures as an increase in travel time impedance
      for a given origin-destination  pair. The increment in time is cpmputed  by using traveler value of
      time to convert cost to minutes of equal value, which can be interpreted from the coefficients in the
      mode choice model.40

    ,  Time equivalents can vary  by situation and type  of market measure to reflect variations in traveler
      price sensitivity.  For example, a pricing  action that affects  automobile operating  cost would
      probably translate as a lower  value of, time than an  out-of-pocket cost  such as a parking charge.
     Also, non-work travel would be expected to exhibit a "lower value of time/higher value of cost"
     ratio than work travel.


     3.3.5.4  Over-Allocation of Attractions

     With a doubly-constrained trip distribution procedure, such as the standard gravity model, it may
     prove necessary for trip attraction estimates to, be sensitive to user costs. Otherwise, trip pattern
     distortions could occur under certain pricing scenarios as the gravity  model tries to reallocate
     attractions  in  response to cost changes,  but the attraction  model would maintain the  same
     attractions  in a zone.  This could result in over-allocation of attractions.

     This problem  can  be minimized  by including  some type of price sensitive  variable, such  as
     composite  impedance, in the attraction model.  Another option is to switch to a logit destination
     choice model to allocate attractions, as discussed below.
     40 In the Washington, D.C. Council of Governments model, for example, the ratio of the home-based work
      auto running time coefficient to the auto operating cost coefficient = 0.0173/0.0035, or about $0.05 per
      minute.                                .      .                                 •     •     r
U. S. Environmental Protection Agency

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
      3.3.5.5  Trip Distribution as "Destination Choice"

      Trip distribution analysis attempts to represent the process of travelers choosing destinations.
      Ultimately, it approximates behavior by matching origin and destination " potentials," offset by the
      degree of difficulty in traveling between zone pairs.

      Some areas have implemented a more advanced procedure  that explicitly models destination
      choice as a behavioral process, much like mode choice.  These models estimate the probability that
      a traveler will choose a particular destination based on the "utility" or attractiveness of that place
      relative to other places; attractiveness is defined by characteristics of the destination, the traveler,
      and travel conditions, including cost.

      The  behavioral approach  offers  a more realistic  evaluation of destination  choice  behavior,
      particularly  with  regard to cost  factors, than  conventional gravity models, which  require
      considerable manipulation with adjustment factors  to match "behavior" with  observed origin-
      destination trip volumes. Portland, Oregon uses such a model,  with separate  formulations  for
      home-based work, school, college, and other; non-home based work-related; and non-work.41

      Some areas also  have found  it advantageous to model destination  choice and  mode choice
      simultaneously, particularly for non-work travel where these decisions are  so interrelated.  The
      Bay Area exemplifies this approach with its logit probability model for shopping trips.42
       Greig Harvey et al., "A Manual of Regional Transportation Modeling Practice for Air Quality Analysis,"
      National Association of Regional Councils, December .1993, Sec. 3.3.6, pp. 3-45,47.
     42 Greig Harvey et al., "A Manual of Regional Transportation Modeling Practice for Air Quality Analysis,"
      National Association of Regional Councils, December 1993, Sec. 3.6, pp. 3-47,48.
3-52
                                                                     U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                        •             to Reduce Transportation Emissions


      3.3.5.6  Importance of Short Trips

      Pricing measures place a great premium on being able to effectively model short-distance trips,
      since short vehicle trips contribute substantial emissions from cold starts. These are the trips that
      are perhaps most amenable to diversion to transit or non-motorized  modes.  The importance of
      short trips in emissions policies raises concerns about the existing  level of detail  in  analytical
      representation of transportation highway and zonal networks.

      In fact the importance of short trips could justify an increased level of network detail, down to the
      level of minor collectors and perhaps local streets in applicable areas,  accompanied by smaller
      traffic  analysis zones to  overcome problems with  non-representation , of  " internal"   trips.
      Moreover, this same concern  might argue  for better  data on walk and bicycle  trips and
      infrastructure, transit accessibility, and information on the mix'of land uses.


      3.3.6  Mode Choice


      The mode choice step estimates the share  of trips made using each mode of travel.  Within  the
      four-step process, mode choice models generally stand out as  being the most rigorously developed
      and  directly price-sensitive analytic procedure.   Many larger urban areas  have developed
      sophisticated discrete-choice logit models that estimate the share of person trips by mode, based on
      the socioeconomic level of the traveler, and the time and user cost attributes of the various modes.
      Models typically address the major vehicular modes—autos and transit—with  shared-ride  often
      treated as a separate mode.  More advanced models also include non-motorized  modes (walk and
      bike), which is an important consideration for pricing.

      There are a number of characteristics or application procedures that can  maximize the sensitivity
      and accuracy of mode choice models with regard to pricing. Topics in this section include:

               •   Satisfactory model coefficients;                    .       ,           ,            ,
               •   Effects of inflation;                  . -         '

               •   Income stratification;,

               •   Differentiation by purpose;

               •   Integrating mode and destination choices;

               •   Nested versus multinomial models;
               •   Special modes;        ' '          '                 '   "      ;           '  .   ' .

               •   Sample enumeration methods;

               •   Stated preference surveys; and           "
               •   Off-line methods.                                                             .

     3^3.6.1   Satisfactory Model Coefficients
           i          '                            •   -                      .   '               f
     The ultimate product from a mode choice analysis depends most directly on the value and validity
     of the coefficients that are used to indicate traveler sensitivity tos pricing.  An area can take one of

U. S. Environmental Protection Agency                                                                3 53

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions


     several approaches if it is not comfortable with its coefficients, wishes to confirm its coefficients,
     or does not have particular coefficients:

            •   Develop  new coefficients/models:  Development of a new  mode choice model is a
                data- and time-intensive task. Yet, in conjunction with the 1990 Census, many areas
                commissioned new regional travel surveys that, taken together with Census data, serve
                as an excellent opportunity for model update or upgrade. A number of references can
                provide guidance to areas interested in developing new coefficients and/or models (see
                NARC Manual  Section 3,3.7.  for  information on models,  or Section  3.6.2  for
                information on data sources and requirements43). Areas .interested in developing new
                models or  greatly modifying current  models  should consider incorporating such
                features as: nesting,  income  stratification, non-motorized modes, and special pricing
             .   variables; these topics are discussed elsewhere in this section.
                 ,,„'' '     '    !  ' ,   '        'i'1 '       ,»'•.            i                 _
            • Conduct  surveys:   Coefficients  can be estimated  from formal  (cross-sectional or
              longitudinal) travel  surveys of revealed preference.  Assuming these surveys correctly
              capture the  appropriate travel conditions, they may be fairly accurate  in describing
              travel behavior response.44 However, revealed preference surveys must be conducted in
              response to actual pricing changes or signals.

                 Stated  preference  surveys, which  derive  estimates of response  sensitivity  from
              reactions to carefully staged "hypothetical" pricing situations (or from a combination of
              real and stated preference data),  may be an  acceptable alternative or complementary
              approach.  Stated preference surveys are discussed in more detail in section 3.3.6.8

            •  Transfer  coefficients: An area that is not comfortable with its cost coefficients for
              pricing analyses may consider transferring model coefficients from another urban area.
              The disaggregate nature of logit models generally supports transferability. Analysis of
              coefficients from different areas indicates that the variation among these coefficients is
              not as great as might be expected, suggesting that careful transference is possible.

              The most critical issue is ensuring that independent variables are consistent within the
              same model, and that the " donor" model itself is the result of a well-documented
              development and application  process. Inflation adjustment of price coefficients (to .
              reflect  the  year in which  the  original model was  developed) is  another  major
              consideration in transferring  pricing coefficients (inflation adjustment is discussed
              below).  For example, if the original model was calibrated using 1980 prices, the new
              model must either:  (1) have its prices converted to 1980 dollars; or (2) have its original
              cost coefficients adjusted to reflect the effect of inflation since 1980.

            •  Synthesize coefficients:  If it is not possible to derive a particular cost sensitivity from
              direct sources,  it may be acceptable to synthesize a coefficient value and then perform a
              series of sensitivity tests on it  The model would be run for a series of parameter values
              above and  below the hypothesized value;  the results of these simulations (mode choice,
              trip length, VMT, etc.) would be charted for each run.
    43 Greig Harvey et al, "A Manual of Regional Transportation Modeling Practice for Air Quality Analysis,"
     National Association of Regional Councils, December 1993.
      Cambridge Systematics, Inc., "Travel Model Improvement Program:  Short Term  Model Improvements,"
     prepared for the Federal Highway Administration, U.S. Department of Transportation, 1994!
                                                                   U. S. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                                                   ..      .  .  . ••        to Reduce Transportation Emissions

               Comparing the variation in results with theoretical values, coupled with judgment from
               experienced practitioners, can often help identify an acceptable value for a parameter.
               •Although such an analysis cannot conclusively identify the proper value for an unknown
               parameter (even though carefully reasoned and conducted), it  may provide a suitable
               interim result until better information can be found.

            •  Develop specialized coefficients: Some models have different  coefficients for parking
               cost than for regular automobile operating cost; these may be higher by a factor of 1.5 to
               2.0.  This difference may reflect a higher sensitivity to parking as an out-of-pocket cost,
               or it may be a function of higher density/more restricted parking supply in areas that are
               priced.   Parking cost coefficients may be synthesized  from other models.   Separate
             .  coefficients'for roadway  tolls  are  not in widespread  use.    Both parking  and toll
               coefficients might  be approximated initially by such  methods  as  stated  preference
               surveys.


     3.3.6.2  Effects of Inflation

     Traveler sensitivity to a given dollar price, reflected in model  coefficients, will change over time
     with the general rate of inflation.  As a general rule, the monetary value of a pricing policy should
     be deflated to reflect the change  in prices between the forecast year and the year in which the
     model is calibrated.   Extreme care should be taken in forecasting what the value of a price will be
     in a. future year. It is commonly assumed that all travel prices will increase'in the future at the
     same rate as inflation, and, thus, that there is no change in the relative price among travel choices.
     This may be an acceptable assumption, though it should be noted that automobile operating costs
     have generally been dropping in real terms since WWII.45                      '

     3.3.6.3  Income Stratification

     One of the biggest  issues raised  in evaluating pricing strategies is how  impacts vary with the
     income level of the traveler. Two concerns are raised:  assuring an accurate estimate of traveler
     response in light of real income differences and  accounting for the distribution  of winners and
     losers. Since mode choice is such a significant  step in travel demand analysis of pricing policies
     (in terms of impact and cost specificity), it is important to account directly for income differences
     when applying mode choice models.
                                   • ^  I                                  -       .
    -f Generally,  income  differences can  be  accommodated by  applying  purpose-specific models
     independently to each income stratum (usually  quartiles or quintiles). It is essential that the trip
     tables brought into mode choice from trip distribution already have  this segmentation. Readers are
     encouraged to consult both appendices for examples of how this procedure is performed.  Another
       It should be noted, however, that the environment for non-auto modes like walking, biking and transit, has
      steadily eroded, further decreasing the utility of these modes relative to the auto which has been increasing
     , in terms of vehicular and environmental design features. This suggests that the generalized cost for auto
      may continue  to decrease in its rate over time, le'ading to greater demand than would occur under the
      assumption of constant pricing.
U. S. Environmental Protection Agency                                                                3.55

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions


      way income differences can be  accommodated  is by developing mode choice  models with
      coefficients that are directly sensitive to traveler income.46
      3.3.6.4 Differentiation by Purpose

      Traveler response is likely to vary by trip purpose and mode choice. Coefficients valid for one
      purpose may be invalid for another type of trip.  Some model systems include a true mode choice
      model only for home-based work (HBW) trips.  Non-work trips, usually home-based other (HBO)
      and non-home based (NHB)  are calculated as a distance-based factor  multiplied by ,the HBW
      shares. This implies that the non-work mode shares move up or down in the same proportion as
      the HBW mode shares, which is unlikely.

      Evidence suggests that mode choice in non-work trips is more sensitive to price and less sensitive
      to time than for work trips. Therefore, it is preferable to develop separate mode choice coefficients
      for work and non-work trips.  This is particularly important if a composite impedance measure is
      developed from the mode choice model to be used in trip distribution and possibly elsewhere.

      While the Census of Transportation (CTPP) has been a robust source of calibration data for work
     , trips, more extensive home interview surveys (perhaps as large as 1.0 to  1.5%) may be  needed to
      furnish enough observations of non-work trips to permit calibration of mode  choice coefficients.
      The NPTS  micro-sample provides one source  for non-work model estimation.  Montgomery
      County, MD has such a model system, developed from about a 0.5% to 1.0% sample.47


      3.3.6.5 Nested versus Multinomial Models

      Most first-generation mode choice models have been multinomial, models, where all modes in the
      alternative set are  assumed to represent  independent options in  the model.  However, when
      alternatives  are closely related, straight multinomial logit models  can  pose problems because they
      assume that the alternatives represent realistic independent choices (e.g., a red bus versus a blue
      bus on the same service schedule).

      Probit model formulations are one way of addressing this issue, but they are highly complex and
      have difficulty incorporating more than two modes. The preferred approach has become the nested
      logit, which is easier and more practical to develop and apply. The nested model splits the choice
      process into tiers, where the choices at any given level are relatively independent. This ensures
      that two similar modes, for example, do not capture an independent share of the total, but rather
      split the subtotal after the " generic" share for that type of mode is determined.

    .  There are  many resources on logit and probit models for the interested reader  (consult the NARC
      Manual, Section 3.3.7, or Ben-Akiva and Lerman for references48); examples of nested logit mode
     46 See M. Ben-Akiva and S. Lerman, Discrete Choice Analysis: Theory and Application to Travel Demand
      MIT Press, Cambridge, MA, 1985.
                               >   >,>                |                     |
     47 "Montgomery County, Maryland:  Travel2 model deals with chaining in a form that is sensitive to land
      use," Maryland National Capital Parks and Planning Commission, 1991.
     45 Greig Harvey et al., "A Manual of Regional Transportation Modeling Practice for Air Quality Analysis,"
      National Association of Regional Councils, December 1993.

(Footnote continued on next page...)

3~*°                                                             ' U. S. Environmental Protection Agency

-------
                                                         Technical Methods for Analyzing Pricing Measures
                              ,                                       to Reduce Transportation Emissions


      choice models are used in Portland, Oregon and San Francisco.  The reader also  is advised to       )
      consult the illustration in the Case Study Model in Appendix A, in which a nested logit model was
      adapted and applied for mode choice.                       '


      3.3.6.6 Special Modes

      Some modes are not well accommodated by convention^mode  choice  procedures; this is a
      function of the models as well as the data available.  This is a second-order, but important, issue
      for pricing because these modes are often key options.  Examples include special transit services or
      improvements that depart from the coded network, including auto-access/park-and-ride, and non-
      motorized modes.  Bike and walk modes are considered in only a small number of metropolitan
      area models, with Portland being an example.                         •                •   "

      Modeling bike and walk is made even more challenging because many of the trips are intra-zonal
      (and hence not well represented), may not be well represented in the base data on trip rates (field
      and subsequent trip generation), and require special descriptors in the models. Portland, Oregon
      includes non-motorized modes in their mode choice model,49 as do Montgomery County, Maryland
      and Sacramento, California.

      Most areas will have to analyze these modes "off-model," usually relying on evidence from other
      implementations  and adjusting regional trip tables  accordingly. 50  Another possible way of
      representing these modes might be through sample enumeration methods, discussed below.

      Another example of a "special  mode"  is any mode  used  to  access transit.  Access modes can
      include walking,  bicycling, auto travel, and transit itself. Many of the difficulties associated with
      modeling  transit-access modes are described above, with the additional complication that transit
      access points may need to be treated as trip attractors themselves. Mode of access to transit has
      been modeled in  some cases as part of the nested logit mode choice model. In the mode choice
      model, walk and  drive access are modeled at the same level or nest. The Washington Council of
      Governments (WashCOG) mode  choice model is an example of transit access modeled in this
      way.       ,                          '. •


      3.3.6.7 Sample Enumeration Methods

      Some analysts have suggested that they  can realize a higher level of flexibility and  realism, and
     possibly accuracy, if they evaluate pricing actions through a sample of households, rather'than
      M. Ben-Akiva and S. Lerman, Discrete Choice Analysis: Theory and Application to Travel Demand MIT
      Press, Cambridge, MA, 1985                                  .              /

     49 Cambridge Systematics, Inc., Calthorpe Associates, and Parsons Brinckerhoff Quade & Douglas, Inc
      Making the Land Use Transportation Air Quality Connection:  The Pedestrian Environment Volume 4A*
      for 1,000 Friends of Oregon, December 1993.
     50 Delaware Valley Regional  Planning Commission,  "An Analysis of Potential Transportation Control
      Measures for Implementation in the DVRPC Region," May 1994.

     51 FY 94 Development Program for MWCOG Travel Forecasting Models; Volume A: Current Applications
      June 1994                 '
U. S. Environmental Protection Agency                         •, -„

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
      work with the zonal averages of the conventional 4-step model.  In this approach, called sample
      enumeration, a sample of households (generally derived from a regional home interview survey,
      although Census data may be a source) is used to represent the regional travel base, and becomes
      the level at which the models are applied.

      While this method is being introduced in connection with mode choice, in fact it may be applied to
      various other steps in the travel demand hierarchy.  The household sample is drawn to represent
      the range  of geographic,  socioeconomic  and travel situations that would occur  in the regional
      population.  Generally, the sample is related (by zonal location) to the 4-step modeling process
      through  the regional trip tables and the corresponding travel times and  costs of the transportation
      alternatives.

      Policies  are then  "applied"  directly to the households; evaluation takes advantage of important
      information on household composition, vehicle ownership, income, and  proximity to activities and
      alternatives to generate fairly robust estimates  of response to the given policy.  These results are
      then "weighted" back to reflect the behavior of the population.
                                                                 1 /
      The process may or may not involve actual modification of regional trip tables and performance of
      traffic assignment in estimating emissions.  In the view of practitioners, this approach implies a
      tradeoff between the accuracy and flexibility that can be derived from a household-level analysis
      and  the  intricacies of a  network-based  traffic assignment, though traffic assignment can be
      performed.

      This technique has been  applied in a number of locations, including California and  Washington
      state, and  has shown good potential-52* 53» 54   It offers the additional advantage of flexibility in
      analyzing certain  alternatives or behaviors that can  be traced at a household level, but not at a
      zonal aggregation level.  For example, analysis at the household level is particularly  useful for
      tracing  short bike/walk trips,  access   to  special alternatives,  automobile   ownership/fleet
      composition/usage issues, and greater specificity of particular pricing issues (such  as parking cash
      out, etc.). The reader is encouraged to consult Appendix B, which reports on an application of this
      approach.


      3.3.6.8  Stated Preference Surveys
                                                                 ii                  '         '
      Stated preference surveys represent an interesting option  for gaining  insight  into mode  choice
      response to policies, alternatives, and situations when solid empirical information does not exist.55
     52 Greig Harvey, Curbing Gridlock:  Peak Period Fees to Relieve Congestion,  Transportation Research
      Board Special Report 242, Vol. 2, National Academy Press, Washington, D.C., 1994.
     53 Greig Harvey, "Transportation Pricing and Travel Behavior," presented at 69th Annual Western Economic
      Association International Conference, June 1994.
       Greig Harvey et al., "Transportation Pricing for California: An Assessment of the Air Quality, Congestion,
      Energy, and Equity Impacts," Vol.  1: Summary Report, Draft, California Air Resources Board, June 1995.
       Cambridge Systematics,  Inc. and Barton-Aschman Associates, Travel Model Improvement Program: Short
      Term Travel Model Improvements, U.S. Department of Transportation, U.S.  Environmental Protection
      Agency, U.S. Department of Energy, October 1994, pp. 1-7,1-8.
3-58                                                  '               if. S. Environmental Protection Agency
54
55

-------
                                                           Technical Methods for Analyzing Pricing Measures
                                     •   ""                              to Reduce Transportation Emissions

      Stated preference surveys might be used, for example, because no data has yet been generated on a
      new type of travel mode or a special type of pricing instrument with unique characteristics.

      In a stated preference approach, it is possible to derive statistical  estimates of,"tradeoff rates
      between various alternatives or their attributes by! making respondents choose from among them in
      measured ways that indicate the relative importance of key attributes.  These rates can then be
      evaluated in relation to the type of traveler and his/her circumstance.

      The validity of the derived statistical relationships relies on how well the alternatives are portrayed
      to (and understood by) the respondent, and  itheir comparison to known " standards."  While stated
      preference surveys rely on  hypothetical  situations,  comparison  of "elasticity"  relationships
      derived from stated preference with more conventional revealed preference  surveys or models
      have shown surprising corroboration.  The results from these surveys should be used with caution,
      but they  offer'an important interim tool for agencies to estimate relationships between pricing
      instruments and travel behavior response, not just in mode  choice but in relation to destination,
      time of day, route choice, etc.

      Stated preference  methods were developed by the private market research industry, and have been
      used successfully for many years to aid companies  in  identifying the critical attributes of their.
      product, and  maximizing  those attributes  to gain market  share over competitors.   Use of the
      techniques in  transportation is a fairly  recent development, but there are numerous examples of its
      successful application-56, 57, 58, 59,  60   Portland and Denver  have both utilized stated preference
      methods in relation to pricing. Portland has used stated preference methods to explore time of day
     .choice^ conducting its survey in conjunction with its  recent home interview'survey, .while Denver
      used such a survey to assist in development of a route choice model  (described under traffic
    .  assignment, below).


     3.3.6.9  Off-Line Methods

     A number of planning agencies may have limited ability to perform mode choice analysis,  either
     because they  do  not have mode choice models or the models they have are  insufficient for
     evaluation of the particular measure.  The NARC Manual cites a number of "off-line" methods
     that may be valuable as substitute or complementary approaches.  Examples include:

            •  TDM Evaluation Model: The Federal Highway Administration is distributing a special
               software product that can substitute for or  enhance a conventional mode choice model.
  M. Ben-Akiva and S. Lerman,  Discrete Choice Analysis:  Theory and Application to Travel Demand
 MIT Press, Cambridge, Massachusetts, 1985.                    ,                                ;'
57 M. Ben-Akiva et al.,  "Combining Revealed and Stated-Preference Data," prepared for publication in
 Marketing Better.       .
  Cambridge Systematics, Inc. and Hague Consulting Group, VFT Feasibility Study:  Market Analysis Final
 Report, July 1988.                                             ,   '     '             *  >  .

  David A. Hensher, P. Truong, and P.O. Barnard, "The Role of Stated Preference Methods in Studies of
 Travel Choice," Journal of Transport Economics and Policy; January 1988, pp. 45-57.
  J. Bates,  "Econometric Issues in Stated Preference Analysis," Journal  of Transport Economics  and
 Policy, January 1988,
U: S. Environmental Protection Agency
     58
     59
     60

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
               The TDM Model is a spreadsheet approach that incorporates a "pivot-point" procedure
               to estimate mode choice changes. The model forecasts the change from an initial mode
               share based on information regarding changes in decision variables associated with the
               policy action.   A  wide variety of pricing  and other strategies can be considered,
               individually or grouped into programs,  with  great flexibility to vary .the levels of the
               strategies and target their application.

               The starting base may be trip tables from an existing planning process, special trip
               tables formed from survey data, or simply aggregate estimates of person, vehicle and/or
               transit trips.  The coefficients in this model  have been "synthesized"  from national
               experience and can be  altered by the user to use local or other coefficient estimates.
               Results are both aggregate and in trip table format.  The trip table results can be returned
               to the four-step process (if applicable) for traffic assignment and emissions!6'
               The TDM Model has special capabilities for trip table manipulation or creation.  The
               model interfaces  directly with the common planning software packages (TRANPLAN,
               MinUTP and EMME/2), enabling easy  interactions with the core model system.  The
               TDM Model allows analysts to alter starting trip tables, or  create a trip table from
               scratch (e.g., for transit or vanpool trips).  Trip tables may be  created entirely from
               scratch using survey data, if desired, or more simplistic estimates  of total tripmaking
               and starting mode shares may be used to fuel the model.  For areas with vehicle-only
               trip  tables,  it  may be feasible  to use the  model  "in  reverse," knowing average
               occupancy rate,  to create a person  trip table suitable for performing mode  choice
               analysis. (Unfortunately, the TDM model does not incorporate non-motorized modes,
               and this should be adjusted for in its application.)

               Quick Response System:  The quick response system procedure (QRS) consists  of a
               collection of short-cut, parametric methods that may be used for transportation analysis
               in the absence of a local model system.  Procedures exist to cover the steps  of trip
               generation  through traffic assignment,  and include mode  choice.   It  exists in either
               microcomputer or manual form, and allows  users with minimum  data and  computer
               capability to form reasonable estimates of travel response.62;63

               Elasticity Factoring Methods:  Agencies without access  to computerized planning
               methods or to either of the above techniques may consider using factoring methods to
               derive estimates  of mode choice  responses  to pricing strategies. Such an analysis
               requires the following travel estimates:
                  -  Person trips
                  -  Vehicle trips (or average occupancy rate)
                  7  Transit trips (if applicable)
                  -  Vehicle miles of travel
     61  COMSIS  Corporation,' Users Guide:  Travel Demand Evaluation Model, for the Federal Highway
      Administration, 1993 (For information contact McTRANS, distributing agent for FHWA at 904-392-0378).
     62 COMSIS Corporation, "NCHRP Report 187: Quick Response Urban Travel Estimation Techniques and
      Transferable Parameters," 1978.
     63 COMSIS Corporation, ORS User's Manual, for the Federal Highway Administration, January 1984.
3-60                                                          .      U. S. Environmental Protection Agency

-------
                                                            Technical Methods for Analyzing Pricing Measures
                                                                        to Reduce Transportation Emissions

                 Ideally, these estimates would be in the form of trip tables, separated by'trip purpose
                 (work, non-work).  Given information on trip  length for each origin-destination trip
                 combination and starting modal shares, analysts should be able to estimate the change in
                 mode shares triggered by the respective pricing action.
               .  The report " Transportation Air Quality Analysis Sketch Planning Methods" M provides
                 a convenient spreadsheet method for performing this analysis.  The reader may also
                 consult various sources for estimates of elasticities, but should be careful to choose a
                 value or values appropriate to local conditions.65* 66>61

                 In the event that the agency does not have trip tables, it may be acceptable to  estimate
                 policy responses  from a more aggregate estimate of regional trip making.  Baseline
                 estimates can come from:                  ,                              .

                        -   Local travel survey data;                ,                           ;
                        -   Census CTPP estimates;                                    '
                        -   Population or activity  estimates factored by measured or ITE vehicle trip
                            rates; and
                        -   HPMS estimates.

                 Obviously, the more coarse the background estimate, the less accurate the estimated
                 response and the more simplistic must be the strategy definition.  The above references
                 are recommended as sources for elasticities and application methods.,

       3.3.7   Time of Day

       The ability of a model system to simulate differences in travel volumes and conditions at different
       times of the day is critical to realistic estimates  of speeds, congestion  levels, and emissions,
       particularly in  larger urban  areas with pronounced travel peaks.  Yet,  time of day allocation of
       travel is done  in a very approximate way,  using peaking factors derived from local surveys to
       apportion a certain percentage of daily trips into one or more peak hours or periods.

       Time of day is an important dimension for many TCMs, since changes in the level of congestion
       and time distribution of travel affect important calculations regarding speeds and travel volumes,
       which fuel the emissions model. Time of day takes on special  importance in relation to pricing
       measures that vary by time  of day, for example, roadway congestion pricing.  With major  price
       differentials between peak and off-peak, it is likely that travelers who are sensitive to cost and/or
       who have flexibility in trip  timing will consider shifting their departure to  another time period;
       alternatively, persons without flexibility and/or with a high value of time might pay a Higher  price
       to travel in the peak (and possibly even enjoy a higher level of service).
       64 Cambridge Systematics, Transportation Air Quality Analysis Sketch Planning Methods, Volume I:. Analysis
        Methods,  prepared for the U.S. Environmental Protection Agency, 1979.
         COMSIS Corporation, Users Guide:  Travel  Demand Evaluation Model, for the  Federal Highway
        Administration, 1993 (For information contact McTRANS, distributing agent for FHWA at 904-392-0378).
         R.H: Pratt Associates, Traveler Response to Transportation System  Changes, prepared for the Federal
        Highway Administration, Second Edition, 1981.          ,
         Apogee  Research, Inc.,  Costs and Effectiveness of Transportation Control  Measures:  A Review and
        Analysis of the Literature, prepared for the National Association of Regional Councils, January 1994.
.  U. S. Environmental Protection Agency      '                                                          3-61
65
66
67

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions


     This is a fairly complex choice that is not well handled by conventional peaking factor approaches.
     Research is underway in this area; sites interested  in time of day forecasting (because of their
     interest in congestion pricing types of measures) might investigate the following activities or
     options:

               •   Approximating methods;
               •   Time of day choice models;

               •   Time of day factors; and
               •   Stated preference surveys
                          "                   '     '           '           i
     3.3.7.1  Approximating Methods

     Current travel models do not attempt  to accommodate potential traveler  response  to  pricing
     measures that vary by time of day. Such analysis requires procedures that are sensitive to travel by
     time period, which in turn would require input data on price by period. This type of information
     would be cumbersome to create, but would not be outside the reach of current software.
                   "'               ,                        ' '             i

     For example, consider a model to estimate the percent of daily trips that begin in the morning peak
     three hours, which could be applied in two phases.  The first phase would calculate one percentage
     split for all O/D pairs, separately by trip purpose. The second phase would  consider the cost of
     making each  trip during  the morning peak and during the off-peak, and then adjust the initial
     percentage split based on the peak/off-peak cost differential for that O/D pair.

     Non-work trips would be expected to exhibit greater sensitivity to time of day price differentials
     than work trips because of greater traveler flexibility. The exact sensitivity of trip start times to
     price differentials is speculative in light of current knowledge, but it should be possible to make a
     reasoned estimate  using stated preference methods (see discussion below), which can  be  refined
     later using observed data.


     3.3.7.2  Time of Day Choice Models

     Domestic modeling practices have not featured formal modeling of time of day as a travel choice.
     Several researchers have devoted particular attention  to this element and performed work that may
     serve as a basis for agencies that wish to advance their capability.68>69

     3.3.7.3  Time of Day Factors
                                             1 „ '              »,          , i
     If trip generation models can be made to  include a measure ofaccessibility (composite impedance)
     that incorporates cosfand travel conditions at different times of day, it may be possible to reflect
     time of day preferences through the trip rates themselves, prior to  trip distribution  and mode
     68 Kenneth A. Small, "Fundamentals of Pure and Applied Economics," Urban Transportation Economics,
      Vol. 51, Harwood Academic Publishers, 1993.
     69 P.R. Stopher and A.H. Meyburg, Urban Transportation Modeling and Planning, Lexington Books, 1975.
3-62                    ,                                            u_ S. Environmental Protection Agency

-------
                                                            Technical Methods for Analyzing Pricing Measures
                                                           "  .          to Reduce Transportation Emissions


       choice.70  If overly  simple measures of accessibility are  used, however, the value of such an
       approach would be minimal. It is also possible to develop separate trip generation models by trip
       purpose by time of day, followed by separate trip distribution and mode choice analysis; this has
       been done by Montgomery County, Maryland.


       3.3.7.4 Stated Preference Surveys

       Traveler response to time of day pricing/service differences may be approximated through stated
       preference surveys. Sites also might consider synthesizing  time of day relationships from other
       travel contexts that employed differential time of day pricing, such as bridge tolls or peak/off-peak
       transit fare differentials.       '                                                              •
       3.3.8  Traffic Assignment/Route Choice


       The traffic assignment step estimates  the  traffic volumes for each  link in the transportation
       network, for each time period that is considered. Traffic assignment is critical to the transportation
       analysis and emissions process; it produces the estimates of link volumes, speeds/travel times, and
       vehicle  miles of travel (VMT) that  are  key inputs to earlier  planning steps, particularly trip
       distribution and mode choice, and are the primary determinants of emissions when input to the
       emissions factor model.

       Traffic  assignment consists  of assigning trips from a trip table to a transportation network
       according to  specified criteria, almost exclusively the minimum travel time path.  Travel cost is
       generally not included as a criterion,  although  some models account for tolls.71 It is feasible to
       include cost in assignment, and therefore extend its effect more broadly through, the other steps of
       the modeling process (see discussion below).

       The details of the traffic assignment process are not of central concern  in this manual.  It is clear,
       however, that the accuracy of travel speeds generated by traffic assignment are of key importance:
       First, travel speeds are critical to effective  analysis of all transportation alternatives, including
       those affected by pricing mechanisms.   Second,  travel speeds are  a direct input to emissions
       analysis.            '                    ,

       Additionally, there is concern about how well conventional assignment models deal with the issue
       of route choice when a roadway pricing system creates competing Opportunities.

       Techniques in use or under development that may be of interest in addressing these issues include:

                 •   Incremental loading and equilibrium assignment methods; , ,  "
                 •   -Dynamic assignment; .                            •
       70
         Greig Harvey et aL, "A Manual of Regional Transportation Modeling Practice for Air Quality Analysis,"
        National Association of Regional Councils, December 1993, p. 3-58.
         Cambridge Systematics, Inc. and Barton-Aschman Associates, " Travel Model Improvement Program:
        Short Term Travel  Model   Improvements,"  prepared  for U.S. Department of Transportation,' U.S.'
        Environmental Protection Agency, U.S. Department of Energy, October 1994, Sec. 4.0, p 4.1.
•  £/. S. Environmental Protection Agency                                          '                      3.53

-------
 Technical Methods for Analyzing Pricing Measures '
 to Reduce Transportation Emissions
                •   Traffic microsimulation;

                •   Speed feedback and recursion; and
               •   Path choice.


      3.3.8.1  Incremental Loading and Equilibrium Assignment Methods72

      Incremental loading and equilibrium assignment methods are largely replacing the all-or-nothing
      methods and provide much more realism in allocation of trips to paths. Among other things, these
      new methods greatly increase the realism with which level of service variations are represented
      and, in principle, provide a means for incorporating the cost of highway travel as -well as travel
      time, which can then be used in the earlier steps of the modeling process.


      3.3.8.2  Dynamic Assignment73

      Dynamic assignment, which is an improvement on conventional "static" assignment procedures,
      addresses problems with demand variations within an analysis time period and begins to account
      for queuing impacts on speed.


      3.3.8.3  Traffic Microsimulation74'75'76

      The outputs of a typical traffic  assignment analysis reflect average  link speeds, and  do not
      document variations  in those speeds that would result from queuing at intersections, entrance
      ramps, or under congested stop/go conditions.  Such variations in speed can  significantly affect
      vehicular emissions; valid estimation can be critical to proper computation of emissions. Research
      is ongoing in this area to improve the linkage between the assignment models and emissions
      analysis through use of traffic microsimulation techniques. The current NCHRP Project No. 25-7:
      "Improving Transportation Data for Mobile Source Emissions Estimates," is developing emission
      profiles that can be linked to a microsimulation to account for these effects.
     72 Greig Harvey et al, "A Manual of Regional Transportation Modeling Practice for Air Quality Analysis,"
      National Association of Regional Councils; December 1993, Section 3.3.9, p. 3-66.
       Cambridge Systematics, Inc. and Barton-Aschman Associates, "Travel Model Improvement Program:
      Short Term Travel Model  Improvements," prepared for the U.S. Department of Transportation, U.S.
      Environmental Protection Agency, U.S. Department of Energy, October 1994, Section 4.0, pp. 4-1 to 4-7.
       Cambridge Systematics, Inc. and Barton-Aschman Associates, "Travel Model Improvement Program:
      Short Term Travel Model  Improvements," prepared for the U.S. Department of Transportation, U.S.
      Environmental Protection Agency, U.S. Department of Energy, October 1994, Sec. 5.3, pp. 5-4 to 5-7.
     75 Patrick DeCorla-Souza et al., "A Trip-Based Approach to  Estimate Emissions  with EPA's MOBILE
      Model," Paper No. 940376, Transportation Research Board, 1994.
     76 E. Ruiter,  "Highway Vehicle Speed Estimation Procedures,"   prepared for the U.S. Environmental
      Protection Agency, 1991.
3-64                                              .                .  U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                                     to Reduce Transportation Emissions
     3.3.8.4  Speed Feedback and Recursion77
     Modeling accuracy would benefit from a formal recursive procedure that cycles information from
     assignment back to the earlier steps, thus ensuring compatible values among  the  steps when
     equilibrium is reached. If this is not done the inconsistency between the  assumptions that are
     inputs .to different modeling steps raises serious questions about the .reliability of the final outputs.
     If pricing is introduced as a factor in more of the modeling steps through a measure like composite
     impedance, then it. becomes even more important to ensure passage of this information through the
     ; relevant steps. Methods are currently under development by the Federal Highway Administration
     in conjunction with the Travel Model Improvement Program, and are in use in various places.78

     3.3.8.5  Path Choice        .

     Few conventional models are able to adequately model the effect of the cost of driving on route
     choice, which becomes a particular issue in evaluating roadway pricing. A common procedure is
     to convert the toll value to  an equivalent number of extra minutes by dividing it by some value of
     travel time. Thus, when a route is  considered as a potential path, the effect of the toll  would be
     reflected in an increased travel time.  This may be acceptable as a default analysis, but it greatly
     clouds  the issue of choice discrimination  among classes of travelers, some of whom would
     recognize the toll as a signal to avoid the facility and some of whom might be expected to pay the
     toll in exchange for a higher level of service (shorter travel time).

     Toll diversion models are a more realistic way to account for route choice responses to  pricing
     measures.  Toll diversion models split vehicle trips for each O/D pair into "toll" and "free" trips.
     This split is based on a comparison of toll and free path characteristics—mainly travel time and
     cost (at least one model also  includes "number of toll stops" as a variable).  The toll path is the
     one that would be taken if the trip were forced to use the toll facility and the free path is the one
     that  would be taken if the trip were prohibited from using the toll facility.   Popular planning
     software packages such  as TRANPLAN, MinUTP, and EMME2 are able to create and skim these
     paths.                                                 , "•   •

     Simplified versions of these models have been used for years by toll road planners, most often by
     applying manual  adjustments (based on rough estimates of travel time savings) to estimated toll
     road traffic volumes. The, process has been implemented within a computerized traffic forecasting
     model for use in planning a toll road in the Denver region.79

     This same procedure can be adapted to analysis of priced roadways.  Unfortunately, this method is
     subject to the same limitation as methods to deal with the mode choice and time of day effects of
     78
77 Greig Harvey et al. "A Manual of Regional Transportation Modeling Practice for Air Quality Analysis,"
 National Association of Regional Councils, December 1993, Sec. 3.3.9, pp. 3-70, 71.
 8 Montgomery County, Maryland has documented the effects of incorporating full feedback in a composite
 equilibrium trip distribution/mode choice/network assignment model (Levinson & Kuman, 1992)
 Incorporating Feedback in Travel Forecasting: Methods, .Pitfalls,  and Common Concerns, Travel Model
 Improvement Program, March 1996.                              ;           '
79 J. Heisler et al.,  "Estimating Toll Diversion Using Existing Transportation Planning Software,"  presented
 at the Second Conference on the Application of Transportation Planning Methods, April 1989.          "
U. S. Environmental Protection Agency                .                                .'•''<  3-65

-------
 Technical Methods for Analyzing Pricing Measures *	
 to Reduce Transportation Emissions


      congestion pricing: lack of observed data with which to calibrate a model. The Denver model is
      based on a small-sample stated preference survey of Denver area residents, most of whom had
      never used a toll road. The model has undergone extensive reasonableness checks and its results
      have been accepted as plausible.

      The reader is referred to the case study model application detailed in Appendix A for an example
      of this procedure.


      3.3.9  Emissions Modeling


      The final step in the analysis process is the translation of the transportation effects estimated in the
      other analytic steps into quantities of the respective mobile source pollutants: VOC, NOX and CO,
      and  PM.  Areas with comprehensive regional  model  systems generally perform this  step by
      applying EPA's MOBILESa model or,  in California,  the EMFAC7F model, and PART5 for
      reentrained dust emissions.   These models accept input from the traffic  assignment process on
      vehicle trip ends, VMT and  speeds, sorted by facility type, geography and time of day.  These
      inputs are then translated into emissions through the application of emissions factors, which are
      based on the regional vehicle fleet mix, average annual distance traveled by class, average travel
      speeds,  ambient temperatures, inspection  and maintenance programs, fuel policies and other
      factors.

      This manual does not detail the development and specific  operating features of the MOBILE or
      EMFAC models or the PART 5 model. Readers who desire more information can consult existing
      reference documents on these models.80*81*82 What is of particular concern here is the effect that
      particular pricing actions may have on the vehicle mix that is included in the models.  Pricing
      measures such as fuel taxes and emissions fees would have their principal effect on changes in the
      composition of the regional fleet.  It is this change in vehicle mix, separate  from (or in addition to)
      any  change in travel  level, that would alter  emissions levels.   Section 3.2.2.3 addressed issues
      associated with changes in vehicle mix and suggested methodologies and sources for estimating
      these changes.

      Areas that analyze fuel price or registration fee measures should use their analysis of the probable
      changes in  regional vehicle, fleet mix to alter the base distribution in MOBILE (or EMFAC).  The
      Case Study Model application in Appendix A presents an example of how this can be done.

      Users should recognize that changes in vehicle mix may be the compounded result of two shifts:

               •  Vehicles may shift in proportion by model year; and
      80 Sierra Research, Inc. and Jack Faucett Associates, "Evaluation of MOBILE Vehicle Emission Model,"
       prepared for Volpe National Transportation Systems Center, U.S. Department of Transportation, December
       1994.
      81 California Air Resources Board, "Methodology for Estimating Emissions from On-Road Motor Vehicles
       Volume I: EMFAC7F,"  Draft Report, June 1993.
      32 U.S.  Environmental Protection Agency,  Office of Mobile Sources, "Draft User's Guide to PART5:  A
       Program for Calculating Particle Emissions from Motor Vehicles," February 1995 (EPA-AA-AQAB-94-2).
1 3-66                                                          ,  ,    U. S. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                        .     .                            ,             to Reduce Transportation Emissions


               •   Vehicles may also shift by relative use (VMT).

     Historical trends, such as reflected  in the Nationwide Personal Transportation Survey arid the
     Motor Vehicle Manufacturers Association statistics, indicate a tendency to use new cars more than
     older ones; this  is reflected  in the annual VMT assigned to  each vehicle age-category in the
     analysis.  Thus, an analytic process that re-apportions fleet rnix by age (to favor newer vehicles),
     yet maintains mileage rates (VMT) in each age-stratum at their prior levels implicitly assumes that
     a shift to newer vehicles means more VMT each year — an  unlikely result.  Analytically, if not
     corrected, this  would result in newer, cleaner models in the MOBILE distribution being given an
     unrealistically high VMT weighting.

     Unless some type qf analysis or survey research is performed to support adjustments by year and
     mileage, users should normalize the distribution of mileages to  account for this effect.  In the
     example below, the ratio of pre-shift [VMT*share] to  post-shift [unadjusted VMT* share]  is 0.97
     (i.e., 6.69/6.89).  Adjusted  VMT is obtained by applying this ratio to the original VMT figures for
     each age stratum.
                                    i&Efcunpte of VMT Adjitstnieirt
Age
(yrs)
1
2
3
4
5 •
Pet of
Vehicles
15
12
10
8
5
' 50%
Annual
VMT
000
15
14
13
12
10
Revised
Veh. Pet
20
13
9
5
3
50%
Effect.
VMT
000
15
14
13
12
10
Adjusted
VMT (*)
000
14.6
13,6
.12.6
11.6
9.7 :
                                   Sum Original Veh. Dist. x Ann VMT = 6.69k;
                                   Sum Revised Veh. Dist. x Ann VMT = 6.89k
                                   Adjustment = 6.69/6.89 = 0.97
U. S. Environmental Protection Agency
                                                                                               3-67

-------

-------
                                                       Technical Methods for- Analyzing Pi-icing Measures'
                                                                 to Reduce Transportation Emissions
4.0    Implementation Issues
     4.1  Introduction
     The  ability  of market-based measures  to achieve their  intended  objectives  depends  on:
     (1) measures  that are- technically  effective;  and (2) an  implementation program  that  is
     administratively and  politically practicable.  Applying sound  analysis to the estimation of travel
     and emission impacts of market-based measures is the necessary first step in receiving emissions
     reduction credit for a proposed program. Submitting agencies "should also provide complementary
     documentation to assure reviewing agencies that reasonable progress can be expected in program
     implementation.

     In short, agencies proposing to incorporate and receive credit for market-based measures should
     address  the  implementation  issues outlined  in this section  to  receive EPA approval.  The
     significance of specific implementation issues will  vary from one area to another, and from one
     measure to another; this chapter presents generic guidelines that may hot apply in some instances.
     For example, this section notes that any needed technology must be shown to be available, yet.
     some measures require no technology deployment at all.

     "Implementation" covers a broad array of qoncerns. It includes substantive issues such as legal
     authority and administrative capacity.  It also includes more " politically-oriented" issues; clearly,
     the method and form in which market-based measures are packaged and  introduced influences
     public acceptance and ultimate program success. A sampling of implementation questions might
     include:

          •   What organization will  bear responsibility for ^implementation,  administration and
              enforcement?

          •   What other measures will be combined with the subject measure?  Would these other
              measures offer attractive and meaningful choices?   ,
          •   How significant are technical and financial uncertainties? How would public acceptance
              and program goals be affected if a "partial" program were implemented?

          •   How are the  resulting revenues allocated  — back to complementary transportation
              services, or used for other societal  purposes?. Should  fees be set so revenues equal the
              cost of these complementary uses/services, .that is, a revenue-neutral policy?
           •  How will equity issues be handled?

     The way in which issues such as these are addressed will determine: (1) how the measure .itself is
     defined and  its impact quantified; and (2) the likelihood that the measure will actually be
     implemented as proposed. In turn, these factors will  influence  the emissions credit determination.
U. S. Environmental Protection Agency                       .                                     4_j

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
      The Clean Air Act requires  EPA to recognize and  approve SIP  credit for any  quantifiable,
      permanent, and enforceable emissions reduction; it does not give EPA  authority to advise on
      implementation issues that do not"(directly) affect air quality. Thus, although all implementation
      issues contribute to ultimate program success, this section discusses only those implementation
      issues that affect an implementing body's ability to produce predicted air quality benefits. Several
      research institutions are available to advise cities and states on the non-air quality  aspects of
      implementing market-based measures. EPA and the Department of Transportation also are able to
      provide research and related assistance.1

      Experience with implementation of market-based measures is growing but is  still limited. EPA is
      committed to working with states and their agencies to resolve any implementation uncertainty and
      to respond to reasonable assumptions where experience has not yet provided a sure guide.  This
      chapter of the Guidance summarizes EPA's position on supporting documentation to address the
      following types of implementation issues:

                •   Public policy context;

                •   Legal authority;

                •   Administrative capacity;

                •   Technology availability; and

                •   Revenue generation and reinvestment.
     1 Much of the information in this chapter.has been summarized from Special Report 242, Volume 2, Curbing
      Gridlock: Peak-Period Fees to Relieve Traffic Congestion, Transportation Research Board, Washington
      D.C., 1994.
4-2
                                                                    U. S. Environmental Protection Agency

-------
                                                         Technical Methods for Analyzing Pricing Measures
                                                                    to Reduce Transportation Emissions
     4.2  EPA Position on Documentation


     States and their subsidiary agencies and implementing authorities requesting emissions credit for
     market-based  measures should provide a narrative summary of critical  implementation issues
     along with their technical analysis of travel and emissions impacts.

     While there are no standard analytical approaches for examining implementation issues, EPA
     needs enough information to  (1) be assured that  the intended market-based measures can be
     implemented over the timetable proposed and (2)  understand uncertainties that may affect  the
     implementation timetable and total emissions reductions.

     The discussion of implementation issues identified below should, at a minimum, describe:

               •    Current circumstances likely to affect  implementation in each issue area;

               •    Actions currently underway to advance implementation of proposed measures;
               •    Planned or proposed future actions, with projected timetables; and

               •    Implementation issues as yet unresolved, remaining uncertainties, and their potential
                   implications for the implementation of proposed measures.

     In each  case, the submitting agency should include relevant empirical or quantitative information
     or reference the appropriate sources. In short, if pricing measures are to be credible and creditable,
     sponsors  should   provide reasonable   evidence  that  the  actions  underway  will   lead  to
     implementation within the specified timetable.


     4.2.1  Public Policy Context


     Like any transportation or emissions policy, successful implementation  of proposed market-based
     measures will depend, in part, on how well they can  be introduced and integrated into the city's or
     state's general priorities.   It  is important  to  acknowledge  that  public relations and  public
     participation may be as influential  as any technical  analysis  in the implementation of pricing
     mechanisms.   F6r instance, the current consumer costs of travel may be widely perceived as the
     actual market costs. An increase in consumer costs would consequently be regarded  in a negative
     light, rather than as a movement toward  a more equitable cost structure.  Therefore, educating and
     involving the public is a very important element of any pricing strategy.

     Sponsors  should  briefly  describe  the  public policy setting in their  region, priorities  among
     competing objectives, and the extent to which implementation,of market-based measures has been
     integrated with major public .policy objectives.  Sponsors also should address issues related to
U. S. Environmental Protection Agency                                                               4.3

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions


     public  involvement  and  education  regarding  planned  implementation  of  market-based
     mechanisms.2
     4.2.2  Legal Authority


     Because this class of measures is new to many areas, and may involve multiple agencies, questions
     of legal authority may be raised that were not raised for other kinds of emissions-control strategies.
     Local units of government sponsoring, endorsing, or implementing market-based measures should
     document that they have the full legal power to do so, based on statutory or delegated authority.
     The credibility of a measure depends on the availability of such authority. Sponsors should review
     and confirm legal authority to ensure that they are able to enact and enforce the proposed market-
     based measures.

     Sponsoring agencies should answer the following questions:3

               •   What is the nature and source of the legal authority?
               •   Are the proposed measures a reasonable exercise of this authority?

               •   Are the revenue-raising provisions and pricing structures consistent with existing
                  taxing authority?                              '

     If any measures affect multiple jurisdictions, sponsors should detail the mechanisms to  be used
     and the basis for crafting agreements among multiple units of government.

     If implementation requires legislative action,  sponsors must detail the process,  prospects  and
     timetable for any required legislative action to establish or clarify legal authority.


     4.2.3  Administrative Capacity


     In addition to legal authority, sponsoring agencies  should demonstrate that capacity exists to
     effectively administer all aspects of the proposed measure.

     Sponsors of market-based measures should describe the administration that will  support  each
     measure, noting areas in which increased competence or capability may be needed and outlining
     the approach, and timetable for making necessary improvements.  Depending on the measure,
     implementation will require adequacy in some or all of the following areas:

              •  Staffing levels and skills;

              •  Computer resources;
      For additional discussion of issues related to public involvement, public acceptance, and equity,  see:
     USEPA, Opportunities to Improve Air Quality through Transportation Pricing Programs(EPA 420-R-97-
     004), September 1997.  This  document is also available at  the following world wide web  address:
     http://www.epa.gov/OMSWWW/gopher/Market/pricing.pdf
      See articles by Olson and by Pietrzyk in Transportation Research Board Special Report 242, Vol. 2, 1994.
                                                                   U. S. Environmental Protection Agency

-------
                                                         Technical Methods for Analyzing Pricing Measures
                                                                    to, Reduce transportation Emissions

               •  Budget support;

               •  Audit capability;            ,              .  .     '                      '      •

               •  Accounting capability;

               •  Ability to process revenues;

               •  Contracting capability; and

             -  •  Leasing and procurement capability.4


     4.2.4  Technology Availability


     Some measures, such  as  congestion  pricing, require technology  deployment.  In these  cases,
     sponsors should do the following in their submission:

               •  Document access to the required technology, either available currently off-the-shelf
                  or available in the future in time for scheduled implementation;

               •  Evaluate the technologies to be deployed  to ensure that they (1) are capable of
                  achieving the objectives established  for each measure, and (2) operate effectively
                  and efficiently;

               •  Summarize the results of technical analyses and assessments that have been carried
                  but for the proposed measures; and

               •  Note constraints, areas of uncertainty, and areas needing further development in the
                  deployment of new technologies.                         •     •       /   .      '

     Assessments for congestion pricing, for example, would cover the following technology areas: in-
    . vehicle  units, vehicle to roadside communication  systems, vehicle detection  and classification
     systems, and payment and accounting systems.


     4.2.5  Revenue Generation and Reinvestment


     In a number of cases, pricing measures generate new revenue. This revenue  could be used in two
     basic ways: (1) to reduce other fees or taxes; or (2) to invest in public programs, particularly in
     transportation. In the case, of fee or tax offset, sponsors do not need to detail revenue use.

     In general, sponsors should describe:

               •  Anticipated programs for which new revenues will be used;

               • 'Anticipated impacts of those inyestments on travel behavior and emissions;
               •  Analysis  used to estimate travel and  emissions impacts, including consideration of
                  synergism and overlaps;
     i                                 •               i
      See Olson in Transportation Research Board Special Report 242 Vol. 2, 1994.
U. S. Environmental Protection Agency

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
               *   Act'ons anc* timetables anticipated in implementing the reinvestment strategy and
                   programs; and

               •   Anticipated difficulties or uncertainties in executing the reinvestment strategy and
                   how they will be overcome.                             .

      If revenues are to be reinvested in any way that affects transportation demand  or emissions,
      sponsors should detail demand and the final emissions impacts. Reinvestments with such impacts
      include:            '

               •   Investment in transit facilities: generates shifts in mode share and  reductions in
                   Vehicle trips and VMT;
                   '":           ,    ••' '•,•'<•         '      '   • ' i
               •   Investment in traffic flow improvements or preferential facilities for high-occupancy
                   vehicles: generate increases in travel speed and shifts in mode share;

               •   Revenue used to support transit fares:  produce shifts in mode share, reductions in
                   Vehicle trips and VMT;

               •   Investment in vehicle buy-back programs: induce changes in fleet mix and emission
                   characteristics;

               •   Financial  support of local  government  and business  trip reduction programs:
                   support reductions in vehicle trips and VMT; and                     -

               •   Enhancement of pedestrian  and bicycle environment, or non-motorized access to
                   public transportation: support reductions in vehicle trips and VMT.
                   »                         •  '              • •     .  •    j             .
      Many market-based measures gain effectiveness when supported by  non-market measures, for
      example, expanding the diversity of transportation choices through increased transit options or
      enhanced pedestrian environment.  Land use  strategies may likewise  be  supportive. These are
      mentioned here because some may be potential recipients of program-generated revenue.

      Emissions effects can be established by iterating through the modeling process described  in earlier
      sections, adjusting the processes and values for effects of the reinvestment, where they can be
      determined or estimated. Alternatively, effects can be estimated independently.

      If planned reinvestment — whether in fixed capital, or operations and maintenance — is expected
      to generate these  types  of  impacts at a significant level, additional iterations  in the modeling
      analysis may be required to capture the full effects and claim full emissions credit.
'*""                                                                 U. S. Environmental Protection Agency

-------
                                                        Technical Methods for Analyzing Pricing Measures
                                                                   to Reduce Transportation Emissions
 Appendix A
         Modeling Pricing Measures in a Traditional Modeling Environment
                                 A Case Study Approach
      A.1      Introduction

      Almost all existing travel models were developed not for policy analysis, but for planning major
      transportation facility improvements:  One of the unintended consequences of this is that most
      models are not well suited for analyzing the full range of travel impacts  of different transport
      pricing strategies.  Although these models are generally not structured for assessing the impacts of
      such public and private policy actions, they still  represent the best available data base for travel
      forecasting within  most urban areas. Thus, there is a strong need either to modify existing models
      or develop new models that can meet a wider range of analysis requirements.

      It was not the goal of this research effort to develop and calibrate a new model set, because that
      would have required time and data resources that were not available. Instead, this project took the
      approach that many good models exist in different urban areas and that it should be possible to
      create a model set that could serve the study's purposes by " borrowing". appropriate parts from
      various models and assembling them into one model set that would exhibit reasonable sensitivities
      towards pricing measures. A side benefit of this approach is that the resulting model would not be
      biased towards any one urban area, but would represent a composite picture of a variety of areas.

      The model would,  Jiowever, b& applied and tested using actual data from one metropolitan area, in
      order to produce  more realistic results than  an  analysis using a hypothetical urban area. The
   .   resulting model set is referred to as the Case Study Model, because it,was used to analyze the
      different market-based case studies for this project.  This model is described further in the next
      section.

      It should be noted  that the Case Study Model is not intended to be one that can be picked up and
      applied "as is" to any particular area for  use in analyzing market-based  TCMs.   It was only
      intended for use in this national-level overview analysis of the potential emission  reductions of
      market mechanism;?. However, the model documentation is presented here as part of the technical
      guidance for urban transportationTjlanning staffs who may seek to adapt their local models to better
      analyze pricing strategies. As such, the following is intended to assist such  staffs to develop and
      use similar modeling advances so as to conduct more credible forecasts of the effects of market
      mechanisms in their areas.                                     ',

      A.1.1    Case Study Model Specification

      As  noted above, the Case Study Model set is adapted from various four-step travel forecasting
      models from urban areas across the U.S. Despite some of the problems associated with the four-'
      step process, it is still the most widely used and readily understood modeling approach. It was an
      explicit premise of this project that most of the problems with the four-step process  are related to

•••^	,	-_ •• '        *	;	
 U. S. Environmental Protection Agency • '   ',                   ,                                    ^_j

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions             ,          '            '

      the way in which the steps are applied and other deficiencies in the input data, rather than flaws
      that are inherent in the process itself. The project's researchers hypothesized that an enhanced
      model  set could be developed that would satisfy the need for specific  sensitivity to pricing
      measures within the context of the four-step process.

      Figure  A-l  is a flowchart depicting the  overall structure of the Case Study Model  set.  The
      following sections describe how each component of the model was crafted., with emphasis on the
      nature of the sensitivity to pricing.  It bears emphasis that although each component of the Case
      Study Model set might not be considered the most advanced " state of the art" technique, the way
      they are combined, as shown in Figure A-l, is  unique and is the most important feature of the
      model set.

      The Case Study Model is applied to data representing 1996 conditions in the Washington, D.C.
      metropolitan area. This city and scenario were selected because the data were readily available,
      courtesy of the Metropolitan Washington Council of Governments (MWCOG).  However, neither
      the Case Study Model nor any results of this analysis are endorsed or supported by MWCOG.  It
      also bears emphasis that this model has not been fully calibrated to any set of data and cannot be
      said to  represent the specific sensitivity of Washington area travelers (or any other travelers) to
      pricing.  Its purpose is to demonstrate the structure of a pricing-sensitive model and to produce
      results that are reasonably illustrative of the range of emission reductions that might be associated
      with the assumed levels of transport pricing changes. Neither EPA nor the authors warrant that the
      results estimated by this model would be achieved in actual practice.

      A.1.2    Submodels

      The Case Study Model set includes two important submodels: households by vehicles and size, and
      parking cost.

      Households by Vehicles and Size  Submodel: Most existing trip  generation and  distribution
      models are handicapped  in many cases by working  with zonal average data.  One of the most
      substantial modeling improvements in recent years has been the. increasingly  popular practice of
      subdividing certain items of zonal data by those characteristics which influence travel.  A common
      practice is to split households jointly by size  and vehicles available (or household income).
      Washington, D.C., Detroit, Philadelphia, San Francisco, and Portland (Ore.) are examples of urban
      areas which use vehicles, while New Orleans,  Atlanta,  Northern New Jersey, Dallas, Denver,
      Phoenix, and Minneapolis-St. Paul use income.  This produces  " cells" of households which are
      more homogeneous in nature, reducing the variance in trip rates. Using income or vehicles as one
      of the dimensions also allows the explicit identification of trips by some measure of wealth, which
      can be useful for modeling the differential effect of pricing on the different groups.

      The submodel operates by using Census data to establish  the relationship between  zonal averages
      and the distribution of households by integer values.  The size submodel splits  households by
      number of persons: 1, 2, 3, 4, and  5+, and the vehicle  submodel splits households by number of
      vehicles: 0,  1, 2+.  For the Case Study Model, data on  households by vehicles  available were
      provided by MWCOG, and so' only a size submodel was needed.  Figure A-2 presents the size
      submodel.
A-2                                             ,            '      (7. S. Environmental Protectipn Agency

-------
                                                         Technical Methods for Analyzing Pricing Measures
                                                          ,          to Reduce Transportation Emissions

     Once the individual distributions of households by size and by vehicles are known for a zone, the
     region's joint distribution of households  by size and  vehicles is used  to  estimate the joint
     distribution for each zone.  The size and vehicle submodels and the joint distribution submodel
     used  in the Case Study Model were developed from a 1987-88 home  interview survey in the
     Washington, D.C. metropolitan region. The regional joint distribution is shown in Table A-l.  This
     distribution is used as a "seed" for each zone and then modified to match the desired distributions
     of households by size and by vehicles for that zone.

     Parking Cost  Submodel: The  Case Study Model's parking costs were estimated  using the
     MWCOG parking cost  model.  This model estimates average daily parking cost in  a zone as a
     function of the number of home-based work person trip attractions per square mile in that zone. A
     threshold value is included, so that no parking cost is estimated for most zones.' The parking cost
     submodel  is implemented as a look-up table, shown graphically in Figure A-3.  The values in this
     chart  represent the average cost of parking between those who pay something and those who pay
     nothing and are expressed in cents per day,  in 1980  year dollars.  (Throughout  the Case Study
     Model set, all costs are expressed in 1980 dollars.)  The  average parking cost value estimated by
     this submodel can be overridden by the user for any zone.
U. S. Environmental Protection Agency
                                                                                             A-3

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
                                         Figure A-1
                   Case Study Travel Forecasting Model Flowchart
      highway network
transit times
 and feres,
parking cost
      households and
        employment
      Trip generation
        by purpose
                          Build peak and off-peak highway toll
                                  and free path skims
                        Apply toll path choice model to calculate
                             weighted peak highway skims
 Calculate peak and
 off-peak composite
     impedance
                               . Trip distribution by
                                                                        vehicle
                                                                    ownership data
 transit times
  and fares,
 parking cost
         toll/free
      highway skims
Mode choice by purpose
                                         I
                           Toll path choice model by purpose
                                                                         vehicle
                                                                     ownership data
      Misc. PPAQ input files:
          Traffic patterns
       District equivalencies
               etc.
                              Combine, balance, and assign vehicle trips
                                                \
                             Calculate final VMT, speeds,
                                    and emissions
                               Report of daily mobile source emissions
                                 Spreadsheets for
                               estimating fee impacts
                                 on vehicle age mix
A-4
                                                                    U. S. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                                                                      to Reduce Transportation Emissions
                                           Figure A-2
                                   Household Size Submodel
                          Proportion of Households by Size vs. Average Persons/Household
          0.00%
              1.05
                      1.55      .2.05      2.55      3.05      3.55
                                      Average Persons per Household
                                          4.05
                                                  4.55
                                                          5.05
Size
1
2
3 ,
4
5+
Total
                                          Table A-l
        Regional Joint Percentage Distribution of Households by Size and Vehicles
Vehicles
0
5.7%
33.0%
Source: 1987/88 MWCOG Home Interview Survey.
                   2+
4.2%
1.1%-
0.3%
0.1%
0.0%
18.6%
9.7%
2.6%
1.5%
0.6%
. 2.1%
' 24.7%
14.9%
• - 13.4%
6;2%
                                                          61.3%
 Total
 24.9%
 35.5%
 17.8%
 15.0%
 6.8%
. 100.0%
U. S. Environmental Protection Agency
                                                                                               A-5

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions

      A.1.3    Trip Generation Specifications

      This component  is the most traditional part of the model set.   Daily home-based person trips
      produced by households are estimated as a function of the joint number of households by size and
      vehicles ownership. Daily person  trips attracted to a zone are estimated as a linear function  of
      households  and employment by type.  Four trip purposes are used: home-based work (HBW),
      home-based other (HBO),  non-home-based (NHB), and truck. NHB and Truck vehicle trips are
      calculated as trip  ends  only; i.e., origins = destinations = trip ends. The trip rates include all trips
      and the split of trips that begin or end outside the region (external) are estimated as a percentage  of
      total trip ends.   External  trips  are then split off as a separate  trip purpose.  Internal/external
      productions are distributed  to external stations in proportion to the 1990 traffic count at the external
      station.                                                   ,

                                          Figure A-3
                                   Parking Cost Submodel
                                         Parking Cost Submodel
         350
                   so
                         100     150  .   200    250     3QO     350     400
                             Home-Based Work Trip Attractions per Square Mile (Thousands)
450
       500
               Source: MWCOG.            .      •

     Work trip ends are balanced to the attraction total and trip ends for the other purposes are balanced
     to production totals.  The trip rates and attraction equations were derived  from the models of
     Washington, D.C., Dallas, and Minneapolis-St. Paul.   The trip generation model is presented in
     Tables A-2 to A-4.  This is a very simplistic model, not so much calibrated as synthesized, whose
     main focus was on matching the Washington, D.C. region's total number of trips by purpose.  This
     trip generation model is not sensitive to pricing.
A-6
                                                                    U. S. Environmental Protection Agency

-------
                                                     Technical Methods for Analyzing Pricing Measures
                                                               to Reduce Transportation Emissions
                                        Table A-2
                                 Trip Production Model
HBWtrips/HH
Size
1
2
3' •'• ' ' '
4
5+
HBO trips/HH
1
2
3
4 ' • •• ~
5+ ' .
0
0.372
1.105
1.912
1.912
3.038 ,
Vehicles
1
• 1.147
1.540
2.347
: . . ' • 2.347
3.187
2+
1.583
2.581
3.367
3.367
3.951
-.- . .
3.074
4.437
4.930
6.293
8.207

, 3.190
5.220
8.120
12.180
15.950 ,

3.335
5.452
8.497
12.644
19.372

• '
HBW Attractions =
HBO Attractions =

NHB Trip Ends =

Truck Trip Ends =
                   TabIeA-3
             Trip Attraction Model

1.484 * total employment
0.45 * office employment + 13.50 * retail employment + 3,00 * other
employment + 0.75 * households
0.555 * office employment + 2.59 * retail employment +1.11'* other
employment + 0.'3 7 * households                                -
0.011 * office employment + 0.212 * retail employment + 0.212 * other
employment + 0.042 * households
                                      Table A-4
                              External Trip Percentages
                                   HBW
                                   HBO,
                                   NHB'
                                   Truck
                           5.9%
                           1.5%
                           2.0%
                           5.9%
U. S. Environmental Protection Agency
                                                                                       A-7

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
      A.1.4    Trip Distribution Specifications

      In most areas, the distribution  of trips from an origin zone to  potential  destination zones is
      performed using a gravity model. This model distributes trips as a function of the number of trip
      attractions and a measure of the separation of the zones.  Almost all areas use highway time as this
      measure of separation.  However, researchers have long known that factors other than highway
      time play a role  in the allocation of trips to destination zones.  For example, there is considerable
      evidence that the presence of good transit service between two zones will increase the number of
      person trips between those zones. A logical extension of this concept is that other components of
      this separation, such as prices that users pay for transportation, should also  influence destination
      choice and thus be accounted for. This effect appears to have an intuitive and empirical basis that
      cannot be ignored.

      So far, very few urban areas have developed distribution models that are sensitive to  the prices
      travelers pay and service  levels  of all travel modes (also referred to as composite impedance).
      Some examples include San Francisco, Boston, New Orleans, Atlanta, and Denver. Most areas that
      use this formulation use  it only  for work trips,  but in theory it should be applicable to all trip
      purposes. This has become a well documented process and was adopted for the Case Study Model.

      The Case Study Model distributes trips for all purposes with a gravity model that uses composite
      impedance (CI) as its measure of zonal separation.  This impedance is defined as the log sum from
      the mode choice model, i.e., the natural logarithm of the inverse of the sum  of the exponentiated
      disutilities of all available travel modes (the denominator  of the mode choice equation).   This
      method was adopted from the New Orleans regional model.  This version of composite impedance
      includes the following travel costs: auto operating cost, tolls, transit fares, and parking fees.

      The use of this function makes the allocation of trips to destination zones sensitive to differences in
      those user costs.  For example, if transit fares were to decrease in a certain corridor, not only would
      the transit share increase for those trips (from the mode choice model), but the number of person
      trips in that corridor would increase slightly, because the decrease in fare causes a decrease in the
      separation between zones in that corridor. Further, using information from the mode choice model
      addresses                 the                 need                 for                 model
A-8
                                                                ,.  U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                                     to Reduce Transportation Emissions

      connectivity, even though the models are still applied sequentially. Equation A-l presents the log
      sum equation as used in this model.
                                      1.0
        where:
        Cly   =
        U(tr)  =
        U(da) =
        U(cp) =
                                   e~U(da) + e~U(cp}
composite impedance from zone i to zone j
disutility of Transit mode
disutility of Drive Alone mode
disutility of Carpool mode
                                                                          (A-l)
     The disutility equations are presented in the discussion of the mode choice model, below. Peak
     (congested) travel times and user costs were used for HBW trips; off-peak times and user costs
     were used for the other purposes.  A separate CI value, was calculated for each of three vehicle
     ownership groups and three transit access areas (walk to transit, drive to transit, no transit access).
     A weighted average of these nine values was then calculated. The value of 2 in equation A-l is to
     ensure that the resulting value is greater than zero and the value of 25 is to spread the CI values
     over the range of 1 to 254.                                  '
                                                                               -''"'/
     In the New Orleans model, composite impedance is used to distribute work trips, but not non-work
     trips.  In the  calibration of that model,  it was found that travel prices  and transit service were
     uncertain influences on non-work destination choice.  However, it seems intuitive that travel cost to
     users should  affect non-work trip patterns, in some way  and so  the Case  Study  Model  uses
     composite impedance to distribute home-based other, non-home-based, truck, and external trips as
     well.                                                                                      .

     Special friction factors (" F-factors" ) were developed for use with the composite impedance values.
     These were estimated such that the resulting trip tables would have the same average trip length
     and  average CI as the "known"  trip tables from the Washington, D.C. area.  The F-factor curves
     are shown in Figure A-4.  In order to  adhere to MINUTP requirements, the F-factors are scaled to
     be integers between the values of 1 and 999,999.       .

     Various trip distribution calibration statistics are shown in Table A-5. The "observed" data are
     based on  1996 trip  tables previously  estimated by MWCOG.   These results suggest that the  CI-
     based gravity models reproduce known trip patterns fairly well.         "'.'..
U. S. Environmental Protection Agency
                                                                                              A-9

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
                                         Figure A-4
                                      F-Factor Curves
                                            F-Factors
           1,000,000,
                             so
                                        100          150
                                        ^ •
                                        Composite .Impedance
                                  200
               Purpose
                                           Table A-5
                                  Trip Distribution Statistics
Avg. Composite Imped.
'Observed"    Estimated
Intrazonal Trips
 Observed"
HBW
HBO
NHB
Truck
I/X75.24
	 ; 	 r— • 	 ; 	
68.10
45.58
45.81
61.88
75.29
67.90
47.52
47.04
61.84
	 '
1 16,300
538,500
314,100
10,000
	 	
133,300
581,500
269,800
11,200

     The resulting trip tables were found to sharply overestimate travel across the Potomac River and so
     K factors were developed to force a better correspondence with the traffic counts on the bridge
     crossings. HBW trips across the river were factored by 0.20, while all other purposes were factored
     by 0.25. The need for such factors is supported by the existing MWCOG model, which uses extra
     time penalties on the Potomac River bridges to accomplish essentially the same function.

     A.1.5    Mode Choice Specifications
                                                                                  *
     Within the four-step process, current mode choice models stand out as being the most rigorously
     developed  and properly  price-sensitive component.  Many larger urban areas  have developed
     sophisticated logit  models which estimate  the share  of person trips  by mode,  based on  the
     socioeconomic level of the traveler and the time and  user cost attributes of the various modes.  The
     Case Study Model mode choice model is based on the approach used in the Washington, D.C. area,
     which is considered fairly representative of good practice.
A-10
                                                                  U. S. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
            •                                           •               to Reduce Transportation Emissions
                                       •' ' '  i •         •                     '
      The HBW mode choice model incorporates a Carpbol model which estimates the split among 2-
      person carpools,  3-person carpbols, and 4+-persoh carpools.   Those  percentages are used to
      estimate the "average" attributes of the, Carpool  mode, allowing the model to then split total
      person trips into Drive Alone, Carpool, and Transit modes. Separate walk-access and drive-access
      "markets" are  used to calculate the Transit split.  This split is sensitive to various auto mode
      attributes, including terminal time, driving time, auto operating cost, tolls, and parking cost, as well
      as transit attributes, including walk time, initial wait time, transfer time, in-vehicle time, and transit
      fare.   A special high-occupancy vehicle (HOV) feature allows the modeler to  define HOVs as
      having either 2, 3, or 4+ persons per vehicle and uses special travel times and costs for such trips.
      The Washington model's sensitivity coefficients were replaced with those representing an average
      of experience from logit models in numerous urban areas around.the. country (which-turned out to
      be quite similar to the Washington area's coefficients).
     Equation A-2 presents the logit mode choice model structure.
        p    =
        where:
                        >-U(ma)
e-U(tra)+e-U(daa)
                               +e
                                  ~U(cpa)
                                                                                      (A-2)
       U(daa).
       U(cPa)
          - probability of choosing mode m for a trip from zone i to zone j by '  •
            travelers with auto availability a
          = disutility of Transit mode for travelers with auto availability a
          = disutility of Drive Alone mode for travelers with auto availability a
          = disutility of Carpool mode for travelers with auto availability a
          = disutility of mode m (Transit, Drive Alone, or Carpool) for travelers with
          availability a                          *
                                                                                          auto
     The disutility of the Carpool mode is a weighted average of the disutility of each integer Carpool
     occupancy mode:
U(cp) = P2*U(2) + P3*U(3) + P4*U(4)                  ,

where:                                           •
U(2), U(3), U(4)   =    disutility of 2-person, 3-person, and 4+-person carpools
P2> ?3 = P4        =    probability of the trip being made in a 2-person, 3-person, or 4+-
                       person carpool, given that the traveler is in a carpool, calculated
                  as:
                                                                                             (A-3)
•'ijoa
                                                                                             (A-4)
U. S. Environmental Protection Agency
                                                                                               A-ll

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions

        where:
        Pijoa =  probability of the trip being made in a vehicle with occupancy o for a trip from zone i to
                 zone j by carpoolers with auto availability a
        U(Oa) =  disutility of carpool occupancy p for travelers with auto availability a

      This is considered a "quasi-nested" logit structure in that the carpool model (equation A-4) is not
      used to directly calculate the carpool share, but rather to  calculate the average carpool disutility
      (equation A-3), which is then used in the primary model (equation A-2). -Table A-6 presents the
      disutility equations for the HBW primary model and the carpool model.

      For home-based other and non-home-based  trips, the MWCOG model does not  use the logit
      equation to calculate a mode share.  Instead, it multiplies the estimated HBW mode shares by a
      factor that varies with distance (and, for HBO trips, by vehicle ownership). A similar approach was
      used in the Case Study Model, except that the non-work mode shares  are  based on off-peak
      highway and transit times.  The factors are shown in Table A-7.

      A very important point to note about the cost coefficients is that they were originally calibrated
      using prices expressed in 1980 year dollars.  Thus, when this model is applied, all prices must be
      adjusted to represent the equivalent price in 1980 dollars. For example, prices in 1993 dollars must
      be deflated to 1980.  This is done by multiplying them by a deflation factor of 0.57, which is the
      ratio of the U.S. Urban  Consumer Price Indices of 1993 to 1980 (= 82.4/144.4).  Neglecting to
      account for this effect will result in the model being improperly sensitive to price.

      This model was not calibrated, per se,  but the bias coefficients were  adjusted so as to achieve
      approximately the  same transit trips and auto  occupancy that  had been forecasted  for the
      Washington, D.C. region for 1996. Those statistics are shown in Table A-8,

      Prices were carefully specified in this model.  All  prices were expressed in  cents per one-way trip,
      in 1980 year dollars, because that is the year for which the original coefficients were, developed!
      Auto operating cost was calculated as 11.0 cents/mile, multiplied by the minimum path distance for
      each O/D pair.  This is the incremental cost of using a vehicle for each trip and does not include
      any  of the fixed costs of vehicle ownership. Parking cost to users was calculated using the density-
      based model shown  above in Figure A-3 and divided by two to represent half the daily price. For
      non-work trips, the average parking cost to users is assumed to be 20% of the daily parking price.
      The toll (roadway price) input represents the weighted average toil for each O/D.
A-I2
                                                                   U. S. Environmental Protection Agency

-------
                                                    Technical Methods for Analyzing Pricing Measures
                                                               to Reduce Transportation Emissions
                                       table A-6
                  Home-Based Work Mode Choice Disutility Equations

 Prime Mode'Choice

 U(tr) = 0.05 * (WLK + WT1 + WT2) + 0.02 * RUN + 0.004 * FARE + 0.02 * AAC + ACBIAS
 U(da) = 0.1001 * TERM + 0.02* RUN + 0.004 *OPCST + 0.0095 * (PKCST + TOLL) + DABIAS
 U(cp) = 6.0519.* TERM +,0.02•* (RUN+'l.l*(CPOCC-l)) + 0.004 * OPCST/CPOCC + 0 0095 *
         (PKCST + TOLL)/CPOCC + CPBIAS
 CPOCC =
                  1.0
           2.0

Carpool Share
           P(2)   P(3)   P(4)
           	-)	1	
3.0   4.394
U(2) = 0.1823 * TERM + 0.07 * (RUN + 1.1) + 0.014 * OPCST/2 + 0.0333 * (PKCST + TOLL)/2
U(3) = 0.1823 * TERM + 0.07 * (RUN + 2.2) + 0.014 * OPCST/3 +0.0333 * (PKCST + TOLLV3
       3PBIAS                                                   ',,'••
U(4) = 0.1823 * TERM + 0.07 * (RUN + 3.733) + 0.014 * OPCST/4.394 + 0 0333 * (PKCST +
       TOLL)/4.394 + 4PBIAS
Bias Constants by Vehicles.Owned
Vehicles      	     0
                1
 2+
ACBIAS
DABIAS
CPBIAS
3PBIAS
4PBIAS
       2.1333  0,9360
       3.3077-0.0439
       2.9882  1.0989
       •1.2951  1.7186
       2.1106  2.5532
 0.9501 (if auto used to access transit system)
 -1.0552,
 0.4200 ,              •   .  !   '  '•  ••
'•1.6952 ..'         •.-.-.•
 2.1909
Abbreviations:
WLK = transit walk time
WT1  =? transit initial wait time
WT2  = transit transfer time

RUN  = transit in-vehicle time
FARE== transit fare
AAC  = auto access time to transit
                         TERM   = highway terminal time
                         OPCST  = auto operating costtp the user
                         PKCST  = daily parking cost to the user (divided by

                         TOLL    = average toll paid
                         CPOCC  = average occupancy of Carpool mode
All times are in minutes and costs are in cents, expressed in 1980 year dollars.
U. S. Environmental Protection Agency
                                                                                    A-13

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
                                            Table A-7
                                 Non-Work Mode Share Factors


Distance (mi.)
0-1


1-2


2-3


3-4


4-5


5-6


6-7"


7-10


10-30


30-50



Vehicles
Owned
0
V
2+
0
1
2+
0
1
2+
0
1
2+
0
1
2+
0
1
2+
0
1
2+
0
1
2+
0
1
2+
0
1
2+
HBO
Transit
Factor
0.5261
0.2374
Q.2374
0.6428
0.1807 -
0.1807
0-6428
0.1807
0.1807
0.6428
0.1807
0.1807
0.6428
0.1807
0.1807
0.6236
0.1607
0.1607
0.6236
OJ607
0.1607
0.6236
0.1607
.0.1607
" 0.5156
O.'l514
0.1514
0.0
0.0
0.0
HBO
Auto Pass.
Factor
2.1790
1-5515 .
2.7676
2.1790
1.5515
2.7676
3.7090
2.0419
3.3290
2.9104
2.0419
3.3290
2.1036
2.0419
3.3290
2.0978
1.9051
3.3290
1.8674 .
1.9051
3.3290
1.7769
1.9051
3.3290
1.5292
1.8150
3.3290
1.5.057
1.7791
3.3290
Table A-8 Summary of Mode Choice
-
Transit Trips*
Percent Transit
Vehicle Trips




Auto Occupancy
* Linked trips.

HBW
675,900
15.4%
3,309,700
1.125

HBO
460,300
4.5%
7,513,700
1.314

NHB
Transit
Factor
0.1334
0.1334
0.1334
0.1334
0.1334 •
0.1334
0.1334
0.1334
0,1334
0.1334
0.1334
0.1334
0.1334
0.1334
0.1334
0-1334
0.1334
0.1334
0.1334
0.1334
0.1334
0.1334
0.1334
0-1334
0.1274
0.1274
' 0.1274
0.0
0.0
0.0
Results
NHB
110,300
2.8% ]
3,052,600
1.239
•,
NHB
Auto Pass.
Factor
2.0081
2.0081
2.0081
2.0081
2.0081
2.0081
2.0081
2.0081
2.0081
2.0081
2.0081
2.0081
2.0081
2.0081
2.0081
2.0081
2.0081
2.0081
2.0081
2.0081
2.0081
2,0081
2.0081
2.0081
2.0081
2.0081
2.0081
2.0081 '
2.0081
2.0081

Total
1,246,500
6.7%
13,876,000
1.252

A-14
                                                                     U. S. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                                                                      to Reduce Transportation Emissions

      pair, which is the skimmed toll value multiplied by the percent of drivers which are estimated to
      use the toll path for that trip.

      The'mode choice model outputs vehicle trips by LOV and HOV occupancy categories. The HOV
      occupancy criterion  can be 2, 3, or 4 persons/vehicle; in this project it was set  to  3.   The
      Washington area does have HOV highway facilities and their impact was included in the analysis.

      A;L6     Path Choice and Traffic Assignment Specifications

      Studies  of toll  roads focus  on  drivers' trade-off  between  paying  a toll and saving  time.
      Traditionally, relatively less attention has been paid to drivers'  path choices,.as planners have
      relied mainly on traffic assignment software to handle that task. However, recent toll road studies
      have discovered that such software is inadequate for modeling complex toll vs. time trade-offs and
      have developed more sophisticated models of path choice. These models determine the best "free"
      path for each zone-to-zone pair, i.e., the best path that does not use the toll facility.  They then
      determine the best "toll" path, which is the best path that includes the toll facility. Those paths are
      analyzed to determine their time and toll difference, which is then used in a logit model to estimate
      the  split of trips between the two paths.  This  is done for every zone-zone pair in the network.
      Separate toll  and time sensitivities are  used  for work and non-work trips.   Recent advances  in
      assignment software permit the two resulting trip tables to be assigned simultaneously, but each  to
      its own set of paths.  Within the multiple iterations of assignment, trips are allowed to "migrate"
      between paths to a limited degree in response to congestion.  The result is  a more  realistic
      assignment of trips to toll facilities, in a manner that is sensitive to the level of toll as well as to the
      capacity of the alternative non-toll routes.                                         "

      This kind of process has been recently used in toll road studies in Denver and New Jersey, and was
      adapted for use in  the Case  Study Model, under the assumption that it is suitable for analyzing
      roadway pricing measures.  The resulting toll, values affect not  only the path choice, but also the
      choice of mode, which uses toll as an input, as noted above.  Because toll is part of the composite
      impedance calculation, toll values affect the distribution of trips as well.

      The toll diversion model is a logit formulation that splits all vehicle trips for each O/D pair into
      "toll" and "free" trips. The equation is as follows:
        P(toll), = •
                      1.0
                   xAU(toII-free)
(A-5)
       where:,
       P(toll)jj=        probability of choosing the toll path for a trip from zone i to zone j   -
       AU(toll-free)  = difference in disutility between toll and free path
                     = 0.301 * (toll time - free time) + 0.0127 * tollpath toll + 1.5 (HBW.trips)
                     = 0.191* (toll time - free time) + 0.0272 * toll path toll + 0.3 (all other trips) toll
                        is in cents and time is in minutes            ,     •                      ,   .

     The toll time is computed based on paths that are built by temporarily reducing the time on all
     tolled links by 10%.  This encourages any trip which is near the toll link to use it. The free path
     time is computed  by  prohibiting the use of all tolled links.  Throughout the Case Study Model,
U. S. Environmental Protection Agency
                                                                                              A-15

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions

     highway paths for skimming and for assignment are built on the basis of minimizing impedance,
     where impedance is calculated in equivalent cents as:

               impedance (in cents) = 10.0 * time (in minutes) + 11.0 * distance (in miles)

     The 10.0 factor on time is derived from an assumed effective value of travel time of $6.00 per hour
     (roughly 22% of the Washington, D.C. area's mean 1990 household income).  The  11.0 factor on
     distance represents the user cost of operating an automobile, as noted above. This formulation
     ensures that user cost influences highway paths.

     The traffic  assignment procedure  uses  four  iterations  of incremental.,  capacity-restrained
     assignment, with 25% of the trips assigned on each iteration, following current MWCOG practice.
     Thus, the assignment of trips is sensitive to roadway capacity in an incremental fashion: some trips
     "see" an open roadway, while others see a congested one. The input daily vehicle trips are split by
     four categories: low-occupancy vehicle (LOV) free path, LOV toll path, HOV free path, and HOV
     toll path, with each category  of trips being assigned to its specific set of paths, respecting the
     presence of priced roadways and HOV roadways. The output of this process is a loaded network
     with daily traffic volumes on each link.

     A.I.7    Emissions Calculation Specifications

     The estimation  of mobile  source  emissions requires  two basic data  items from  the  traffic
     assignment: VMT and speed.   U.S.  EPA's MOBILESa emission factor program is  applied to
     calculate emission rates in grams/mile for the: criteria pollutants (the EMFAC7F program is used in
     California). These rates are a  function of the mix of vehicles by eight types, the average distance
     they travel  per year, average travel speeds, ambient temperatures, inspection and maintenance
     (I/M) programs, fuel policies, and other factors.  The PPAQ (Post-Processor for Air Quality)
     program developed by Garmen Associates  is used to read the  loaded network, revise the speed
     calculations by facility type and time period, summarize VMT by facility type and time period, and
     apply the MOBILESa emission factors. PPAQ requires a series of input tables that reflect the mix
     of vehicle  types by roadway type,  the percentage of traffic by hour, and other parameters that
     describe traffic patterns in more detail.  These parameters have been adopted from  work recently
     performed in the Philadelphia region. The result of a PPAQ run is an estimate of total daily tons of
     HC, CO, and NOX from on-road mobile sources.

     Changes in  most of the pricing measures under study, such as roadway pricing, transit fares, and
     parking costs, will be  reflected  in the assigned link  volumes.  The exceptions are the measures
     involving registration  fees that  are based  on age or emission  level.   It is  assumed that such
     strategies have no measurable impact on the amount of VMT, but will affect the mix of vehicles by
     age. Strategies that make it more expensive to own an older vehicle should result in fewer older
     vehicles on the road. Since older vehicles were generally built to less stringent emission standards
     and are usually less well maintained,  a reduction in such vehicles can be expected to  reduce the
     emission rates calculated by MOBILESa.
                     .        ,J   "*            *          .               !       '      •         \
     Spreadsheets have been developed in this study to analyze age-based fees and emission-based fees.
     These spreadsheets  estimate the impact of different fee  structures on the default MOBILESa
     vehicle mix by year.  The spreadsheets use a very simplistic relationship between the annual user
     cost of owning  a vehicle and the percentage of vehicles in each age category.   For each age

A-16                                ,                              U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                                      to Reduce Transportation Emissions

      category, a,value is computed representing the annual fee divided by the estimated market value of
      the vehicle. This value is the change in fee as a percent  of the vehicle's value. An elasticity of -
      1.0, derived from literature on the price elasticity of new'car sales, is applied to that change in fee
      The resulting value is the estimated percentage change in fleet share for that age category - This is
      done separately for each of the eight MOBILESa vehicle types and MOBILESa defaults are used
      for the initial age mix. It is assumed that none of the registration fee policies would affect the total
      number of vehicles; only the age mix. Thus, all of the vehicles removed from older age categories
      are shifted to newer age categories, in their existing proportions.

      The output of these spreadsheets is a revised.set of vehicle age mixes by vehicle type that can be
      input directly into MOBILESa.  Table A-9 shows an example spreadsheet, analyzing a fee schedule
      for light-duty gasoline vehicles (passenger cars) that varies from $100 for 4 year old vehicles to
      $900 for vehicles of 20 years or more.
                                         .                                            '       '•  r   '
      The revised age  mix is entered into PPAQ, along with the loaded highway network from the
      assignment step.  PPAQ splits the link volumes by time  period, adjusts the speeds  on each link
      summarizes VMT and average speed  by time period,  facility type, and jurisdiction, applies
      MOBILESa to each "scenario", multiplies the resulting emission rates by the VMT, and produces
      a concise summary of the regional mobile source emissions of interest.  Table A-10 is a sample of
      the summary page of PPAQ output.
U. o. Environmental Protection Agency
                                                                                             A-17

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
Table A-9
Sample Vehicle Age Mix Adjustment Spreadsheet (Age Fee)
ILDGV |

year
25
24
23
22
21
20
19
18
17
16
IS
14
13
12
11
10
9
S
7
6
5
4
3
2
1


estimated
value
$400
$500
$600
5700
$900
51,000
SI ,200
$1,500
51,800
$2,100
$2,500
$3,000
$3,500
54,100
54,900
$5,700
$6,700
57,900
59,100
510,500
512,000
513,500
514,700
514,800
517,922


elasticity:
change in fee
5900
5900
5900
$900
5900
5900
5800
$800
$700
$700
$600
$600
$500
5500
5400
$400
$300
$300
5200
5200
$100
5100
$0
$0
SO


-1.00
fee change
as % of value
225.0%
180.0%
150.6%
128.6%
100.0%
90.0%
66.7%
53.3%
38.9%
33.3%
24.0%
20.0%
14.3%
12.2%
8.2%
7.0%
4.5%
3.8%
2.2%
1.9%
0.8%
0.7%
0.0%
0.0%
0.0%

avg age:
MOBILESa registration mix by year (starting 1 July 90) new/base
base % by yr
1.10%
0.30%
0.40%
0.50%
0.60%
0.80%
1.10%
1.50%
1.40%
1.90%
2.40%,
3.70%
4.70%
5.40%
5.00%
5.10%
5,00%
5.60%
7.70%
8.10%
8.40%
8.20%
'8.30%
,7.90%
4.90%
100.0%
8.2
new % by yr
0.00%
0.00%
0.00%
0.00%
0.01%
0.10%
0.41%
0.77%
0.93%
1.38%
1.98%
3.23%
4.41%
5.23%
5.09%
'5.28%
5.33%
6.03%
8.44%
8.93%
9.39%
9.19%
9.39%
8.94%
5.54%
100.0%
7.1
%left
1.10%
0.30%
0.40%
0.50%
0.59%
0.70%'
0.69%
0.73%
0.47%
0.52%
0.42%
0.47%
0.29%
0.17%
-0.09%
-0.18%
-0.33%
-0.43%
-0.74%
-0.83%
-0.99%
-0.99%
-1.09%
-1.04%
,-0.64%
0.0%

cum % left
1.10%
1.40%
1.80%
2.30%
2.89%
3.58%
4.27%-
5.01%
5.48%
, 6.00%
6.41%
6.88%
7.17%
7.34%
7.26%
7.08%
6,75%
6.32%
5.58%
4.75%
3.76%
2.77%
1.68%
0.64%
0.00%

avg fee:
ratio
0.0%
0.0%
0.0%
0.0%
., 2.4%
13.0%
3.7.1%
51.2%
66.5%
72.6%
82.6%
87.3%
93.9%
96.8%
101.7%
103.5%
106.6%
107.6%
109.6%
110.3%
111.8%
112.1%
113.1%
113.1%
113.1,%

$235
Note: "LDGV" = light-duty gasoline vehicle
A-18
U. S. Environmental Protection Agency

-------
                                                             Technical Methods for Analyzing Pricing Measures
                                                                         to Reduce Transportation Emissions
                                             Table A-10
                                      Sample PPAQ Output
EPA Market-Based Measures Study:   1996 Base
                                                                                         Page 36
• Total Emissions by Scenario: • '. ' . [ ~ ' ~~- — : 	 — 	



Scenario
A 67F
A 67F
A. 67F
A 67F
A 67F
• : A '67F
A 67F
A 67F
• A 67F
• A 67F
A 67F
"A 67F
A 67F
A 67F
A 67F
A- 67F
A 67F
A 67F
A '-67F
A 67F
A 67F
A 67F
A 67F
A .67F
TOTAL -
IT
IT
IT
IT
2T
2T
2T
2T
3T
3T
3T
3T
4T
4T
4T
4T
5T
5T
5T
5T
6T
6T
6T
6T
kg
1
2
3
4
1
2
3
4 ,
1'
2
3
4
1
2
3 '
4
1 '
2
3 .
4
1 •
2
3
4

VMT
(veh-mi)
6959230
9368194
17085802
11197222
1 819904
1285529
. 2004210
• 951494
6361328
9973934
15.549907
7382279
4687831
7350054
•11459147
5440198
1564851
1 2745873
3572584 .-
1958525
2094284
3674875 .
4781289
1 2621150
140889696 ' -
VHT Avg Sp'd -
.(veh-hr)
1 141448
324159. ,-
'411706
1 189783 '
20345 '
76520 .
49856 ,
17818
340178
1049888 .
822746
232147
393935
1080890
962954
256613
• 115915
450143,
. 24'4698
, . 99925
66910
117784
152757
83743
.7702861
(mph)
49.
.28.
• 41.
59.
40.
' 16.
40.
53.
18.
9.
18.
'31'.
11.
6.
11.
-' 21,
13.
, 6.
. 14.
.19.
31.
31.
31.
31.
18.

2,
9
5
0
3 •
8
2
4
7
5
9
8'
9
8
•9
2
5
1
6
6 '
3
2
3
3
3
- tons ,
VOC HC
(kg.)
7289
16466
22108
12909
942
3435
2563
926
12872
42272
35269
9719
13371
'40505..
" 38330
9549
40.71
17039
10154
3616
- 2903
5964
7395
- 3457
323124-
•356.179
, CO
, , (kg) -
49933
107271
1.42505
13i713 •
6434
• , 22543
16282
• ' 6571
99395
273395
• 249739
70717 •
102474 '
'267663
260217
74506
. 31003
110003
' 69201
28669
20651 "
,37679
.48753 •
• ' 25087
• 2252406
2482.827
NOx
• (kg)
13.690 •
1,6541
31823
32602
1404
2085
3452'
2052
10507
17313
• 25847
12602 ;
8178
13550
:19982
8818
2687
5153
6048
3129
3575'
6094 ,
8159 •'
' 4355
259645
286.207
                             HC Emissions  Breakdown:                              .  •  > '
                                             Exhaust     186606  kg  205.696 tons.    57.8%
                                         Evaporative      33070  kg   36.453 tons    10.2%
                                           Refueling           0  kg    0.000 tons      0.0%
                                             Running,     9li80  kg  100.507 tons    28.2%
                                             Resting    .. 12268  kg  '13.523 tons      3.8%
PPAQ/MOBILE5                    '       .            .   .
04-21-1994. /  dO : 5.7:18'           •


Note:  .  A 67 refers to the summary jurisdiction, of which there is only one.
        F x refers to the facility type.
        T y refers to the time period: 1=AM peak, 2=PM peak, 3=midday, 4=night.
U. S. Environmental Protection Agency .

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions

      Post-processor programs such as PPAQ are commonly used in air quality modeling. This is mainly
      because historically, the constrained speeds that result from most traffic assignments are not
      sufficiently accurate.  These speeds represent just one particular value and they tend to include
      other elements of impedance, as needed to obtain accurate assigned volumes.  Many planning
      agencies have developed their own speed processor programs to adjust these constrained speeds to
      be more realistic and  to summarize VMT by various stratifications, in preparation  for running
      MOBILESa.

      A.1.8    Model Application

      The case study model set is applied in  a series of 15  program steps.  The  MINUTP planning
      software system is used for most of these.  Custom FORTRAN programs were written to prepare
      the land use data and to apply the mode choice model.  PPAQ  and MOBILESa are "stand alone"
      programs and the age mix spreadsheets are in Microsoft Excel.  The full model  set requires about 9
      hours to apply using an 80486-based computer running at 66 MHz.

      The case study model set was applied using basic land use and network data from the Washington,
      B.C. area (1,478 zones), representing approximate 1996 forecast conditions.   However, since the
      model incorporates components from various cities, the results do not reflect actual (or forecasted)
      conditions in Washington and cannot be compared to the results from the Washington area's own
      model set

      The Washington  area is projected to comprise about 1.8 million households, 4.0 million persons,
      and 3.0 million jobs. The area has an extensive Interstate system, including a Beltway around the
      District of Columbia and the close-in Maryland and Virginia suburbs (I-95/I-495) .. In addition to
      an extensive bus network, the area is served by the Metrorail system  and four commuter rail lines.
      Major HOV facilities exist in Virginia on the Shirley Highway (1-395 and 1-95  to the south) and I-
      66 to the west.   There is one existing toll road, connecting the Beltway to Dulles International
      Airport to the west.
                 *
      As Figure A-l shows, the model set is applied " backwards":

               •  The path choice model is applied first, to derive weighted average highway time,
                  distance, and toll values.  Peak period values are used for work trips and off-peak
                  values are used for all other purposes.

               • _The mode choice  program is then applied to calculate the composite impedance
                  value by zone-zone pair.
                   ''.,••                      ' .'      .     l        '
               •  Trip generation and distribution are applied to estimate  person trips by purpose.
               •  The mode choice program is applied again, this time to  split person trips by mode.
               •  The path choice model is applied again to split vehicle trips by toll vs. free path.
               •  The vehicle trips are assigned to the highway network.
               •  The age mix spreadsheets are applied to determine changes in the age mix.
               •  PPAQ and MOBILESa are applied to compute emissions from the loaded network.

      The case study model has not been " calibrated" in the true sense, since it was developed from data
      representing several urban areas.  However, the results of the various components were checked for

A-20                              .                               U. £.' Environmental Protection Agency

-------
                                                         Technical Methods for Analyzing Pricing Measures
                                               ,   •      .   -         to Reduce Transportation Emissions

     internal consistency and to ensure an approximate level of correspondence with 1996 total VMT
     estimates for the Washington, D.C, area.

     A.1.9    Summary of Advanced Features of the Case Study Model

     As noted above, the Case Study Model set is hot based on new research and does not represent any
     breakthrough in-the-state of the practice in travel demand modeling. Its advancement is that it was
     created from the best pricing-related features from several other well-documented model sets that
     have been extensively tested throughout the years. It is very likely the first time that these various
     components have been assembled in this exact manner  and  demonstrates one  way in which the
     four-step process can be enhanced to be sensitive to policy issues such as pricing.

     The most noteworthy features of this model set are:

               •   Use of composite impedance for trip distribution for all trip purposes
               •   Integration of a toll/free path choice mode] within a four-step process
               •   Mode choice model, including a nested carpool occupancy model

               •   Parking price sub-model within the mode choice model that permits separate
                  specification 6f,parking prices for LOVs and HOVs

               •   Assignment procedure that simultaneously handles LOV/HOV and toll/free  path
                  trips   -. -    .                 •

              •   Ability to easily calculate effects of vehicle age mix changes
              •   Integration of mobile source emission calculations within a four-step travel model

     A.2      Enhancements to the Case Study Model

     The resource constraints of this project limited the development of the Case Study Model. As this
     work progressed, several shortcomings became apparent. These are described here as guidance for
     enhancing this particular model set, and other models in general.

     Income Stratification: It would be preferable to stratify the entire model set by income level. This
     would  permit the identification  of differential  price sensitivities by  income  level and  would
     facilitate  the examination of  the  differential   effect  of pricing policies by  income  level.
     Unfortunately, time  did not permit the development of such  a model, nor was  the  1990 Census
     Transportation Planning Package (CTPP) available at the time this work was conducted. Now that
     CTPP files have become more widely available, income stratification should be pursued  more
     vigorously. This is important because there is considerable evidence which suggests that people of
     different income levels respond to pricing measures differently, and public policy considerations
     suggest the need to analyze how the implementation of such measures affects each income group.
     More investigation into the combined effects of income and vehicle ownership on traveler  behavior
     should  also be conducted.  Traveler surveys need to collect information on household income (or
     income category, such as quartile). Progress is being made in the difficult area  of estimating the
     stratification of trip attractions by income.

     Vehicle Ownership: There is growing evidence that the effect of vehicle ownership on person trip
     generation cannot be discounted, even after considering the effect of income.  Vehicle availability

/. S. Environmental Protection Agency ,                                                              . _
                                                                                           • ~  '

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
     may well have  a measurable effect on discretionary trips and on the level of trip chaining.
     Considerable attention is now being paid to developing vehicle ownership models that are sensitive
     to income, locational, density, accessibility, and/or transit service variables.

     In order to analyze the  effect of registration fees or one-time purchase surcharges/rebates, these
     models must also be at  least marginally sensitive to the price of buying a vehicle and the annual
     user cost of owning one. In the context of travel modeling, increases in ownership cost should be
     reflected in decreased numbers  of multiple-vehicle  households and increased numbers  of one-
     vehicle households. There might even be an increase in zero-vehicle households, but that is much
     less clear.  At least one  vehicle ownership model sensitive to user cost has been well documented
     [5].              .                                                .....

     Speed Feedback: The Case Study Model now begins with peak and off-peak speed values, the
     peak speeds having been derived from previous MWCOG model runs.  It would be preferable to
     run the entire model set at least one more time in each application, using the speeds  from the first
     run (modified as necessary to more closely match observed data) as peak speeds in the second run.
     This is particularly important for estimating congestion pricing scenarios.  However, in this model,
     two  iterations would require a total of  18  hours (this  is  one of  the  disadvantages of  an
     interconnected model set — changing one thing changes everything and so the entire  model set
     must be iterated to ensure equilibrium). Doing this would have severely limited the ability to meet
     this project's schedule and so was not done.

     In addition, numerous studies have documented the sensitivity of emission rates to average speeds.
     Thus, it is vital to have  reasonably consistent and accurate speeds for emission modeling.  In this
     study, the PPAQ program was used to adjust assignment speeds by facility type and time period.

     Auto-Access to Transit Trips: V.ery few models include such trips in their vehicle trip table.  To
     do so requires access to transit network data (such as park-and-ride [PnR] lot locations) that was
     unavailable for this project,  as well as additional processing steps and time. Still, this phenomenon
     should not be overlooked, because some improvements to transit service can  increase emissions by
     enticing some who carpool  or who walk to transit to switch to driving to transit.  Although most
     drive-access trips are short, they almost always involve a cold engine, and therefore cold start
     emissions.  Increasing the number of such trips might not compensate for the passenger car VMT
     that is removed from the network.
             1                     .            • •                      '
          «        "            .
     Parking Price Measurement: The  cost of parking to the user  is  one of the most  important
     variables in determining mode choice.  Thus, it would be preferable to measure it more carefully,
     such as specifically modeling the proportion  pf travelers who have free parking (instead  of
     accounting for that effect in the average value, as the Case Study Model does). This would require
     some additional data, which  is difficult to obtain, but it would improve the  model's ability to
     respond directly to policies which change the percent of travelers who park for free.  It would also
     be preferable to include variables other than density in the parking cost submodel, to increase its
     accuracy.

     In addition,  if strategies to change  non-work parking pricing are to  be properly modeled,  an
     improved method of estimating non-work parking prices is needed.  This submodel would probably
     be based  on  employment  density and locational variables.   The  cost of all-day parking to
     commuters is relatively  stable and closely related to work trip end density, as suggested above in
A-22            '                                               .    U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                                     'to Reduce Transportation Emissions
      Figure A-3. Of course, non-work travelers have a shorter parking duration time and are even more
      likely to park for free than commuters5 but otherwise little is known about their average parking
      costs.
     Effect of Price on Trip Generation: In this model set, the number of person trips per household is
     completely insensitive to the incremental user cost of travel by any or all modes.  As was noted
     above, that might be remedied by- including some kind of composite accessibility measure at the
     zonal or household level, which would require additional research.  This  would make travel
     estimates sensitive  .to area-wide pricing  measures such as a gasoline tax  increase.  It would also
     address the phenomenon of "induced travel"  that is a concern to some planners by making trip
     rates sensitive to the general level of development in a region and the general quality of highway
     and transit service.                           ,
         *      •               .                             "    •    .     '     •
     With a doubly-constrained trip distribution procedure such as the standard gravity model, it would
     probably prove necessary for the trip attraction estimate to be sensitive to user cost (if the  gravity
     model uses composite impedance).  Otherwise, trip  pattern distortions could  occur under  certain
     pricing scenarios as the gravity model tries to reallocate attractions in response to cost changes, but
     the attraction model would maintain the same attractions in a zone.

     Time of Day: This model set estimates.total daily travel only.  Although separate .peak and off-
     peak impedances are used to represent those periods, the model does not account for the possible
     "migration" of trips from one period to the other.  Such migration might occur due to congestion,
     pricing, or employer policy. ^A basic model that splits daily travel into three or four periods is not
     conceptually  difficult;  such models  are  currently used in several cities.  Further refinements  to
     make that model sensitive to peak/off-peak time savings should be readily feasible.* However,  to
     analyze peak-only pricing or congestion pricing measures* the model also  needs to be sensitive  to
     the peak/off-peak travel, cost  differential  and it is  uncertain whether the data to define such
     sensitivity  exists.  However, it may be necessary to synthesize. such a  sensitivity (or perhaps
     estimate it from stated preference surveys or the experience of transit systems with peak/off-peak
     fares) and re-examine it in the context of an  actual  congestion pricing experiment.  This  is
     important,  because  a likely major response to peak-only pricing measures is  to shift the time  of
     travel.             s                                                               .,

     Trip Chaining:  As in most other travel forecasting models, this model does not directly estimate
     any effects on trip  "chaining".  Many  researchers believe that a logical and likely response  to
     increases in the price of travel (or reductions in the number of vehicles available to a household) is
     for people to use their vehicles more efficiently:   This  could have  some effect  on  VMT and
     potentially a greater impact on cold  starts. It would  be helpful if the model set could reflect this
     phenomenon.       .                       ...                             '
     Nested Logit Model: Although  the multinomial logit model is the most widely used formula for
     mode choice modeling, it is starting to be replaced by the  nested logit model.  This is because the
     true nested model is more adept at handling sub-mode splits (e.g., bus vs. rail, within the  transit
     mode).  There is some evidence that the nested logit structure more closely represents travelers'
     trade-offs of attributes when  selecting  a travel  mode.   The nested  logit  also  eliminates the
     " independence of irrelevant alternatives" property, which can  cause mode share estimation errors
     for travel modes with similar characteristics.
U. S. Environmental Protection Agency         .  ,  ..                      ,--.-'-'           A-23

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions

      Separate Mode Choice Models by Purpose: The Case Study Model includes a true mode choice
      model only for HBW trips. HBO and NHB mode shares are calculated as a distance-based factor
      multiplied by the HBW shares.  This implies that the non-work mode shares move up or down in
      the same proportion as the HBW mode shares, which is unlikely.  There  is evidence from  around
      the country that the mode share of non-work trips is more sensitive to price and less sensitive to
      time than for work trips, which is logical. It would be preferable to develop separate mode choice
      coefficients for work  and non-work trips.  This is  particularly important, given  that the mode
      choice coefficients influence trip distribution in this model as well.

      One of the reasons why there is so much attention to work trips is because the CTPP provides a
      potentially robust source of calibration data for commuting.  More  extensive home interview
      surveys, perhaps as large as 1.0 - 1.5%, may be needed to capture enough observations of HBO and
      NHB trips (especially using transit) to permit the calibration of usable coefficients.

      Improved Vehicle Age Mix Submodel: The Case  Study  Model's vehicle age mix submodel is
      overly simplistic and based on rough approximations. For example, its assumption that the age mix
      is sensitive to the  ratio of the annual fee divided by the value of the vehicle is open to question.
      The submodel that estimates the average value of the vehicle by age and type was not rigorously
      calibrated due to time constraints but should be redeveloped based on known data.  Further,  for the
   •   emission-based fee, some  provision in the.submodel  should be made to allow motorists the option
      of repairing the vehicle to improve its emissions, thus avoiding some or all of the fee.  Additional
      research is necessary to  determine  if an annual or quarterly fee would truly  affect vehicle
      ownership decisions or day-to-day travel choices (or both). s Finally, some research should be
      conducted into the effect of leasing vs. buying on vehicle replacement decisions.

      Long-Term Land Use Allocation Effects: The Case Study Model, as with most  travel demand
      models,  simulates  travel  choices that are made in the  short-term;  i.e., hours, days, weeks.
      However, those pricing strategies -which travelers perceive to be permanent can also  have long-
      term  influences; i.e., months or years.  It would be very desirable for the travel model to be
      supplemented by a land use model that could adjust the  allocation of  households and jobs  in
      response to such pricing changes.  For example, a significant change in the toll on a bridge might
      be expected ultimately to  persuade some existing travelers, over time, to  change their  location of
      residence, work, shopping, etc.  Such tolls might also  influence the locational decisions of residents
      and/or employers who are new to the region.  Modifying future  land use in response to pricing
      changes  is one way to respond to the issue of " induced, demand". The  few land use allocation
      models now in use employ travel time in their allocation process, just  as typical trip distribution
      models do. Introducing pricing variables, such as composite impedance, might be one way for land
      use allocation models to be sensitive to the travel cost of users.              •

      It should also be noted that the sensitivity of long-term locational  decisions to pricing is probably
      quite different  than short-term  decisions on where to travel or by what mode.  This  long-term
      sensitivity is probably lower, but additional research is needed to quantify this difference.

      Emissions Estimate Sensitive to Trips: The Post-Processor for Air Quality (PPAQ) program used
      in this project greatly facilitates the calculation  of  the emissions burden  from each scenario's
      loaded network.  However, PPAQ uses  only VMT  in its calculations.  Although  the additional
      emissions from cold starts, and  hot soak and diurnal VOCs are included in the MOBILESa
      emission rates, these should more properly be modeled as a function of the number of trips rather

A-24                                                               U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                       •        ,                      to Reduce Transportation Emissions

      than VMT. This would more accurately differentiate the emissions impact of those strategies that
     .reduce vehicle trips from those that reduce VMT.
                                       •  • •    >        '••'..'

      This can also be partially  addressed through  an alternative  technique  for estimating cold start
      emissions [8]. That method uses purpose-based factors to split the vehicle.trips into cold- and hot-
      start trips. The two trip tables are then assigned and the assignment software keeps track of where
      the cold start VMT occurs on the network.  This provides an  estimate of cold start VMT on each
      link, which improves the estimation of cold start emissions without having to calculate the change
      in vehicle trips.


      A.3      Application of the Case Study Model

      In this project,  market-based measures, to  reduce mobile .source emissions were classified into
      several groups.  Within each group, specific variations of the measures were identified for detailed
      analysis.  The following sections discuss how the Case Study Model was used to evaluate  each
      measure.  This information is presented primarily to provide guidance to travel demand forecasters
      in the development and use of similar techniques for evaluating market-based measures in  their
      own areas.
<*      .        ,                  '               '         ,           '."'•.             '    .

      A.3.1    Parking Pricing Measures
                           IN                     i         •   '                         '      ' •
      Analysis of parking pricing measures involves adjusting parking  costs for a specific part  of the
      region (e.g., the CBD), and/or a specific time period (peak vs. off-peak), and/or a specific segment
      of the travel  market  (e.g.,  work trips).   Parking pricing measures  are represented by a fixed
      incremental change or a percentage adjustment to  existing  parking costs.  Parking costs are
     represented in the model by specifying the average daily cost for  parking in each traffic analysis
     zone. A separate cost can be supplied for each zone for both LOV parking and HOV parking.  This
     allows the specifiqation of strategies with differential pricing based on vehicle occupancy.
                 / .               ^ -          .        .        .            •   .              \
     Parking costs are  used to compute combined  impedances  and to estimate modal shares.    This
     affects the estimation of traveler choice of destination and  mode  of travel (with the potential to
     affect travel by time of day). Five parking scenarios were examined in two broad categories:

               •   Increase in average cost of parking (all  spaces)

            ^          (a)  $4 per day increase in the core area and $2 increase elsewhere
                      (b)  $ 10 per day increase in the.core area and $5 increase elsewhere
                      (c)  $5 per day increase in the core area and no change elsewhere

               •  Increase in average cost of parking (employee parking only)    ".-

                      (a)  same as (a) above but only applied to home based work trips
                      (b)  same as (b) above but only applied to home based work trips

     Each  of the  scenarios are represented by adjusting the  zonal data  file used in the combined
     impedance step and the mode choice step.  All incremental cost adjustments are converted to 1980
     dollars by using the 0.57 factor prior to adjusting the  parking costs  in the zonal data file. Table A-
     11 summarizes the impacts of this scenario.  -                             ' .   .'      .

U. S. Environmental Protection Agency                 ~~'            "                    ~~'   :     A~25

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
                                          Table A-ll
                                     Parking Cost Impacts
     Model Step	Impact
     Trip Distribution       The composite  impedance changes for paths destined to zones that have
                           modified parking costs. This does not change the total number of trips that
                           are attracted to these zones  since  that value  is fixed from the trip
                           generation  step.  It does change the  distribution of trips to zones with
                           modified parking costs.

     Mode Choice          Changes in parking cost affect both the split between auto and transit and
                           the split between LOV auto trips and HOV trips.

     Assignment           VMT decreases due to fewer vehicle trips.

     Emissions        HC, CO and NOX emissions decrease due to the decrease in VMT.


     A.3.2    Roadway Pricing Measures

     Roadway pricing involves charging a fee (or toll) for vehicles  to use certain roadways or to cross a
     border into a designated area.  The toll can vary by time of day, by vehicle occupancy, and/or by
     the level of congestion on the roadway.  Most such proposals would have the toll collected through
     automatic vehicle identification so that no delay would be imposed in collecting the toll.

     It was assumed that this toll would influence travelers' choice  of 1) destination zone, 2) travel
     mode, and 3) highway path. The Case  Study Model does not include a time of day  submodel,  so
     influences on the time of travel could not be estimated.  In effect, the toll is assumed to be assessed
     to all vehicle trips  (LOV and HOV) on the selected  roadways  at all times of the  day.  In this
     scenario, the selected roadways  were those specific  freeway and  expressway  links in the
     Washington, D.C. region that had been  shown, through previous trip  assignments, to have  an
     estimated peak hour volume/capacity (V/C) ratio of 0.9  or higher.  These were almost all of the
     Interstate and major through Primary roads in the region (727 one-way links out of a total of 18,400
     links). Although these priced roadways represent 4% of the total coded network, they carry about
     31% of the total daily VMT.

     One level of toll was analyzed: $0.10/mile.  This  is  a typical high-side value of the toll  being
     charged in newly proposed toll roads in various areas.  The first step was  to convert this into 1980
     year dollars  by multiplying by 0.57,  as noted above.   The equivalent 1980  toll  level is then
     $0.057/mile. The increased cost of using certain roads  was  first reflected in the path skimming
     step, in which  path-building impedance is defined as the sum of auto operating cost (cost/mile
     multiplied by distance), and the money value of travel time. Priced roadway paths and non-priced
     ("free") roadway paths were then built and input to the toll diversion model, which  estimated the
     priced route/free route share for allO/D pairs.  The average toll paid is then the toll on the shortest
     path, multiplied by the proportion of  drivers who are estimated to use that path.   A weighted
     average of the priced and free travel time and the priced and free distance were also computed.
A-26                                                              U. S. Environmental Protection Agency

-------
                                                         Technical Methods for Analyzing Pricing Measures
                                                                     to Reduce Transportation Emissions

      The average time, distance, and toll for each O/D pair was then input to the mode choice program
      in order to calculate composite impedance.  That impedance value was used in the gravity model to
      distribute trips from origins to destinations. The average time, distance, and toll were again input
      to the mode choice program, this time to compute the split among travel.modes.  The toll diversion
      model was applied a second time to  split  the vehicle .trips by priced/free path. The assignment
      program assigned priced trips to the priced paths and free trips to the free paths.  Within the four
      assignment iterations, priced  trips were allowed to "migrate"  to  the free paths (and back) in
      response to changing relative  congestion levels between priced and free roadways; however, free
      trips were required to stay on  a free path.  Table A-12. summarizes the impacts of this scenario on
      the travel choices.
                                          Table A-12
                                  Roadway Pricing Impacts
     Model Step
      Impact
     Highway Skims


     Trip Generation
     Trip Distribution



     Mode Choice

     Toll Diversion

     Assignment

     Emissions
     Highway paths are.modified to select paths that minimize the total cost.
     Average time, distance, and toll values are estimated that reflect the impact
     of the priced roadways.                                       -
     No impact.    .-,.-'

Composite impedance increases for paths using the priced roads due to the toll.
     It also increases slightly for paths using free roads because they are longer
     (drivers are accepting increased distance, and operating cost, in order to
     avoid the toll).

     Cost of driving increases; driving alone decreases, and carpooling and
     transit increase slightly.
     Presence of tolls causes a share of the remaining vehicle trips in affected
     O/D pairs to switch to the free paths.
     VMT decreases due to fewer vehicle trips. However, congestion increases
     on the free roads, resulting in much slower overall average speed.
     HC and CO emissions increase, and NOX emissions decrease due to the
     lower average speed.                          -
     These results fit with the initial expectations and the results of other research, up through the toll
     diversion  model.  The assignment results, however, bear further explanation.  Although some
     switching to carpools and transit occurs, this does not occur in significant numbers.  Some travelers
     switch destinations to one which can be reached via a free route.  The majority of travelers are
     those who dp not change their destination or mode and  the primary method for them to "escape"
     tolls in this model is  by switching to a free route. In this scenario, the free routes are generally the
     arterials, which already had slower average speeds to begin with, and those speeds are made even
     slower with the additional traffic from the priced freeways.  Reducing average network  speeds
     generally has a negative effect on HC and .CO (although  this depends on the existing average speed
     values, as noted above).                    -       . '   F
U. S. Environmental Protection Agency
                                                                                            A-27

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
      Part of the problem may lie in the definition of " congestion".  In this scenario, the priced roads
      were those freeways and expressways with a V/C of 0.9 or more. Although V/C is a good indicator
      of a roadway's congested speed compared to its free-flow  speed, that is a false indicator of
      emissions. That is, MOBILESa's emission rates are not based on congestion, per se, but absolute
      average speed. Consider the case of a freeway section with a free-flow speed of 60 mph. With
      congestion, the speed might drop to 40 mph, a difference of 33%.  But according to MOBILESa,
      vehicles on that roadway will still have lower emission rates than vehicles on a nearby arterial'
      whose congested speed might typically be 35 mph.  This is true even through the arterial's free-
      flow speed might be 40 mph, in which case the arterial is experiencing less congestion than the
      freeway (i.e., a speed drop of 13%).

      Still, this example highlights some of the complexity of roadway pricing analysis.  Even with the
      enhancements made to the Case Study Model, the following shortcomings are still seen:

               •  No price-sensitive time of day model.

               •  Toll diversion model not specifically calibrated to Washington, D.C. area.
               •  No speed feedback in the model set.

      the above problems notwithstanding, a key observation of this analysis is that a network-based
      procedure is essential to the complete analysis of roadway pricing measures.  The interactions
      among  traveler  choices  of destination,  mode, and path are  too  complex  to  model  without
      performing a specific assignment of vehicle trips to a roadway system.

      A.3.3     Vehicle-Based Measures
     A.3.3.1    VMTFee

     A fee would be assessed for each vehicle-mile driven.  It was assumed that this fee would be
     collected in such a manner that travelers would be aware of it on a daily basis and therefore the fee
     would influence their short-term travel decisions.  Thus, the VMT fee was modeled in terms of an
     increment to the automobile operating cost parameter.  No impact on vehicle ownership levels, land
     use, or long-term locational choices was assumed.

     Three levels of VMT fee were analyzed: $0.04/mile, $0.10/mile, and $0.40/mile.  The first step was
     to convert thesp into 1980 year dollars by multiplying by  0.57, as noted above.  The equivalent
     1980  fee levels are then $0.023/mile, $O.Q57/mile? and $0.228/mile. These values were then added
     to the base automobile operating cost of $0.1 I/mile to yield new automobile operating costs of
     $0.133/mile, $0.167/mile, and $0.338/mile.  The mode choice/composite impedance application
     program  includes  a specific user-coded parameter for auto operating  cost per  mile  and this
     parameter was adjusted for each run,

     In addition, the increased cost of driving was reflected in the highway network  path building and
     skimming steps, performed using the MINUTP PTHBLD program. Since paths  are built based on
     an impedance that includes time and operating cost,  changes in the operating cost will affect paths.
     In PTHBLD, this change is achieved by coding DCOST as  133,  167, or 338, for the three pricing


A-28                                                              U. S. Environmental Protection Agency

-------
                                                         Technical Methods for Analyzing Pricing Measures
              •    •                                                   to Reduce Transportation Emissions

      levels.  Table A-13 presents a summary of the travel choices affected by this change, as reflected in
      the Case Study Model.             ,

      It should be noted that the Table A-13 shows no impacts of VMT fees on trip generation. This is
      because the Case Study Model's trip generation module is not sensitive to pricing changes.
     Model Step
                                          Table A-13
                                      VMT Fee Impacts
 Impact
     Highway Skims


     Trip Generation

     Trip Distribution
     Mode Choice


     Toll Diversion

     Assignment


     Emissions
Highway paths are nTodified to select paths that are shorter in distance,
to minimize the cost of travel.
No impact.              ,                                 '

Composite impedance increases with distance (operating cost) and goes
up relatively faster for longer-distance trips; thus, travelers tend to
choose destinations that are slightly closer. Average trip distances
decrease slightly. ,

Cost of driving increases; driving alone decreases, carpooling decreases
slightly, and transit  use increases.                    '
No impact.

VMT decreases due to fewer vehicle trips and shorter trip lengths;
reduced VMT means less congestion, resulting in faster average speed.

Emissions of all pollutants decrease due to less VMT and higher average
speed.
     A.3.3.2  Fuel Tax
                                    f              ,       . -                         '         '

     The tax on all motor fuels would increase.  It was assumed that this increase would-be perceived as
     simply more expensive gasoline and that this would influence the short-term decisions of travelers.
     Thus, the fuel tax was modeled  in terms of an increment to the  automobile operating cost
     parameter.             ,     ..    "

     Three levels of fuel tax were analyzed: $0.20/gallon, Sl.OO/gallon, and $1.50/gallon.  These were
     converted into  1980 year dollars:  $0.114/gallon, $0.570/gallon, and $0.855/gallon.  These were
     then converted into cents/mile by dividing by an average value for on-road fuel economy. A value
     of 19 mpg was calculated from, the  1990 Nationwide Personal Transportation Study (passenger cars
     only), but this was increased to 20 mpg for all scenarios by assuming that a short- to medium-term
     (3-12 months) impact of higher fuel prices would be to encourage the purchase of more fuel-
     efficient vehicles.  The resulting additional cost per mile is then: $0.0057/mile, $0.0285/mile, and
     $0.0428/mile, making the new automobile operating cost parameter: $0.1157/mile, $0.1385/mile,
     and $0.15287mile, respectively.                      .                    .                     •
U. S. Environmental Protection Agency
                                                                                             A-29

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions                                       .      .

      This new cost of driving was then input to the model in exactly the same manner as for the VMT
      fee, as described above.  The impacts were the same as noted in Table A-13, although at a lower
      level due to the lower level of operating cost increase.                     ^

      A.3.3.3   Age-Based Registration Surcharge
                                                       •'           ,1                 .
      A surcharge would be placed on the annual registration cost of all vehicles and would vary with the
      age of the vehicle.  It was assumed that this fee would be  collected annually and that  travelers
      would not be aware of it on a daily basis and  therefore the surcharge would not influence their
      short-term travel decisions. This surcharge was assumed to affect only the age mix of the vehicle
      fleet by encouraging drivers to trade-in their older vehicles fo^newer ones.  Such older vehicles
      were presumed to be removed from the fleet (i.e., scrapped) and replaced with newer vehicles, in
      roughly the same proportion as newer vehicles currently exist in the fleet.  The MOBILESa default
      age mix by vehicle type was used as the initial age mix.

      The analysis methodology presented estimates only the impacts due to changes in  vehicle mix by
      age and assumes that no changes in VMT occur due to the shift from older vehicles to newer
      vehicles.  Since the fee is levied  solely based on vehicle age, no changes in travel characteristics
      are assumed except for changes in vehicle mix by model year. The fee can be structured in various
      combinations of fee variation by each separate model years, or can be  charged by certain ranges of
      vehicle ages, or it can be varied among different vehicle types.

      The changes in registration distribution by model year are estimated through a process in which
      certain assumptions about the value of the vehicles in each age group and changes in car ownership
      behavior are made to quantify the change in number of vehicles under each model year. The basic
      principle in the analysis methodology is that the vehicles are dispensed or scrapped at a rate which
      is proportional to the ratio of the registration fee to the market value for that particular model year
      of vehicle.

      General Assumptions:                   '
                                                  ,.'•••.     i .
               •  It is assumed that market value for a particular model  year of vehicle is the same for
                  all the vehicles under that year.

               •  No drop in the latest model year vehicle registration is assumed because the market
                  value for these vehicles is much higher than the estimated registration fee.

               •  LDGT2, HDGV, LDDT and HDDV are assumed to have lower price elasticity than
                  other lighter duty vehicles because they are mostly used for business purposes and
               •  are less likely to be scrapped or replaced (see Table A-14) •
                  Motorcycles are assumed to have lower  price elasticity because of their unique
                  nature.

      Model Specific Assumptions:
      (these are assumptions that could be replaced by more accurate data if available to the user)
               •  The same estimated market value is assumed for both LDGVs and LDGTls.
               •  The estimated  market value for a late  model year HDDV and Motorcycle are
                  assumed to be $60,000 and $7,500 respectively.


A-30            •           ,                                       [/. 5. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                                                                      to Reduce Transportation Emissions
                                            Table A-14
                                 Price Elasticities by Vehicle Type
Type
Elasticity
LDGV
1 -1.0 .
LDGT1
-1.0
LDGT2
-0.9
HDGV
-0.5
LDDV
-1.0
LDDT
-0.9
HDDV
-0.5
1 • MC
-0.75
      The two major inputs to the analysis are the existing vehicle mix by model year and the average
      market value for each model year of vehicle.  In the Case Study Model,, the. MOBILESa default
      vehicle mix is assumed for the base conditions, but these can be replaced by more accurate local
      data if available or from any other updated versions of MOBILE.                         "

      The other key input to the analysis is the average market value for each model year of vehicle
      which could also be replaced with more accurate local data. For the Case Study Model, the average
      market value for the latest model year (1993) light duty gasoline vehicle is assumed to be $17 922
      (source: Automotive News, 26 May 1993, p.  61) and the value for older models is estimated from
      this number assuming a price deterioration rate which is proportional to the age of the vehicle  The
      following summarizes the specific steps of the analysis:

      Step 1  Percent Registration Fee Cost of the Total Market Value:

               The percentage of the registration fee cost of the total market value for that model year
               of car is calculated. This percentage of registration fee cost of the  total value is assumed
               to be  equal to the percentage of, vehicles scrapped or replaced.
               Percent Registration Fee = Registration Fee    * 100 / Total Market Value
                                                        my
                                                                                   • my
      Step 2  Estimation of Vehicle Scrappage or Replacement:
               It is assumed that the vehicle scrappage or replacement percentage due to the registration
               fee is  same  as the percentage of registration fee  of the total market value.  If the
               registration fee exceeds the market value of the vehicle (as in the case of older cars) then
               all the  vehicles under that model year are replaced by newer model years. For example,
               if the registration fee is 1% of the market value, then 1% of those model year vehicles
               are assumed to be replaced by newer cars.  This approach is applied to all the vehicle
               model  years except for the latest model year which is assumed to be independent and no
               scrappage or replacement for this year is calculated.

               To account for the variation  among different vehicle types, especially the 'differences
               between lighter and heavy duty vehicles, different price elasticities  as'given in Table A-
               14. are assumed for different vehicle types.   Lower elasticities are assumed for heavy
               duty vehicles since most of these vehicle types are generally used for business purposes.
               The estimated, scrappage percentages as calculated above are corrected for these price
               elasticities for each vehicle type.        .  •       ',                         .•
U. S. Environmental Protection Agency
                                                                                             A-31

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions

              Example: Assume a market mechanism with a $900 registration fee for cars between
              ages 15 to 25, $600 for ages 5 to 15, and $300 for ages 0 to 5 years, the estimated market
              value of a 24 year old car is $400, the registration fee for this model year vehicle from
              the above description is $900, the percentage of registration fee is more than 100% of the
              total market value, hence all of these model year cars are  assumed to be replaced by
              newer cars.  Similarly, for a 7 year old model year car with a market value of $9,100, the
              registration  fee for this vehicle is 600, which is about 6.6% of the market value.  Hence
              6.6% of these model year cars are assumed to be replaced with newer cars.

     Step 3  Estimation of Replacement Vehicles-

              In the analysis all the scrapped vehicles are replaced with newer model year vehicles in
              the same proportion as their existing vehicle distribution and weighted  by appropriate
              values to make1 the total of the distributed percentages to 100.

     Step 4  Results from the Analysis

              All the above described steps have been  included in a spreadsheet which performs the
              calculations and estimates the resulting vehicle registration distribution. ' The resulting
              revised age  mix was used in the PPAQ/MOBILE5a run instead of the default age mix.

              Three  levels  of  maximum surcharge  were  assumed:  $100/year,  $500/year,  and
              $900/year.  The maximum surcharge was assessed on vehicles 20 years old or older, no
              surcharge was assessed on vehicles 3 years old or newer, and an intermediate surcharge
              was assessed on vehicles between 3 and 20 years old, varying with age (see Table A-9
              for an example, showing the $900/year surcharge). Table A-15 summarizes the impacts.

                                         Table A-15                           s
                                Age-Based Surcharge Impacts

     Model Step	Impact	;	
     Emissions        Emissions decrease due to shift in age mix of fleet from older to newer
                      vehicles.
     A.3.3.4 Emission-Based Registration Surcharge

              •    .,                  " •'.                 .  '•
     A surcharge would be based on the estimated total annual emissions of all vehicles.  An estimate
     would be made of the annual emissions of each vehicle (by the 8 MOBILESa vehicle types), using
     the default MOBILESa emission rates by age and the default MOBILESa annual mileage by age.
     1993 was assumed as the base year, so a 25-year-old vehicle is of model year 1969, for example.
     It was assumed that this fee would be collected annually and that travelers would not be aware of it
     on a daily basis and therefore the surcharge would not influence their short-term travel decisions.
     This surcharge was assumed to affect only the age mix of the vehicle fleet by encouraging drivers
     to trade in their older vehicles for newer ones. Such older vehicles were presumed to be removed
     from the fleet (i.e., scrapped) and replaced with newer vehicles, in roughly the same proportion as
     newer vehicles currently exist in the fleet. In the definition of this measure, it was assumed that the
     surcharge would be based on the estimated  emissions for all vehicles of a certain age, not measured
                         	      ,	•	'	'	"	j	        _
 A-32   •                                                           U.S. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                                                                      to Reduce Transportation Emissions

      emissions. Vehicle owners would not have the option of repairing and retesting the vehicle. The
      MOBILESa default age mix by vehicle type was used as the initial age mix.

      The emissions fee is similar to the age-based registration surcharge, the only difference being that
      the fee  is levied on the total emissions emitted  by the vehicle.  The emissions fee  has  certain
      advantages over the registration surcharge in terms of equity, since this measure changes the price
      based on total emissions rather than on age. Empirical evidence shows that newer cars travel much
      more than the older cars, although this is more than compensated for by the net emissions per mile
      (emissjon factors)  being  much  lower  for  newer  cars.   This  type of  measure takes these
      characteristics into consideration and distributes the price,uniformly  over all  vehicle  age  groups
      based on their total emissions. Equity is also maintained across different vehicle, types  since the
      heavy-duty vehicles which have higher emission rates and VMT accumulation are assumed to pay
      a higher fee compared to the lighter duty vehicles.

      The changes in registration distribution are based on essentially the same technique as described in
      the previous section on age-based registration surcharges.  The differences in assumptions are as
      follows:                                                    .

      General Assumptions:                            .'•_'-.

               •   An average emission factor and VMT accumulation is used for all the vehicles under
                   a particular model year.
U. i. environmental Protection Agency
                                                                                              A-33

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions                                           ,

     Model Specific Assumptions:
     (these are assumptions that could be replaced by more accurate data if available to the user)
              •   MOBILESa estimated average emission levels by each model year and  default
                  registration distribution by age are used in the analysis.

     The three major inputs to the analysis are the vehicle emission levels by model year, the annual
     mileage accumulation by model year, and the average market value for each model year of vehicle.
     In the Case Study Model, MOBILESa default annual mileage accumulation rates and emission
     rates were usecl for the base conditions, but these can be replaced by more accurate local data if
     avai lable or from any other updated versions of MOBILE.

     The following two steps are used to estimate the change in annual user cost for this measure. From
     that point, the analysis proceeds exactly as in Steps 1 - 4 of the age-based registration surcharge,
     described above.
                             	„      "    ",''*,'            ' '   ' '     '           ' ,     ,  'I,  ,  '
                  !;i  '        •         '     '•'  '      .''',';.     ;         , •   •   •
     Step 1   Estimation of Pollutant Tonnage:

              The initial step in the analysis is the estimation of pollutant levels for each model year of
              vehicle. The emission rates (g/mile) for  each model, accounting for the appropriate
              control measures such as inspection and maintenance,  reformulated gasoline,  etc., as
              well  as local  conditions such  as temperature  and  altitude, • are  estimated using'
              MOBILESa.    A-single  index of pollutant tonnage is calculated  by combining the
              estimated levels of HC, CO and NOX, using equivalency factors for each pollutant based
              on previous EPA studies.  The factors used in this project are shown in Table A-16. The
              level of each pollutant multiplied by its equivalency factor is  summed to  get to the
              combined emission rate for each model year.

              Combined Emission Rate my(gms/mi)
               = EJIC* HC level + ECQ* CO level + E^Qx* NOx level "


                                          Table A-16
                                  Pollutant Equivalency Table
Type
Equivalency
EHC
1.0
ECO
0.1
ENOx,
0.9"-
A-34                                      ,            .            U. S. Environmental Protection Agency

-------
                                                         Technical Methods for Analyzing Pricing Measures
                                                                    to Reduce Transportation Emissions

      The combined emission rate by model year multiplied by the annual mileage accumulation for that
      model year vehicle gives the annual total tonnage of pollutants:
                                '»_                .               -
               Combined Annual Tonnage of Pollutantsmy                        .      ,
               = Combined Emission Ratemy * Annual Mileage Accumulation mv

      Step 2  Estimation of Emissions Fee by Model Year:

               The combined annual tonnage of pollutant as calculated from the previous step for each
               model year is multiplied by the emissions fee set for the market mechanism. This gives
               the additional price paid by each model year's vehicles due to the emissions fee.

               Price due to Emissions Feemy

               = Combined Tonnage of Pollutantmy(tons) * Emissions fee ($/ton)

       (From this point, continue with Step 1, from Section A.3.3.3, described above.)

       Application of the Methodology:

       The calculations involved in the estimation of the changes in vehicle registration distribution due
       to emissions fee are performed in a spreadsheet,  an example of which is  shown in Table A-17
       These revised distributions go into the MOBILE input section of the PPAQ set up file.

       This methodology is a sketch planning technique involving numerous assumptions and factors in
       the process which are based solely on engineering judgment and are not substantiated by
       empirical results.  The  methodology  itself is in  a rudimentary stage and requires further
       enhancement with respect to the factors and values assumed in the process. The key assumption
       regarding changes in vehicle ownership in response to external costs such as emissions fees and
       registration  surcharges should be verified.  The assumed independence between changes  in
       vehicle ownership and changes in the annual vehicle mileage also bears further examination.
'}. S. Environmental Protection Agency
                                                                                           A-35

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions

             • _    '•;          '  :    ;  .'.     TableA-17   '  /   .'.  _   '  |
             Sample Vehicle Age Mix Adjustment Spreadsheet (Emission Fee)
|UH»V|
r»<
2>
24
23
22
21
20
19
IS
17
16
IS
14
13
12
11
10
9
g
7
6
5
4
3
2
1


•xnK
n,KU
J«"
IK6»
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
I9S5
1936
1987
1988
1919
1990
1991
1992
1993


ntngfalarr
MOBILE
4,305
4,305
4,305
4,305
4,305
4,305
4,559
4.829
5.114
5,416
5,737
6,076
6,435
6.815
7,218
7,645
8.096
8.575
9,082
9.619
10.187
10.7S9
11,427
12.102
12.818


1.0 0.1 0.9 elasticity: -1.00
MOBILE MOBILE* MOBILES MOBILES combined combined
earn, mile* othllCtvl CO level NOx level level lons/yr
IV8,36y
174.064
169.759
165.454
161,149
156,844
152,539
147,980
143,151
138.037
132,621
126,884
120.808
114,373
107,558
100.340
92.695
84,599
76.024
66,942
57,323
47.136
36,347
24,920
12.818


8.89
9.44
9.28
6.17
6.10
6.03
5.19
s.oi
4.94
4.80
4.65
1.63
1.70
1.59
1.50
. 1.50
1.31
1.22
1.11
1.00
0.89
0.72
0.60
0.47
0.34


101.82
96.65
95.30
81.15
80.10
79.05
57.74
56.56
55.31
53.98
52.58
15.35
25.08
22.94
21.53
21.76
,19.21
17.62
15.33
13.63
11.83
9.72
7.76
5.68
3.47


M5
4.35
4.35
4.35
3.51
2.91
2.90
2.88
2.99
2,94
2.89
2.39
1.56
1.54
. 1.42
1.43
1.36
1.29
1.29
1.20
1. 10
1.00
0.89
0.78
0.66


22.99
23.02
22.73
18.20
17.27
16.55
13.57
13.32
13.16
12.84
12.51
5.31
5.61
5.28
4.93
4.97
4.46
4.14
3.81
3.45
3.06
2,59
2.18
1.74
1.28


07IT
0.11
0.11
0.09
0.08
0.08
0.07
0.07
0.07
0.08
0.08
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.03
0.03
0.03
0.02
0.02


fee/ton;
cAirnmed
— sw
$500
$60,0
$700
$900'
$1,000
S1.200
$1,500
51,800
$2,100
$2,500
$3,000
$3.500
$4,100
• $4,900
$5,700
$6,700
$7,900
$9,100
$10,500
$12,000
$13,500
$14,700
$14,800
$17,922


SI 0.000
ehungc
into
$1,089
$1,090
$1,076
$862
$818
$784
$681
$707
S740
$765
$790
$355
$397
$396
4392
'$418
$397
$390
$380
$365
$343
$308
$274
$232
$181


Icechg
„•/,„!
vulue
272.1%
218.0%
179.4%
123.1%
90.9%
78.4%
56.7%
47.2%
41.1%
36.4%
31.6%
11.8%
11.3%
9.6%
8.0%
7.3%
5.9%
4.9%
4.2%
3.5%
2,9%
2.3%
1.9%
1.6%
1.0%

avg age:
MOBILESa rcgiKlratiim mix by year
leaning 1 July 1990}
base-/. new 54 eum.%
by year by year %lcft loll
1.10%
P.30%
0.40%
0.50%
0.60%
0.80%
1.10%
1.50%
1.40%
1.90%
2.40%
3.70%
4.70%
5.40%
5.00%
5.10%
5.00%
5.60%
7.70%
8.10%
8.40%
8.20%
8.30%
7.90%
4.90%
100.0%
8.2
0.00%
0.00%
0.00%
0.00%
0.07%
0.20%
0.52%
0.86%
0.89%
1.31%
1.79%
3.53%
4.53%
5.33%
5.06%
5.22%
5.22%
5^93%
8.25%
8.79%
, 9.23%
9.13%
9.34%
9.01%
5.79%
100.0%
7.1
1.10%
0.30%
0.40%
0.50%
0.53%
0.60%
0.58%
0.64%
0.51%
0.59% .
0.61%
0.17%
0.17%
0.07%
-0.06%
-0.12%
-0.22%
-0.33%
-0.55%
-0.69%
-0.83%
-Q.93%
-1.04%
-1.11%
-0.89%
0.0%

1.10%
1.40%
1.80%
2.30%
2.83%
3.43%
4.02%
4.66%
5.17%
5.76%
6.36%
6.53%
6.70%
6.77%
6.71%
6.59%
6.37%
6.05%
5.50%
4.81%
3.98%
3.05%
2,00%
0.89%
0.00%

avg fee:
-new/buse
— CT?T
0.0%
0.0%
0.0%
11.5%
24.5%
46.9%
57.1%
63.9%
69.2%
74.8%
95,4%
96.4%
98.8%
101.1%
102.4%
104.4%
105.8%
107.1%
108.5%
109.9%
111.3%
112.6%
114.1%
118.2%

$360
Nqte: "LDGV" = light-duty gasoline vehicle
     A.3.4    Modal Subsidies

     Analysis of modal subsidy measures involves reducing cost to travelers who use a mode that is
     promoted by subsidy.  Typical subsidies involve incentives  for transit (fare  reduction, on-site
     transit  stops),  cost incentives  for ridesharing  (cost  reimbursement,  reduced parking  costs,
     preferential parking locations, toll reductions) and transportation allowances (parking " cash out").
     Two levels of transit fare reduction were studied:

                  •   50% reduction in transit fares
                  •   100% reduction in transit fares (i.e. free transit service)

     A reduction in transit fares influences a traveler's choice of destination zone, travel mode and
     (indirectly) highway path. Destination zone choice is affected by changing the computation  of
     combined impedance which  is an  input to  the trip  distribution model.  Travel mode choice is
     affected in the mode choice model by making transit more attractive and thereby increasing transit
     share. Highway paths are affected by assigning fewer vehicle trips to the highway network which
  ,   .reduces congestion effects and causes path changes to occur in the traffic assignment model.

     There are two "matrix"  files that reflect zone to zone transit fares  in 1980  dollars.  These files
    . (walk to transit fare matrix  and drive to transit fare matrix) are modified with the MINUTP
     program MATRIX. For the 50% reduction, all cells are factored by 0.50. For the 100% reduction,
     all cells are replaced with zero. These revised fare matrix files are specified in the model setup and

A-36                                      ,                         U. S. Environmental Protection Agency

-------
                                                         Technical Methods for Analyzing Pricing Measures
                                                                    to Reduce Transportation Emissions
      used during both the combined impedance calculation and the mode choice estimation steps.  Table
      A-18 summarizes the impacts of this scenario on the travel choices:
      Model Step
                                           Table A-18
                                     Modal Subsidy Impacts
      Impact
      Trip Distribution



      Mode Choice


      Assignment


      Emissions
     The composite impedance decreases for those paths that show a reduction
     in transit fare under this scenario.  This modifies the distribution of trips in
     favor of transit accessible zone pairs.

     Transit costs decrease causing transit shares to increase and vehicle trips to
     decrease.

     VMT decreases due to fewer vehicle trips.  This can also result in some
     increases in speed for certain highway facilities.

HC, CO and NOX emissions all decrease due to the decrease in VMT.
     A.4      Other Model Enhancement Issues

     A.4.1    Addressing the Unknown Sensitivity to Pricing

    • The sensitivity of urban travel demand to pricing has not received much attention in the past.  The
     best known analysis has been the "Curtin Rule", which states that a 10% increase in transit fare
     leads  to a 3% drop in transit ridership.  In almost all other applications, the price of transportation
     to the traveler has often been assumed to change at the same rate as inflation and thus be of no
     more  or less importance in the future than today.

     The rapid rise in gasoline prices in the mid-1970's and early 1980's, combined with a flurry of
     activity in rail transit planning created a need for mode choice models which were sensitive to auto
     operating cost to the  user.  There is now over a  decade  of experience  with such models,  and
     transportation planners have some confidence that adjusting this parameter to account for increases
     in the price of driving, as described  in sections A.3.3.1 and A.3.3.2, produces reasonable results
     that are not too great an extrapolation beyond present experience. A similar comment can be made
     for the price of parking, which has been identified in the past ten years as a significant influence on
     mode choice.

     However, when it comes to more innovative pricing strategies, such as congestion pricing, planners
     in the U.S. have  little  or no  experience on which  to draw.  Moreover, there is an  intuitive  and
     logical belief that in some way,  strategies such as congestion pricing will have a travel impact far
     beyond the choice of travel mode.  But the problem remains: without actual experience from which
     to observe and compare, how can demand modelers  estimate the impacts of such measures without
     falling back on theory?

     A powerful argument  can be made  that " before-and-after" studies offer the best evidence of
     travelers' response to pricing measures.  The problem, of course, is that this assumes that a strategy
U. S. Environmental Protection Agency
                                                                                           - A-37

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions

     has actually been implemented, in a relatively controlled manner, and without other confounding
     factors that could also influence the outcome. Careful analysis  of the last gasoline crisis might
     yield some insight into motorists' response to  steep and rapid increases in the user cost of driving,
     assuming one could filter out the impact of the apparent  shortage of supply and the psychological
     (and real) effects of long lines at the pump. Analyses of the ridership effects of existing differential
     peak/off-peak transit fares, before and after implementation, might produce some insight into the
     cost sensitivity of trip start time.
                  -          .                '•           /       '  '      -!•''
     Another criticism of the pricing parameters currently used in models is that they are based on
     cross-sectional surveys.  Such parameters are  not based on what a fixed group of people did over
     time, but on what many people did at one time.  Current model calibration theory says that if you
     can  take a "snapshot"  survey of enough people with  different enough travel conditions and
     choices, you  can generalize enough to draw conclusions  as to how one traveler would respond to
     changes over time.  Although this is how almost all mode choice models are presently calibrated,
     there is some sentiment among planners that it is an insufficient basis for identifying true traveler
     sensitivity to pricing.

     One alternative which is gaining popularity is the panel survey,  which follows a fixed  group of
     households and individuals over time and attempts to measure changes in their travel behavior in
     response to actual changes in their travel conditions. In theory, this should produce more accurate
     cost sensitivities, but in practice such surveys  are very expensive and difficult to conduct. Worse,
     they generally require 18-24 months or more, to  allow for three  or more survey "waves".  Still,
     such surveys have been conducted in Seattle, Washington and Montgomery County, Maryland and
     hold considerable promise for the future.

     Another technique that may be useful is the stated preference survey, in which people are asked
     how they would respond to a series of hypothetical situations.  Such  surveys are not as reliable as
     surveys of observed behavior, because of the large gap between what  people say they will do
     (especially under hypothesized conditions which may be far outside  their normal experience) and
     what they actually do. Still, if performed under carefully controlled conditions, stated preference
     surveys can produce estimates of pricing sensitivity for travel decisions, when no other  option is
     available.

     Some mode choice models have been developed by transferring  model coefficients from another
     urban area.  The disaggregate nature (i.e., modeling individual  choices) of logit mode choice
     models generally supports this. Analyses of coefficients from different areas indicates that the
     variation in these coefficients is not terribly great, suggesting that careful transference is  possible.
     Perhaps the most critical issue here is to be aware of how the independent variables were developed
     for the original model and to maintain consistency in the new model.

     Another major trap  to avoid:  in transferring pricing coefficients, one must be aware of the year for
     which the original model was developed and adjust such coefficients for inflation as needed.  For
     example, if the original model was calibrated using 1980 prices,  then the new model must either
     also convert  its prices to 1980 year dollars, or adjust the original cost coefficients to reflect the
     effect of inflation over the years.   For example,  the Case Study Model  uses prices expressed in
     1980 dollars. To convert its operating cost coefficient (0.004) to 1993 dollars, it must be multiplied
     by 0.57, which is the ratio of Urban Consumer Price Index values  for 1980 and 1993, resulting in  a
     1993 equivalent value of 0.0023.

A-38                                                               U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                         ,                            to Reduce Transportation Emissions


      In some cases it might not be possible to derive a particular cost sensitivity from any of the above
      sources. In the  last resort, it will generally be acceptable to synthesize a coefficient value and then
      perform a series of sensitivity tests on it. .The value of the parameter would be varied above and
      below the hypothesized value, the rtiodel would be re-applied, and the results (mode choice, trip
      length, VMT, etc.) would be charted  for each variation.  Comparing this variation with theoretical
      values  and a great deal of common sense from a variety of external reviewers  can often  help
      identify an acceptable value for a parameter.  Although such an analysis, even though carefully
      reasoned and conducted, cannot conclusively identify the proper value for an unknown parameter,
      it may provide a suitable interim result until better information can be found.


      A.4.2    Model Development/Application

      In revising travel models  to  be  more sensitive to pricing, several development and application
      issues must be considered, including the following:                                    '

      Connectivity and Feedback:  Asrthe  Case Study Model demonstrates, in order for pricing to affect
      each simulated  travel choice, the model set must be  "connected"; i.e., air travel prices must'
      influence each, model step. In a four-step model, one implication of this is that the model must be
      applied "backwards" and  "forwards", as in  the  Case Study Model.  This increases  model
      application  time and resource requirements (e.g., hard disk space) but is probably unavoidable
      within the current sequential  model structures.  Advanced programming, such as  combining the
      mode choice and distribution steps, can minimize this, but only at the cost of increased overall
      complexity.                              '

      A related issue is the question of achieving model equilibrium, usually defined as the consistency
      of travel speeds  throughout the entire  model.  It is unclear if or how these improvements to handle.
      pricing  would affect the potential and the need to iterate the model set until speed equilibrium  is
      achieved. It may be that price-sensitive models need to use a different definition of equilibrium,
      possibly one  that  involves achieving  consistency  of  equivalent prices  throughout the  model
      (equivalent price being defined as actual price plus the traveler's cost of time).

      Effects  of Inflation:  As  noted above, the. sensitivity of travelers to price,  as reflected  in the
      model's coefficients,  is expected  to change over time  with the general rate of inflation.  Price
      coefficients that are borrowed from  other areas (or from different years)  must be'adjusted to
      account for the effects of inflation. This means that cost coefficients can be expected to decrease
      numerically.over time, as the purchasing power of money erodes, more or less steadily, over time.

      Extreme care  must be taken in forecasting price information.  It is commonly assumed that all
      travel prices will rise in the future at the same rate as inflation, meaning that there is no change in
      the relative price between travel choices.. This is not always true.  For example, transit fares in
      many cities  have not kept pace with inflation. If a forecast assumes that that trend will continue
      into the future, then that must be reflected in reduced fare values for a future model run.  Similarly,
      as Figure A-5 shows,  the trend in  automobile operating  cost has been dropping since after World
      War II, when expressed in 1994 dollars, and the trend is for continued declines in real dollar terms.
U..S. Environmental Protection Agency                                                                A S

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
                                          Figure A-5
                           Trends in Passenger Car Operating Cost

                              Vehicle Operating Cost Trend and Forecast
18.00 .
«, 16.00 •
« 14.00 .
§ 12.00 .
te
S" 10.00 .
I
<•> 8.00 .
6.00 -
4.00 -









\
^\'
.







A
- \
\/








X

v^ —







/v
f \
^c
""-






	 	 „

•---..






— — 	 ••"'



           1950
1960
                                 1970
                                           1980
                                                      1990
                                                                 2000
                                                                            2010
                                                                                       2020
             Historical
                                  	Forecast: DOE gasoline price- •
                                                 Forecast: gasoline price trend
     Different Strategy Specifications:  -Some market-based strategies have many different variations
     which affect how the strategy is implemented and may affect how it is perceived by the public.
     However, it is unlikely that any travel model will be sophisticated enough to be sensitive to such
     nuances. It is enough of a challenge to create models which are properly sensitive to a change in
     actual travel price, much less be influenced by the context in which the change is implemented.

     For example, two ways of increasing the user cost of driving are: 1) increase the gasoline tax, and
     2) convert the cost of motor vehicle insurance from a semi-annual payment to a pay-at-the-pump
     fee, as has been proposed in California.  Although these two measures are quite different from an
     implementation  perspective, it is questionable if motorists would perceive them as different (as
     long as the increased price per gallon was the same).  It would seem reasonable to estimate their
     quantitative effects on travel in the same manner.

     Model Refinement/Development:   Some of the enhancements described above can be made fairly
     easily, while others require considerable effort. An example of the former would be the Case Study
     Model feature which permits the separate consideration of LOV and  HOV  parking price, thus
     allowing the testing of differential parking price by zone by vehicle occupancy.  It would be very
     easy to incorporate this into a mode  choice model which already uses parking  price and splits trips
     by occupancy level, which quite a few mode choice models already do.  This would require  only a
     few hours of programming to revise the mode choice  application procedure; no new calibration
     would be required.

     At the other end of the spectrum, modifying the distribution model to use composite impedance is a
     much more significant task, requiring the development of a  new program (or modification of an
A-40
                                                                  U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                       •  •       '      '               '                to Reduce transportation Emissions

      existing one) to calculate composite impedance and the complete re-calibration of the distribution
      models to use the new impedance value. This could take from a few weeks to a few months.

      It is important to recognize that the improvement of the pricing sensitivity of most existing travel
      models  is a difficult undertaking which requires time, data,  computing resources, and staff
      expertise which are not universally available in local planning agencies. Such a task is made all the
      more difficult since agencies are usually expected to pursue this work in addition to their usual on-
      going planning and  forecasting duties.  One way to organize the workload is to establish formal
      parallel "tracks", one for on-going work  and another for new model development, to ensure that
      neither effort is neglected.                 ,

      Of course, agencies  which are in the process of updating their models or developing new models
      anyway should take all reasonable steps to improve the sensitivity of their models to pricing. The
      features  of the Case  Study Model and  the enhancements  described in Section A.5 should be
      strongly considered. Once a commitment is, made to develop new models, the incremental effort to
      make such models reasonably price-sensitive is not very large.

      A.5      Summary

      A.5.1    Summary of Desirable Model Features


      The analyses of this appendix lead to the following major conclusions:

               *  Improve the Four-Step Process: Although many current four-step models are now
                  largely inadequate for modeling most pricing measures, this process can be modified
                  to improve its sensitivity to pricing strategies,  as illustrated in this appendix.  Such
                  modifications will address most of the  criticisms of this  process and enable  it to
                  perform  adequately until the next decade, at which time a new model structure may
                  be sufficiently developed and tested to serve as a replacement.

               •  Improvements are Not Easy: A few of the model enhancements described above are
                  fairly simple, quick, and  inexpensive but most of them  are not.   Most of  these
                  changes will require a recalibration of a region's entire travel model chain, which is
                  usually a multi-year effort  with a budget in the high six figures.  If a  region is
                  planning to  pursue such an  effort anyway, then incorporating these improvements
                  does not entail a significant incremental investment of resources.

               *   Improvements are Necessary Anvwav: Most of the model  enhancements discussed
                  here would help improve the sensitivity and credibility of travel forecasts in contexts
                  other than pricing strategies..  For example, analyses of TCMs such as changes in
                  transit supply, land use, ridesharing policy, and HOV lanes would benefit from these
                  enhancements. Thus, there are several reasons to enhance existing models.
U. S.. Environmental Protection Agency                                                               ,  .,

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
                    Recognize the Unknown Sensitivities: The fact that the sensitivity of certain travel
                    choices to some  pricing measures is still largely unknown  cannot be used as an
                    excuse to eschew their analysis.  Additional research, transferable parameters, and
                    sensitivity analyses can be used to make a reasoned estimate, leaving room for future
                    revisions to model results when better information is forthcoming.
A-42
                                                                     U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                                     to Reduce Transportation Emissions
       References
        l.
       •2.
       3.
       5.
       6.
       7.
 W.  Allen,  User's  Guide  for  the  MWCOG Mode  Choice Application  Program
 Microcomputer  Version, for the Metropolitan Washington Council of Governments, June


 J. Heisler et al., Estimating Toll Diversion Using Existing Transportation  Planning
 Software, presented at the Second Conference on Application of Transportation Plannine
 Methods, April 1989.

 The Phase III Travel Demand Forecasting Model: A Summary of Inputs, Algorithms  and
 Coefficients, METRO, Portland, Oregon, January 1994.

 W. Allen, Model Calibration for Employee Commute Options Planning, presented at the
 Fifth Conference on Application of Transportation Planning Methods, April 1995.

 K.  Train,  Qualitative Choice Analysis,  Theory, Econometrics, and an Application to
Automobile Demand, The MIT Press, Cambridge, Massachusetts, 1986.

 Greig Harvey and E. Deakin,  A Manual of Transportation-Air Quality Modeling for
Metropolitan Planning Organizations, for the National Association of Regional Councils
Revised Draft, November 1992.                                                     '

B. Spear, New Approaches to Travel Forecasting Models: A Synthesis of Four Research
Proposals,  Travel Model Improvement Program, John A. Volpe National Transportation
Systems  Center  for the. U.S.  Department  of Transportation,  and U.S. Environmental
Protection Agency, January 1994.

W. Allen and G. Davies, A New Method for Estimating Cold Start VMT, presented at the
Fourth Conference on Application of Transportation Planning Methods, May 1993.
. U. S. Environmental Protection Agency
                                                                                            A-43

-------

-------
                                                        Technical Methods for Analyzing Pricing Measures
                                                                   to Reduce Transportation Emissions
 Appendix  B


      The STEP Analysis Package: Description and Application Examples


       B.I    Overview

     This  appendix  discusses  STEP, a  travel demand modeling package  designed  for  planning
     applications and policy analysis, and shows how the model was used to analyze a wide range of
     transportation pricing measures for four California metropolitan areas:   the San Francisco Bay
     Area, Los Angeles, San Diego, and Sacramento.  We begin with a description of STEP then
     discuss how we used the STEP models to examine the potential impacts of pricing strategies on
     travel behavior, traffic volumes, energy consumption, and emissions for two analysis years 1991
     and 2Q10.   We also briefly  describe other  methods  used to supplement the STEP analyses
     including estimates of the elasticity of vehicle fuel economy with respect to fuel price, and detailed
     regional highway network models. A series of tables presentthe analysis results.


     B.2      Introduction to the STEP Package

     STEP is a travel demand analysis package composed of an  integrated set of travel demand and
     activity analysis models, supplemented by a variety of impact analysis capabilities and a simple
     model of transportation  supply. STEP is based on microsimulation - a modeling technique which
     uses the individual or household as the basic unit of analysis rather than dealing with population
     averages, (cf. Orcutt,  1976).  STEP results are aggregated only after the individual or household
     analyses are completed, allowing the  user great flexibility in  specifying output categories. This is'
     more commonly referred to as "sample  enumeration" in the literature,  or as "discrete choice
     analysis."

     STEP has been applied in a number of Bay Area studies over the years, and has been adapted for
     use  in studies  in  Los Angeles, Sacramento, Chicago, and the Puget Sound region  (Seattle).
     Applications can proceed with model re-estimation specifically for the  region - essentially,  by
     creating a completely new set of models for STEP - but to date nearly all applications outside the
     Bay Area have relied on extensive re-calibration of the default (Bay Area) models plus a limited
     amount of re-estimation as needed to match local conditions.

     Several features of STEP supported  its choice as the  basic modeling tool for the case studies
     presented here. STEP's regional, subarea, and corridor-level analysis capabilities fit well with the
     scope  and scale of the policies under consideration.   Its model formulations can represent a
     comprehensive set of possible price effects, and its models display linkages consistent with travel
     behavior and pricing theory. Its use of microsimulation makes it possible to address many  of the
     questions about equity and the distribution of impacts that frequently arise in debates about pricing.
J. S. Environmental Protection Agency                               •   '                           B 1

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions

      Finally, it is far faster to calibrate STEP for a region than to upgrade the regional models to include
     pricing variables, and far faster and less expensive to run STEP than to apply regional models.

     STEP'S data  analysis  capability  is  another important  asset  in  pricing  studies.    STEP'S
     microsimulation formulation permits the package to be used as  a survey  tabulation  technique
     employing sophisticated data transforms and linkages.  For example, many travel surveys contain
     detailed information about the vehicles each household owns and indicate which vehicle was used
     for each trip made on the  survey day(s).  Using STEP, these vehicle data can be tabulated so that
     exact usage  patterns by model year or vehicle type can be determined.  They also can be related to
     personal and household characteristics to yield useful information  about low-income households'
     dependence  on old vehicles and their contributions to vehicular emissions.

     STEP itself was originally developed for sketch planning analyses in the San Francisco Bay Area
     (Harvey, 1978).  Since that time, all of the models in STEP have been completely re-estimated and
     additional models addressing location choice, time-of-day of travel choice, and congestion effects
     have been added.  The most recent formulations are nested logit.  A number of versions of STEP
     are currently available, including options that permit the analysis of activity data as  well as  travel
     data, and versions that use either MOBILE or California EMFAC emissions data.

     STEP'S models  are applied  using  actual  or  forecast data  on  household  socioeconomic
     characteristics, the  spatial  distribution  of population  and  employment  ("land use"), and
     transportation system  characteristics  for  the selected  analysis year(s).   The  socioeconomic
     characteristics of a sample of households and its members are usually taken from a regional  travel
     survey or from the U.S. Census Public Use Microdata Sample (PUMS).  Population, number of
     households,  and employment by category (type) are taken from the regional " land use" data base.
     Transportation level-of-service data (times and costs) are derived from  the region's travel model
     system. The land use data are provided to STEP for  subareas (which could be zones, districts, or
     corridors) and for the region as a whole; the level-of-service data are provided in the form of large
     matrices of  interzonal times and costs. STEP then reads through the household sample, attaching
     level-of-service and land  use data to each household record as necessary.  For each household,
     STEP uses  its models to predict a daily travel and  activity pattern  for each individual in  the
     household.  Finally, household travel is summed up and household totals  are expanded to represent
     the population as a whole.
                                      .'!'"''             .'        I                 ',
     STEP can analyze  any  change in the  population or  in the transportation system that  1) can  be
     represented  in terms of the variables in its models and 2) can  be  associated with  a specific
     geographic area or grouping of households. Testing the effect of a change in conditions or policies
     is a simple matter of re-analyzing the household sample using the new data values, and comparing
     the results with previous  outputs.  For example, a new highway  or  new transit service can  be
     represented  by changed  travel times and costs for the areas served; a parking  price increase can be
     represented  by an increase in out-of-pocket costs; an increase in income in a particular area or for a
     particular population subgroup can be represented by editing the household file to incorporate the
     revised incomes.   Along similar lines, future years  can be  represented through proportional
     factoring and re-weighting of survey observations to reflect expected regional trends, or can  be
     based upon  a more sophisticated microsimulation of household changes based on cohort survival
     and other methods of demographic forecasting.
B-2                                              •            ,     U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                              .   .                     to Reduce Transportation Emissions

      The  sampling framework  preserves the richness of the underlying distribution of population
      characteristics and permits  tabulation  by  any  subgroup  with  sufficient observations to  be
      statistically significant.  For example, the results  can be disaggregated by income level and age
      which would allow an assessment of effects for, say, various income quintiles among the retired
      population. This is a significant advantage over an aggregate model, which uses zonal averages for
      most socioeconomic-economic data.

      STEP maintains its quick response capability while achieving great detail in representing behavior
      in part by reducing its  detail in  representing transportation networks. STEP does not  have  an
      internal transportation network representation and traffic assignment model, so changes in level of
      service resulting from changes in demand must be calculated in another way.  Both an approximate
      method and a more detailed and conventional network modeling approach have been developed for
      this purpose.  ,                 ,                                      ,

      To approximate the effects of changes in demand on network performance and vice versa, a simple
      routine for estimating level-of-service was incorporated into STEP  in the early 1980s'(Harvey,
      1993).  The simplified level of service model uses peak and off-peak travel times and base  case
      demand estimates to calibrate a supply function for appropriate spatial groupings of trips (i.e., trips
      in broadly-defined  "corridors").  The basic form of this equation is: t=a*(l-f-[V/C]Ab), where t is
      the travel time in minutes per mile; V is1 the volume in vehicles per  hour; C is the "capacity"  in
      vehicles per hour; and a'and b are coefficients fit to each corridor, For each change in demand, the
      calibrated function  can be used to compute a new "equilibrium" in the corridor (This is generally
      known as the Bureau of Public Roads function for determining travel time on congested links).

      While the simplified level of service model is useful for many analyses, it is intended  only as an
      approximation of changes in network performance and  is likely to be inadequate in cases where
      large network perturbations  could occur or where specific route choice changes are at issue. When
      network questions are critical, STEP must be used in conjunction with a more detailed  network
      model.

      In the typical application,  STEP  is "interfaced"  wjth the  region's  detailed highway network.
      STEP'S modal trip outputs are summarized on a district-to-district basis (a district  is defined as an
      aggregate of the zones for which land use data are reported; for example, in the Los Angeles region
      there are 1555 zones and 55 districts defined/by the regional agency).  If the policy under analysis
      results in  any  significant  differences from  the   base-case district-to-district trip tables,  the
      differences,are used to factor the  zone-to-zone trip tables in the aggregate model system.'  The
      network models are then run using these new trip tables, and the results are fed back into STEP as a
      revised set of level  of service inputs. Iterations continue much as is done in a conventional travel
      model system until an acceptable  level of convergence is achieved.   Transit networks also may
      need to be run in conjunction with STEP in cases producing significant differences in highway
      travel times of a sort likely to affect bus operations.

      For certain transportation pricing measures, such as proposals to toll specific links  or facilities  in a
      network, use of the detailed network models together with  STEP is pf particular importance. For
     the case studies presented here, we used the network models for Los Angeles and the Bay Area to
     test the route choice effects  of congestion pricing, interfacing in the riianner described above with
     the versions of STEP developed for each region.                      '
U. S. Environmental Protection Agency
                                                                                               _

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions

     Three major variants of the STEP model system are shown in Figures B-la, b, and c; Figure B-lb
     shows the version used in our pricing studies. The basic data requirements of the STEP model are
     summarized in Figure B-2.  A typical sequence of activities for a STEP application is shown in
     Figure B-3.
                                         Figure B.la:
                        STEP Model Structure with Basic MTC Models
                      Household Characteristics Dependent on Travel:
                           Number of Autos Owned
                      Daily Household Trip Choices:
                          Trip Frequency (HBW, HBS, HBO, NHB)
                          Trip Destination (BS, HBO, NHB)
                          Trip Mode Choice (HBW, HBS, HBO, NHB)
                                  Note:
                                   HBW - Home-Based Work Trips
                                   HBS - Home-Based Shopping Trips
                                   HBO - Home-Based Other Trips
                                   NHB - Non-Home Based Trips
B-4
                                                               U. S. Environmental Protection Agency

-------
                                                    Technical Methods for Analyzing Pricing Measures
                                                               to Reduce Transportation Emissions
                                    Figure B.lb:
             STEP Model Structure with Enhanced MTC Models
                 Major Household Location Choices:
                 Residential Location
                 Primary Workplace Location for Each Worker
                 Household Characteristics Dependent on Travel:
                 Number of Autos Owned
                 Daily Household Trip Choices:
                 Trip Frequency (HBW, HBS, HBO, NHB)
                 Trip Destination (HBS, HBO, NHB)
                 Trip Mode Choice (HBW, HBS, HBO, NHB)
                 Time Characteristics of Household Travel:
                 Work Arrival Time
                 Transportation System Performance:
                 Highway Corridor Delay
                                Note:
                                HBW - Home-Based Work Trips
                                HBS - Home-Based Shopping Trips
                                HBO - Home-Eased Other Trips
                                NHB - Non-Home Based Trips
. Environmental Protection Agency
                                                                                    B-5

-------
   l
•a1 {§
I
              •i
              T3
              O
              §
              •s
              o
              S





              I
               s
              •«-»
              CO

              ~8
              TJ
               O
              CO
              n

w


                                                                                                                                     I
                                                                >>    o o < ^


                                                               f    S  2  2 c
                                                                C    >  O  (U O
                                                               ia    ^ xi -o ™
                                                               2    I  I  I S

                                                    •g'i
Working-Age Adults i
Not Employed •
1 Is
Non-Work Chains I
Number of Chains . m
Number of Aclivilies H
ILocalion of Activities M
Mode 1
1
me
Pa
   2     u


   111

   'SPl


Itlt
1/0 3  M  fc1

1111

££££
                                                                 1
                                                                 ce
                                                                 T3
                                                                 §
                                                                              _ -a
                                           u
                                          T3
                                           O
                                        E  °

                                        ttl
                                                                              -
                                                                               o ta
                                                                               O o
                                                                              -C O

                                                                              ,55 j

                                                            E    ~ ™  >> .2 J3  .2
                                                           -—    -O -^;  H  en e  M
                                                                       s
                                                                       CO

                                                                      .
                                                        CO ^"S  o o ^

                                                           13  2  
                                                                                    Q
                                                                   3o
                                                                   z 2
It
•s s
 I^J Q


? S


II
S a
                                                 £,  a>


                                                  1?


                                                 Il
                                                 & ?
                                                     *•
                                                  I-a

                                                     '
                                                                                                                                     CD

-------
                                               Technical Methods for Analyzing Pricing Measures
                                                        to Reduce Transportation Emissions
                                   Figure B.2:
                      Primary STEP Data Requirements
              Basic Data:
               Regional Household Travel Survey
               1990 US Census Public Use Microdata Sample
              For the Survey Year:
               Geography
                land area, population, housing stock
                for tracts, zones, and/or districts
               Network Level-of-Service
                highway, transit
                a.m. peak, p.m. peak, off-peak
                times, costs;
             For Each Forecast Scenario:
               Geography
                land area, population, housing stock
                for tracts, zones, and/or districts
               Network Level-of-Service
                highway, transit
                a.m. peak, p.m. peak, off-peak
                times, costs
               Economics
                expected real income growth
                expected real fuel price growth
U. S. Environmental Protection Agency
                                                                           B-7

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
                                 Figure B.3:
               Sequence of Activities for a STEP Application
     Prepare Survey Data for Initial Analysis:
      Screen Survey for Unusable Observations
      Reweight Survey to Match Key Census Demographic Characteristics
      Reformat Network and Geographic Data to Match Database Requirements
      Assemble and Test Database
     Calibrate the STEP Models:
      Run STEP for Base Conditions
      Compare STEP Calculations with Actual Household Travel Patterns
      Adjust Constants, Beginning with Upper-Level Models, and Rerun STEP
      Iterate the Adjustment Process Until the Overall Fit is Acceptable
      Prepare STEP for the Forecast Scenario:
        Adjust Household Data to Reflect Changed Conditions
         Income
         Subarea Population
         Household Type Cohorts
       Reformat Network and Geographic Data to Match Database Requirements
       Assemble and Test Database
       Run STEP to Create a Base Case
      Test the Policy Alternative(s) with STEP:
       Alter the Database as Necessary to Represent the Policy Option
       Run STEP to Estimate the Effects of the Policy Option
       Post-Process the STEP Outputs
       Repeat the Analysis Sequence for Variants of the Policy Option
B-8
                                                    U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                             '   ..                            •        to Reduce Transportation Emissions

      B.2.1    Transferring STEP Models to Other Regions

      Although each application of STEP could utilize models estimated specifically for the region being
      studied, a less costly approach is to transfer models estimated in one region to ahother. In the case
      studies presented here, STEP models originally estimated for the Bay Area were transferred to Los
      Angeles, Sacramento, and Sail Diego, with detailed calibrations  and a moderate amount of model
      re-estimation in each case.                   ,
                                               x
      Procedures for transferring models and evaluating their performance are well established - in fact,
      many regions routinely use one or more transferred models in their regional model systems The
      procedure for transferring. STEP to a new region follows  much, the,same.general sequence of
      actions and so will be discussed only briefly here.

      To transfer STEP to a new region, the required data first must be set up. The region's 'most, recent
      household travel survey  is obtained, and checked (incomplete  observations are excluded), and
      network data and land use data for the year of the survey are extracted from the regional modeling
      data bases. The data are then linked and a trial simulation Is carried out to determine how closely
      the models to  be transferred match the actual travel patterns in the survey data. Invariably, a
      sequence of adjustments to model constants (and sometimes to a small number of coefficients) is
      necessary to  achieve an acceptable replication of the base travel pattern. These adjustments serve
      both to capture actual differences in behavior and to compensate for variation in the way  regional
      planning agencies define certain variables such as transit wait times, income ranges, and specific
      categories of land use.                                                     .

      Once an acceptable simulation of the survey year,(the '^base case") has been  obtained in this
      fashion, STEP  should closely reflect travel conditions and behaviors in the  region to which it is
      being transferred, and consequently can be used with local data and forecasts for the full range of
      modeling applications.
     B.3      Applying STEP to Pricing Measures

     B.3.1     Overview

 x    The application of travel forecasting models to specific pricing policies is rarely a straightforward
    " matter.  In nearly every case, both the models themselves and the available data, bases impose some
     limits on the policies that can be tested. For example, the regional transportation data bases (and
     models based on the data bases) typically lack information about the variation of parking price in
     each zone, and may have only approximate information about the vehicle Used for a specific trip.
     In cases where such details would play a large role in determining the impact of a policy being
     studied, only an approximate estimate of the policy's effects can be formally estimated through
     modeling: the analyst must devise a means of representing the policy as well as possible given the
     models and data, and must be  prepared to make off-line calculations and adjustments to improve
     the realism of the analysis, or to do further analyses after gathering additional, information.

    •Some discussion of implementation scenarios  is necessary simply to  determine  how a' proposed
     pricing concept should be analyzed; clearly, however, much more attention to specifics would be
     needed in an actual implementation.   In our  analyses,  for example,  we  implicitly assume that
     evasion  or outright fraud would be insignificant, hence the measures would be fully effective as

U. S. Environmental Protection Agency                                                   ,   . •

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
      proposed.  For most transportation pricing measures, monitoring, enforcement, and audits would be
      needed to assure that.
                                       "          ',•.•'    -          i  •            .      , .  •
      In  our case studies we  found that  it generally was  possible to define transportation pricing
      strategies in ways that were tractable from an analysis perspective and  also yielded information
      which is helpful in  thinking about policies as they might actually be implemented.  The use of
      advanced modeling capabilities, along with  the availability  of good data,  made it possible to
      explore behaviors that often would be omitted from a more conventional analysis. Nevertheless,
      the analyses did require a number of assumptions,  and they have certain limitations that must be
      acknowledged and taken into consideration in policy evaluations.
                   ';  .      .,	  ,     , ,  ,    '   •        ,';,'_        •»   I '                ,
      The following sections detail how the pricing concepts analyzed in our case studies were specified
      and analyzed.  In each case, the underlying rationale for the  pricing  concept is stated,  a specific
      pricing measure is defined, modeling assumptions  to represent the pricing measure are outlined,
      and key implications of the assumptions are noted.
                                          • ,  .       "         .            i
     B.3.2    Congestion Pricing

     Congestion occurs in the highway system when more vehicles attempt to traverse a segment of road
     per unit of time than that segment can accommodate!   Such a location is called a bottleneck.
     Congestion pricing builds on the simple realization that travelers are sensitive to the cost of travel;
     a fee levied at a bottleneck will divert some vehicles from .the traffic  stream, reducing congestion.
     The diversion  of a specific vehicle  might  be to a different  route, time-of-travel, mode,  or
     destination; it could reflect a trip foregone; or, over the long run, it might follow from a change in
     residence or workplace location.

     Two  major design issues arise in thinking -about  how to use pricing to manage congestion at a
     bottleneck:

           •   Price level - Price can be varied over a wide range to achieve different levels of traffic
               improvement.  Economic theory tells us that price should be set to reflect the social cost
               caused by the  marginal user at a bottleneck, less the average variable cost already paid
               by users.  While this  should be the clear goal of any congestion pricing application,
               considerations of implementation and management  ease  may point toward a simpler
               price criterion  based, for example, on achieving and maintaining a conventional level of
               service measure from the literature of traffic engineering.  We know that the "optimal"
               level of congestion reduction will be unique at each bottleneck, but it is much easier to
               explain  a generally applicable congestion reduction goal in the policy-making process,
               and easier to implement and manage facilities based on observed performance.  Hence,
               the actual criterion for setting the  congestion  price may well be framed  in terms  of
               standard traffic  leyel-of-service metrics (e.g.,  B, C, D,  E).  For similar reasons  of
               simplicity and clarity, specific prices might be chosen to reduce the amount of change-
               making required (rounded to the nearest 25 cents  or to the nearest dollar, for instance),
               although with modem road pricing technologies  this would not be strictly necessary.
               Periodic adjustments in price are likely to be needed to maintain effectiveness, and they
               too would likely be done  in simple, rounded increments of 25 cents  or a dollar, unless
               electronic toll collection were in place.
B-10                                                               U. S. Environmental Protection Agency

-------
                                                            Technical Methods for Analyzing Pricing Measures
                                                                       to Reduce Transportation Emissions

            •   Period of application - Some economists have argued forcefully that congestion prices
                should change dynamically in response to traffic conditions,  perhaps. varying from
                minute to  minute  to  achieve the optimal  reduction  in  congestion.  However,  few
                seriously believe that such a dynamic scheme would be implemented any time soon, for
                several reasons: 1) the practical difficulties of creating, testing,  and maintaining'the
                hardware and software required for such a system; 2) the unresolved theoretical question
       '         of whether a truly dynamic system would produce a stable set of prices; 3) the  strong
                revealed  preference .of travelers for  predictable conditions, even  if the  price of
                predictability is a somewhat higher average time or cost; and 4) the question of how to
                treat incident-related  delay in a  dynamic pricing environment  An  initial congestion
                pricing scheme more likely would involve prices that can be. explained through relatively
                simple signage and do not vary from day-to-day (though weekend-weekday and seasonal
                variations might be both desirable and feasible). Hour-to-hour variation might, however,
                be used to avoid large price increases and decreases,at the peak /  off-peak boundaries,'
                and might be designed as a pyramid of prices centered on each peak hour in order to be
                relatively easy for the driver to remember.

      In addition to these basic design issues for pricing at a bottleneck, there is a question about how
      widely congestion pricing would be applied in the highway network.  While pricing would be
      easiest to implement on litnited  access  facilities,  spillover from  priced  freeways to.unpriced
      arterials and collectors could be a problem in some locations. 1 Local communities seem unlikely
      to tolerate significant traffic  diversion to  the  facilities  under their jurisdiction, and could  be
      expected to oppose freeway pricing schemes if they created or worsened congestion on local roads.
      The localities might, however,  accept a broader-based pricing plan which manages traffic on a
      systemwide  basis, especially  if part of the revenues  were  returned to affected jurisdictions.
      Widespread  implementation of congestion pricing hence could mean pricing both freeways and
      parallel routes where significant delay appears.

      The  congestion  pricing  measures  tested  in  our case studies were designed to reflect  these
      observations about the policy  environment. We assumed that some form of electronic payment
      system would be used rather than toll booths, so that there would be no stopping to pay tolls.
      Prices were  applied everywhere delay appeared in the highway network (as  represented  in each
      region's model system - freeways and arterials plus some major collectors).  Price levels were set to
      reduce congestion to meet specific levels of service; we  investigated a range of level-of-service
      targets and eventually chose LOS  D/E for use in all four metropolitan areas.2  Our analyses
      allowed prices to vary by corridor, determined peak definitions by the extent of congestion in each
         The first US congestion pricing project opened in December 1995 on State Route 91 in Orange County
     California; the San Francisco-Oakland Bay Bridge is currently being studied as a second possible application!
      Because of the special characteristics of these two applications, spillover to arterials. is not likely to be a
     major issue. SR 91 pricing will apply only to the new lanes added in each direction, with the original lanes
     left unpriced; in the Bay Bridge case there are .essentially no realistic alternative highway routes.  The
     extension of pricing to other facilities such as I-1Q in the Los Angeles area or 1-80 in the Bay Area would,
     however, have to confront the possibility of spillovers to parallel routes.

        The choice of LOS D/E was based on analyses of benefit measures from the STEP model" which indicated
     that stable, near-capacity flows (about 10 percent below actual capacities) were'the most economically
     efficient traffic regime.  Specifically, we used delay reduction per marginal unit of price as the measure of
     benefit.
U. S. Environmental Protection Agency.            ,."~           B~lT

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
                            "           '••        '                     '•.:'.' I
     corridor, and permitted different prices to be charged in each corridor for each hour of the AM and
     PM peak periods, but we stopped short of dynamic pricing.  Rather, we assumed  that travelers
     would face a fairly simple schedule of prices by time of day, readily, comprehensible to travelers
     and influencing their travel behavior and location choices.^       -

     It is important to note that under this pricing approach, users of the facilities in greatest demand
     still would perceive traffic as heavy and somewhat constrained, with speeds below posted limits (at
     least for the cases considered here, higher speeds would not be as efficient from an economic point
     of view). Note also that we assume that prices would be maintained in constant dollars, meaning
     that from time, to time price adjustments might be necessary.
                  •":      '     ,"  '  • "  ;   '.'•''•   '  ,   "|l'"': •'-•,.    I '        ~  '       '      '   .
     The STEP analyses were  carried out by focusing on highway performance at the corridor level, as
     follows.  In the STEP calibration phase each of the metropolitan areas was divided into major
     corridors based on topography and highway function. Each district-to-district trip interchange was
     assigned to a corridor, and approximate volume-delay relationships (i.e., expressing travel time per
     mile as a function of volume and capacity) were developed for the corridors.^ This was carried out
     for both the AM and the PM peak in each region.
     3   We assumed congestion prices would be in effect on non-holiday weekdays only - 250 days a year.

     4   The shape of the volume/delay curve is a critical determinant of the outcome of the analysis, because it
     indicates how much traffic would have to be removed from the peak in order to achieve a given LOS. To
     represent volume/delay relationships for corridor-level changes, STEP uses an equation roughly analogous to
     the Bureau of Public Roads (BPR) equation relating link-level volumes and travel times. The STEP equation
     was initially developed in a study for the California Energy Commission and later re-estimated in studies for
     the Metropolitan Transportation Commission, the Southern California Association of Governments, and the
     Puget Sound Council of Governments (Seattle region). The equation expresses the relationship between the
     ratio of average peak to average off-peak travel times in each "corridor"  - basically a trip exchange - and the
     aggregate capacity serving that corridor. Separate estimations were done using data from the detailed highway
     networks of the three regions;  because the coefficients of all three models were nearly identical, a single
     equation was implemented in STEP. The specific functional form is t/t0 = 1 + (V/C)2.
                                                ' '         *       • .    '  '   !
     This corridor function, derived from regional network models, shows travel time climbing rather gradually as
     congestion builds. We know from highway operations research that the buildup of congestion for specific
     facilities is more abrupt and steeper in the region of capacity flows than this equation  indicates.  However,
     because the corridor function represents an aggregation of facilities of different types, it reflects the "family"
     of volume-delay relationships  for the freeways, arterials, and  major collectors embedded in the network
     models and producing their travel time estimates.

     The steepness of the buildup of congestion is important in determining what the congestion price would have
     to be.  If the slope is steeper than our equation indicates,  as it would be in a corridor with a single facility,
     congestion prices could be lower for a given  level-of-service  improvement than we report here. This is
     because a steeper slope implies that fewer vehicles would have to be priced off each corridor's facilities to
     achieve a given LOS.  We tested a number of functional forms  in STEP, and the different forms did indeed
     produce some variation in optimal prices. For example, letting the slope  parameter rise to 4, the value used in
     the standard Bureau of Public Roads (BPR) equation, would lower the "optimal" congestion price by about 40
     percent (regional average). Since the BPR curve is for a single  freeway facility,  it is much steeper than any
     corridor curve could be (unless the corridor consisted of a single freeway).  Therefore the BPR value should
     be viewed as an outer limit.
B-12                                                                   U. S. Environmental Protection Agency

-------
                                                             Technical Methods for Analyzing Pricing Measures
                                                                         to Reduce Transportation Emissions
       The level of service target was defined in terms of the volume delay function.  For the generic
       functional form used in this version of STEP, level-of-service D/E corresponds to a travel time that
       is about  85  percent longer than the time  under free-flow conditions.  In other words the target
       level-of-service was represented by a l.SS.ratio of peak to uncongested travel time in a corridor.

       In the Los Angeles region; about 300 aggregate " corridors" were defined in this manner, and about
       220 of them  - 73 percent - were sufficiently congested in the AM peak to justify congestion pricing
       For the  San Francisco Bay Area 150 corridors were  defined, with 90 (60 percent) meeting the
       cntena for pricing in the AM  peak.  San  Diego and Sacramento were both considerably less
       congested; only 15 percent of the 80 corridors analyzed in San Diego and 8 percent of the corridors
       analyzed in Sacramento were candidates for pricing.

       To estimate  the  price needed to achieve the target level of service, STEP was applied to  each
       sample of households and the  average price per mile was  adjusted on a corridor-by-corridor  basis
       until all corridors were at or below the 1.85 peak/off-peak travel time ratio, and no corridor had a
       higher congestion price than necessary.  This took approximately five iterations (model runs) for
       each region and each analysis year.                        .

    - For each region, a specific congestion price was estimated for each corridor and time period 5  For
       1991 conditions, the congestion prices would vary from zero (for the uncongested exchanges)  to as
      much  as  $1.00 per  mile for a very few  corridors,  such as the 1-80 corridor and the Bay  Bridge
      corridor in the San Francisco Bay Area and the 1-405 and HO corridors  in the West Los Angeles -
      Santa Monica area.  In San Diego, the highest corridor level prices would reach about 40 cents per
      mile, whereas in  Sacramento the highest corridor prices would be about 20 cents per mile  By the
      year 2010, congestion  is expected to worsen considerably in all four regions; many more corridors
      would be candidates for pricing, and prices would have to be higher to maintain the LOS  D/E
      target.0
          In the four case study regions, PM peak conditions are less sharply congested but last longer than AM
      peak conditions. Hence evening congestion prices, at least initially, could be somewhat lower but would be in
      effect for a somewhat longer period of time than those in the morning peak.  However, congestion pricing
      would flatten and spread out the AM peak somewhat, diminishing AM-PM differences in prices and hours of
      application.                     •                    ,

         One might ask whether the prices arrived at in this manner are the optimal prices.  The issue is not simple
      to resolve; in the first place it is well  understood that user-optimal  may not be identical to system-optimal
      (Wardrop,1952). User optimality is examined here, although we note in passing that pricing also could be
      used to achieve system rather than user optimality.   The analysis of user-optimal prices is particularly
      complex, because travelers can respond to pricing, in a.number of waysr shifting,trips among corridors and
      altering their frequency and times of travel. It is necessary to account for the possibility that travelers could
      switch to another route, travel at a different time of day, change modes, choose different destinations for some
      trips, increase or reduce the number of trips made, move to a different residence, or change their place of
      work.   STEP, accounts for these phenomena, but because STEP 'is a hybrid mix  of non-linear  demand
      functions  of various types, it is not possible to mathematically prove the existence of a unique set of
      congestion .prices for a given level-of-service criterion.  Simulation  offers an alternative  approach for
      assessing whether model results represent a stable and unique equilibrium, and we used it to investigate the -
      optimality of our corridor prices. We applied a number of procedures designed to determine whether STEP
      would  produce  different sets  of "optimal" congestion prices.  These  included adopting different search
      algorithms in the program code, and starting the searches from different initial corridor prices. All search
      strategies that produced stable outcomes were in agreement with the initial "optimal" prices,  which lends
      some support to the notion of a unique equilibrium.                                 "        '


U. S. Environmental Protection Agency                    '~~~      ~'       '.                  '   '.	g ,,

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
     Estimated  reductions  in travel time, VMT, trips, emissions, and  fuel  use resulting from the
     calculated congestion prices, as well as estimates of the total revenues generated, were calculated
     by summing up the analysis results for each corridor.  To simplify the presentation of price levels
     and provide an indicator of overall price impact, a corridor-weighted average price per mile is
     shown in the tables, and can be thought of as the average price peak period drivers would face
     overall.  It is not necessarily the price any individual traveler would  experience. For example, the
     price necessary to obtain LOS D/E on the San Francisco Bay Bridge in 1991  would have  been
     about $6, or 75 cents a mile for that corridor, in contrast to the average Bay Area AM peak price of
     about 9 cents a mile J

     Corridor-level  results are useful for preliminary planning  purposes,  but for implementation
     planning it is important to translate the results into specific facility charges.  Within the resources
     of this study, we were not able to test congestion pricing in a full network context for each of the
     four case study areas. Instead, we ran STEP for the four areas, then selected two corridors for more
     detailed analysis: 1-80 from the Carquinez Bridge to the Oakland-San Francisco Bay Bridge, and I-
     10 from Santa Monica to Downtown Los Angeles - two of the most congested locations of all those
     we studied. We ran regional network models for the Bay Area and Los Angeles to see how prices
     would need to vary among facilities  in the selected corridors, given the corridor prices and demand
     levels produced by STEP.  The Tranplan network analysis program was used, with an equilibrium
     traffic assignment for the AM peak hour and price incorporated into the route choice criterion.^
                                      '"'       i          „                 '
     Tranplan corridor analyses produce results comparable to STEP if  the per-mile price is applied
     equally across all facilities in the corridor. With the same price per mile on all alternate routes, the
     main effect will be a reduction of overall corridor demand rather than a rearrangement of traffic
     among corridor facilities (absent differential prices, traffic in  a congested corridor  will distribute
     itself such that all routes will have about the same travel times). However, Tranplan analysis made
     it possible to test link-by-link pricing to  more precisely target  bottlenecks in the system. We  went
     through five iterations in which we manually adjusted link prices in the two test corridors,  each
     time increasing the price on  links that remained congested and decreasing the price on links with
     better than D/E level-of-service and  then running Tranplan to evaluate link-level impacts.  Overall
     corridor delay  reduction tended to improve from iteration to  iteration, while overall price levies
     tended to fall.  After the .five iterations,  we judged that the effectiveness of congestion pricing, in
     terms of reduced delay per dollar, might be 10-15  percent  higher in these  corridors  than the
     approximate results of  the  STEP analyses would suggest.  This  should be  considered when
     reviewing the average prices and/or time savings presented in the tables.

     What if prices varied by location, but were set at modest prices initially and were increased only
     gradually to the levels necessary to avoid stop-and-go driving?  This approach would  give people a
     chance to adjust their travel and location behavior under prices that  accurately signal the ultimate
     7    The Bay Bridge congestion pricing  studies  underway at the time of this writing  are discussing
     considerably smaller prices, e.g., a $3.00 peak period toll.  A $3.00 toll in 1996-1997 dollars would be the
     equivalent of a $2.50 toll in 1991 dollars. Such a price increase would be sufficient to cut the queue at the toll
     plaza by about a third, but would not achieve LOS D/E.

     8    It was possible to use the network provided by the Southern California Association of Governments for
     this part of the analysis, but for the Bay Area a new Tranplan network was created as part of the study (MTC
     Uses UTPS networks and the study team did not'have access to this software).
B-14                                                                U. S. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                 ,                              '         .       . .  '    to Reduce Transportation Emissions

      spatial distribution of impacts.  Dynamic models would be necessary to fully explore the changes
      that such a pricing approach would produce over time;  STEP does not currently include such
      dynamic models.  However, STEP is able to evaluate lower-than-" optimal" congestion prices as
      would occur in a pricing phase-in (and perhaps in many cases where prices are set on political as
      well  as technical  grounds).  We tested the  impacts of lower prices by taking the final corridor
      congestion  prices  for the Bay Area and Los  Angeles and applying them in  10 percent increments
      (i.e.,  prices at 10 percent of optimal corridor prices, 20  percent of optimal prices, etc.). The STEP
      results, indicate that the shape of the aggregate demand curve is moderately convex, with slightly
      decreasing  effects for.each price increment. For each of the two  case  analyses, the first price
      increment of 10 percent produced almost twice the 'impact of the final increment of 10  percent.
      This suggests that implementing a constrained price can  still be reasonably effective.
                                                             i       i
      The STEP analyses are for scenarios in which pricing  is used to  manage congestion wherever  it
      occurs on  the  network of highways and  arterials;   how  congestion pricing would  wprk  if
      implemented on a few facilities is a different question.   Even if the ultimate objective is system-
      wide implementation, it is likely that-initial applications would be "spot pricing" - pricing applied
      to just a few facilities or corridors. As we discussed earlier, however, closely-parallel routes could
      receive significant amounts of diverted traffic if a single congested facility is priced; such traffic
      diversion could lead to significant congestion on the parallel routes; and opposition from affected
     jurisdictions might well be enough to halt implementation, unless the parallel routes can be priced
      as well.   -                                                    .

      Even where diversion to parallel routes is infeasible for most travelers, as is the case for the San
      Francisco-Oakland Bay Bridge, or where each facility in a corridor can be differentially priced, as
      our analyses x>f the 1-80, 1-405, and  I-10 corridors considered, a number of concerns about "spot
      pricing" remain.  For example, our analyses  indicate that implementation  at a single .highly-
      congested location or .in a single corridor will alter regional patterns of trip distribution, residential
      location, and workplace location, with  specific effects varying with household income level.  The
      result  of spot pricing could lead to a distortion of the spatial structure of the region, because the
      spot pricing leads  to exaggerated locational impacts.  Thus single  facility pricing may produce  a
      misleading view of the eventual areawide effects of congestion pricing.


     B.3.3     Employee Parking Charges

     In most metropolitan areas, parking is  commonly provided to its  users free of charge, although
     providing such parking can be quite expensive and presumably is recouped in other ways (e.g.,
     through  the prices  charged for  goods  and services, for private parking, or  through public  tax
     subsidies, for public parking).  Charging for parking, whether done through private initiative or in
     response to government incentives or mandates, would make the costs of parking more apparent to
     travelers and would likely reduce auto use somewhat.      ,

     Parking could be priced for all users, and sometimes is  (at many commercial garages, e.g., or by
    . local governments who install on-street meters).  However, proposals for the implementation of
     parking pricing  often focus on daytime employee parking, since the associated employee travel
     typically occurs  during the costly peak periods. If employees had to pay for parking, it is reasoned,
     they would  be  more likely to use  alternative  commute modes such as transit,  carpooling, or  '
     walking. In the case studies we present here we analyze only employee parking charges.
U., S. Environmental Protection Agency         •~~           ^"""~~~       '     -                   g ,-

-------
 Teffinical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions

      In comparison with congestion pricing, parking pricing is a relatively simple measure to analyze
      using STEP. The average zonal parking price (daily, for work trips, and hourly, for non-work trips)
      is a variable in  each  of the STEP  mode choice models, and zone-lever parking price data are
      available for each of the four metropolitan areas studied here.  Thus, any parking scenario that can
      be expressed as a change in an average zonal price can be analyzed using STEP.
                                              1111                   'i
      Proposed parking price changes  do  not always target the average zonal parking price, however.
      Consider a city in which a substantial amount (varying by zone) of the all-day parking is provided
      by a private operator, who  charges a daily fee for use. The operator, perhaps given an incentive by
      local or state tax policy, decides  to raise the fee by $1.00 per day.  To analyze the impact of this
      increase, it is necessary to have an estimate of the percent of all-day parking in each zone that is
      provided by this operator and hence will be affected by the increase.  A number of cities maintain a
      parking inventory which could provide this information, although many other cities would have to
      conduct a special survey to produce this estimate.

      Other parking pricing proposals  can be far more complicated to  analyze.  Consider a $3.00/day
      parking surcharge which applies only at employment sites with 100 or more employees.  In order to
      translate this surcharge into zonal average price estimates, we would  need information about the
      fraction of workers in each zone who work at sites with 100 or more employees.  We would need to
      account for the possibility that some of those employees do not provide any parking now, in order
      to figure out what share of each zone's employees would be subject, to the fee. The possibility that
      some employees could avoid a fee  at their workplace by parking elsewhere should already  be
      reflected in the calculation pf zonal average parking cost, but we also must consider the possibility
      that employers will simply pay the fee themselves rather than passing it on to the employee, again
      reducing the number  of affected workers (note that certain  implementation  strategies, such as
      treating parking as a taxable benefit or requiring the surcharge to be collected from the employee as
      a payroll deduction, would reduce the  likelihood and the impact of this latter concern). Very few
      cities have an employer and parking  data base organized to support such an analysis, and we have
      found none that has information  on  likely alternative  parking sites or on employer responses to
      such policies.  Hence, calculating the actual increase  in zonal average  parking charges that our
      surcharge would produce could require either a great deal of data "collection.  Nevertheless, for
      preliminary  planning  purposes it usually will suffice to  make  some  simple assumptions  in
      developing the data inputs or in interpreting the  results. For example, we could analyze the parking
      surcharge as if it applied to all employees and then factor the results downward to account for its
      more,restricted reach: if regional employment data indicate that only 40 percent of the region's jobs
      are provided by employers  with 100 or more workers, then our impact estimates should be reduced
      by about 60 percent.

      For  our  four  case studies, we utilized  parking  cost  data files developed by  the regional
      transportation agencies.  These files present  only the estimated average  employee parking price
      (nominal price) by zone. Given the  data we had available, we chose to model two general policy
      options: a flat daily charge  on all  employees who drive alone and do not currently pay for parking,
      as well  as a daily surcharge on all employee parking, paid or not. The first option could be thought
      of as a rough approximation of what prices might be like if free parking were no longer provided to
      employees; or it might be thought of as the result of a policy that imposes an impact fee or tax on
      free  employee parking but waives the fee on  parking that  is already priced  at or above some
      threshold level.  The second option  would be a flat impact fee (or tax,  depending on how it is
      structured and applied).
B-16                                                     .         U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                                     to Reduce Transportation Emissions
                 r       •        i                                    '
      Using STEP, a range of daily employee parking charges from $1.00 to $10.00 was examined for
      each of the four metropolitan areas.  To model the minimum price threshold option, drive-alone
      parking fees for ail workers in each sample were set to the specified minimum or to current levels,
      whichever, was higher - fees in zones  where existing zonal average parking fees exceeded the
      threshold charge were held constant. The second option we evaluated, a flat fee or surcharge on all"
      employee parking, was even easier to represent than the minimum price option; the fee was simply
      added to the employee parking price in effect in each zone.  In both analyses, we assumed that the
      employees would personally pay the parking charges (hence we treated the charges as out-of-
      pocket expenses). We also assumed that carpool and vanpools would be permitted to park for free
      at their destinations, and that no charges would b,e imposed for park-and-ride parking. These latter
      assumptions are generally consistent with the current treatment of HQVs. and park-and-ride in the
     , four case study regions.        '

      STEP accounts for the full set of travel effects we would expect parking pricing to have, including
      impacts on highway performance, but to verify that STEP'S simplified level-of-service functions
      provide an adequate representation of the latter, the peak period trip tables  from STEP were
      assigned using Tranplan to the relevant networks for Los Angeles  and the Bay Area, and  the
      resulting travel times' were  cycled back through the STEP  model.  No significant changes from
      STEP aggregate performance measures were identified.

      Results for $1.00 and $3.00 parking price increases are reported here. Given the ubiquitousness of
      free parking in  each of the.four regions, the differences between the two  policy options were
      minimal: the estimated impacts of the parking fees varied by 10 percent or less (i.e., a reduction of
      1 percent in VMT for the minimum price option, a  1.1 percent VMT reduction  for the surcharge).

      Our assumptions that prices would apply to all drive-alone vehicles? and that  HOV parking would
      be  exempt from  charges  maximize the  impact of-the .employee  parking fees.   In  actual
      implementation, a number of factors could reduce these impacts. For example, as our earlier
      discussion pointed out, exemptions of certain employers would reduce the number of employees in
      each zone who actually would pay a parking fee, with the impact varying widely among zones.

      In addition, in situations where parking is differentially available to or subsidized for different
      income or occupation groups, the impacts of price changes may vary  from those we have shown.
      Our results assume that a parking fee would be paid by all who drive  alone. But under some
      conditions  the fee might actually be absorbed by the employer; for example, some blue  collar
      workers have negotiated for free parking as part of their labor agreements, and a parking surcharge
      would have to be paid for by the employer or compensated through offsetting salary increases.  In
      cases such as these, the fee on parking could vary systematically with, income  group, and hence be
      disproportionate to the number of workers affected.

      Finally, the impact  of free parking for high-occupancy vehicles deserves special attention.  Free
      HOV parking is a common measure in our case study regions and might well be permitted under a
     policy to charge for parking; but it is not a necessary feature of the analysis.  If the parking fees
      apply equally to HOVs, HOV users still experience an advantage over solo drivers because they
      can split the cost among all passengers,'but the price differential between drive-alone  and HOV
         To calculate impacts on an annual basis, we assumed employee parking charges would apply 250 days a
     year.              .              .
U. S. Environmental Protection Agency                                                              B 17

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions

     decreases - by about 40 percent on average.  Based on STEP runs for all four metropolitan regions,
     this diminished advantage would cut the impact of the parking fee by  about 15 percent, because
     fewer current drivers would switch to HOV and some of those who currently are HOV users would
     decide to drive to work.
                  •          '     •• •      '       .     '        •      • • •   " i    .
                  •                  •                                    i
     B.3.4    Fuel Tax Increases
                                                                   •.

     A fuel tax increase would be a direct approach for reducing fuel consumption and also for reducing
     greenhouse gas emissions (because CC>2 emissions are proportional to fuel consumed). Its effects
     on other emissions and travel are muted, though still significant, because auto purchase decisions
     and usage patterns can lead to a more efficient vehicle fleet and reduced per-mile operating costs.

     The fuel tax increases  analyzed here are expressed as straightforward additions  to the at-the-pump
     price of gasoline and diesel fuel. For our base case, vehicle fleet fuel economy is about 22 miles
     per gallon (.0364 gallons per mile).  Base-case fuel cost is about $ 1.20 per gallon, or 5.45  cents per
     mile at average fuel economy. With no increase in fleet fuel  economy, a 50 cent per gallon fuel tax
     increase would add  about 2.3 cents and a $2.00 per gallon tax (or other form of price  increase)
     would add about 9.1 cents to the average per-mile cost of driving.  However, empirical  evidence
     and common sense  suggest that the in-use vehicle fleet would become more efficient under  a
     significant fuel price increase. In the many households with more than one car, household members
     could quickly arrange to make more use of their fuel-efficient vehicles and less use of their "gas
     guzzlers," cutting fuel consumption considerably.  Over time, both single-vehicle households and
     multi-vehicle households could be expected to increase  vehicle fuel efficiency as they replace some
     vehicles and retire others.

     How fast and to what degree such vehicle substitutions, replacements, and retirements might occur
     in response to fuel price increases has been a matter of considerable dispute. The issue is important
     to our analysis because it could significantly affect the impact  of a fuel tax.  Travel and location
     choices are undoubtedly affected by the costs of vehicle ownership and operation, i.e., by both the
     number of vehicles a household chooses to own and the type and age of its vehicle(s). Faced with
     higher fuel  costs, a household which  for  whatever  reason does not reduce its per-mile fuel
     consumption (by changing its vehicle holdings or changing which vehicles it uses most) will have
     to devote more of its income to fuel purchases, or take steps  to reduce its vehicular travel  (or some
     combination of the two).  If on the other hand the household finds it possible to reduce  the price
     effect through  vehicle substitution and replacement,  fuel efficiency improvements will have  a
     smaller effect on  travel.10

     STEP includes a model of the number of vehicles a household chooses to own, so we were able to
     capture the effects of fuel price increases on auto ownership  in our analyses.  However,  STEP
    . currently does not address the type or age of the vehicles owned, information which is needed to
     estimate the cost per mile under different fuel price scenarios.  We did not have direct access to  a
     model of household vehicle purchase decisions for this study, so to account for the broader range of
     10   A household's ability to change vehicle holdings is related to its current and expected income, its current
     vehicle holdings, ownership and operating costs of the alternatives, etc. The household's willingness to
     change its vehicle holdings depends on many additional factors, such as vehicle seating capacity, comfort,
     handling, and safety; fuel economy, an element of operating cost, is but one influence.
B-18                                                               U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                                     to Reduce Transportation Emissions

      impacts, we turned to outside sources for evidence on the elasticity of fleet fuel economy with
      respect to fuel price.                                             -

      The literature from the U.S. and abroad suggests that fleet fuel economy (miles per gallon) is quite
      sensitive to the price of fuel. Pickrell's recent research (Pickrell, 1993) and his syntheses for the
      Presidential Commission on Greenhouse Gas  Reduction (a group  known popularly  as  "Car
      Talk")(Pickrell, 1995) examine the impact of fuel prices and report findings from a wide range of
      reputable U.S. and international  studies in advanced economies.  He cites numerous estimates of
      long-run average elasticity of fleet fuel economy with respect to fuel price in the .5 - .6 range, with
      estimates as low as .2 to .3 and some higher than 1.0. An elasticity of 0.5 means that a 25 percent
      increase  in real fuel price (e.g:,  from $1.20 to $1.50) would increase, long run average fleet fuel
      economy from 22 miles per gallon (mpg) to almost 25 mpg; ,a 167 percent increase in  real fuel
      price (e.g., from $1.20 to $3, .20) would increase long run average fleet fuel economy from 22 mpg
      to about 40 mpg (82 percent).   A 40 mpg fleet average sounds high  for U.S. conditions, but it
      cannot be dismissed outof-hand, especially for a longer-term scenario (2010 or later) and/or one in
      which  the price,increase was implemented nationwide or  in a majority of urban states (so that
      manufacturers would have sufficient time and incentive to offer more fuel-efficient vehicles).

      Substantial fuel  economy improvements could,  in  fact, be obtained through shifts in consumer
      choices among the vehicles currently available for purchase:  for example, by purchasing the four
      cylinder rather than the six cylinder version of a midsize sedan, a consumer could  obtain a 10-15
      percent improvement in mpg. This percent increase in fuel economy is about what a 25-50 cents
      per gallon price increase would require, at a .5 elasticity. However, for large fuel price increases,
      an elasticity of .5 would imply that at least some consumers also would have, to change the type of
      vehicles they own and,use, i.e., greater numbers would  have to purchase and use highly  efficient
      vehicles and restrain their purchase and  use of the least efficient ones.  Currently over a dozen
      vehicles are sold in the U.S. which obtain over 40 mpg, so this seems technically, feasible, and may
      become more so if gradual improvements in technical efficiency, averaging perhaps 1-2 percent a
      year, are  forthcoming over the  next  decade or so, as many analysts expect (Pickrell, 1995).
      Whether buying habits in fact would change in the necessary fashion could be debated.

      For further evidence of how fuel prices might affect fleet composition and use, we turned to models
      of the  vehicle fleet.   Since our  case study regions were all in  California, we were particularly
      interested in an analysis tool known as the Personal Vehicle  Model (PVM), which the California
     Energy Commission  has used to estimate the composition of the state's vehicle fleet by size and
     age, as a function  of the price of fuel and other factors.11   We asked the CEC to provide some
      indication of the PVM elasticity  of fuel economy with respect to fuel price, as evidence  for
     California fleet conditions.  A run of the PVM made for this study by the CEC in January 1995
      indicated that a $2.00 fuel surcharge would lead to a 2 mpg  increase in fuel consumption (from 22
     to 24 mpg), for an average elasticity of .05.

     The PVM-estimated elasticity is much lower than the elasticities reported by Pickrell.  A partial
     reason  for the difference is that most national and international long-term elasticity estimates allow
     "    The PVM was developed more than a decade ago, and at the time of our study the CEC was engaged in
     a multi-million dollar project to replace it with an updated package based'on  new data and state-of-the-art
     modeling concepts. Hence we chose to treat the PVM as one source of evidence rather than to rely solely on
     it. '         .                   .      •     .,    .          .       .                   •   •
U. S. Environmental Protection Agency                               '    •                            B-19

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
                                                                     •,    i    '
      for changes in the products manufacturers offer in response to fuel price increases. In contrast, the
      PVM  analysis  assumed  that  the price  increase  would only  apply  in California,  and  that
      manufacturers would not increase the fuel economies of the cars they offer in response to a change
      in only one state, even a state as large as California. The PVM analysis does allow consumers to
      purchase more efficient vehicles from those  otherwise available. It does  not. consider increased
      relative use of the more fuel efficient vehicles within each'household's existing vehicle holdings.

      We discussed the fuel economy - fuel  price elasticity issue with a number of researchers and
      ultimately settled on testing a range of assumptions about the fleet response to fuel price, expressed
      in terms of the elasticity of fuel economy (miles per gallon) with respect to  price. Results for three
      elasticity levels are reported here:  0.5,  0.16, and 0.05.  The researchers we contacted felt (and we
      agreed) that the .05 PVM elasticity should be used as a lower boundary, and that a 0.5 elasticity,
      i.e., the lower end of the .5-.6 estimates from the national studies, was a reasonable upper boundary
      for a California-only policy. 12
                 1 " 	"     i      ' i            I,             ' '         , • '  •   'I      i                  '

      The fuel economy elasticities can be  used to compute average  mpg and  out-of-pocket vehicle
      operating costs per mile resulting from a fuel price  increase.  For example, consider a two dollar
      per gallon  increase, i.e., a fuel price of $3.20 per gallon.  In  comparison to the current $1.20  per
      gallon, for  which average out-of-pocket expenditure  is about 5.5 cents per mile, the estimated mpg
      and cents-per-mile costs would be:

               Elasticity              MPG                  Cents per Mile
               0.00                   22                     14.6
               0.05                   24                     13.3
               0.16               -28                     11.4
               0.50                   40                      8.0
                  :t •  .    ,     ••     , '    '.'•'•''':       ' ••:    ,  '•  "',   .  . •  •  •      •
      It is clear from this table why fleet response to fuel price is such an important issue. At a .05 fuel
      economy elasticity, the average fuel cost  per mile increases by more than 140 percent; this would
      result in large reductions in travel. By comparison, at an elasticity of .5, the average fuel  cost per
      mile increases by about 45 percent. In  the first case, trip and VMT reductions account for most of
      the drop in  fuel use, while in the second case, improved fleet fuel economy accounts for most of the
      drop in fuel use.  Since both the incidence and the economic, implications of the fuel price increase
      differ markedly between these two cases, forming a more  precise  understanding of fleet fuel
      economy sensitivity to fuel price is of some importance.
      12   A California-only gas tax increase seems more plausible for small to moderate tax increases (25 cents or
      less) than for higher ones, especially those of a dollar or more. Of course, it is not necessary to assume that a
      fuel tax or other, fuel price increase would be implemented in California only: the analyses could equally well
      represent the impacts of scenarios involving federal fuel tax increases or state tax increases implemented in
      many states. Also, for the analyses presented here, at-the-pump price increases implemented by sellers would
      have the same effects as a fuel tax increase.  A California-only interpretation of our analyses  does not
      necessarily require new, highly efficient vehicles to be produced for the state market (though it might make
      California an attractive test bed  for such vehicles, including ones currently  sold overseas but  not now
      marketed  in the U.S.).  It does  however presume that,  of the vehicles produced for  the U.S. market,
      manufacturers wpuld sell a higher share of the most efficient vehicles in California. Also, the used car market
      would be affected; demand for low mpg cars would decline in the state, and such cars would likely be  retired
      earlier or perhaps shipped to other states or countries for sale there.
B-20                                             ,                     U. S. Environmental Protection Agency

-------
                                                             Technical Methods for Analyzing Pricing Measures
                                                                        to Reduce Transportation Emissions

      Using our three elasticities, we studied a range of fuel price, increases'from $0.10 to $3 00 in 10
      cent increments.  The results.for the $2,00 fuel price increase under different elasticity assumptions
      are presented here, along .with some results for a $0.50 price increase. Results for these two price
      levels are sufficient to support generalization about price effects over the full range. 13

      It is worth noting that for some policy  objectives, the fuel price  (fuel tax) might be adjusted
      penodically to maintain the per-mile  cost, i.e., to reduce the impact of improved fuel economy
      Such  tax  adjustments would  make  sense  in terms  of paying for road  maintenance,  since
      maintenance costs do not decline proportional to fuel use. Similarly, if pay-at-the-pump insurance
      policies were implemented, it would be necessary for the component of the fuel "tax"  designated
      for insurance to be de-coupled from fleet efficiency.  If for either reason: the fuel.tax were adjusted
      to compensate for revenue losses due to fleet efficiency improvements, its effects oh VMT, trip
      rates,  delay, and emissions would be greater  than  we have estimated here.  Essentially  such
      adjustments would make the fuel tax very much like the VMT fee discussed below.


      B.3.5     VMT Fees

      A fee  oh vehicle-miles of travel (VMT) would directly charge users for the amount of vehicular
      travel consumed. A VMT fee therefore could be used  to reduce VMT-related impacts.14  Such a
      fee also would be a better targeted road user payment mechanism than the fuel taxes  we now use
      because drivers could  not reduce  their exposure to the fee by purchasing  more fuel efficient
      vehicles.1-5          ,                        ,

      Currently, the easiest way of collecting a VMT fee would be through a charge determined at the
      time of vehicle registration or vehicle inspection, based on owner-reported or inspector-recorded
      odometer readings.  However, if one goal of  a VMT fee is to reduce vehicular travel  and  its
      negative externalities, the fee should  be linked as closely as possible to  day-to-day use of the
      vehicle.  Collecting the VMT fee  as  part of an  annual payment for vehicle  registration would
      probably be less effective in reducing VMT than more frequent charges: an annual fee is remote
      from individual drivers' thinking about their day-to-day driving behavior, and may be less effective
      in influencing it.  Also, drivers would "discount" annual payments  compared to more frequent
      levies.
         We calculated:impacts on the basis of 250 times the average weekday rate plus 1.15 weekend-and holiday
     days at 95 percent of the weekday rate.
                                           "-*         '-  .              ' •
         VMT is roughly related to congestion, though a VMT fee would have a bigger effect on non-work travel
     than on work trips, which make up the majority of VMT during the congested peak periods  VMT is also
     roughly related to fuel use and to hydrocarbon, NOx, and carbon monoxide emissions.  In contrast,  PM
     emissions from on-road transportation are closely related to VMT.
     15
         Used as a road user payment mechanism, the VMT fee would have to be adjusted periodically or indexed
     to reflect costs of road construction, operations, and maintenance, or if such road costs increase the fee's
     percent cost coverage would decline.  Nevertheless, costs to each user would remain proportional to use Per-
     gallon fuel taxes also suffer from declining cost coverage unless adjusted or indexed, but are far less directly
     related to use of the roads because of divergent vehicle fuel efficiencies.            .    •
U. S. Environmental Protection Agency                    •                                             _«

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions

     There is no reason, of course, that a VMT fee tied to registration or I/M programs would have to be
     paid annually.  One can imagine a variety of alternative arrangements, including ones in which the
     registration or I/M fee itself is paid in (monthly or  quarterly .installments.  One approach might
     mimic the  billing method used by public utilities, in which monthly or quarterly bills are based on
     estimated usage, and a periodic reading (or report) is  used to calculate the additional increment due
     or credit earned. 16
                               ,    •   .    •           ...-.,.     |  •

     Recent technological developments offer other ways to frequently measure and collect a VMT fee.
      It  is currently feasible to put in place  a VMT monitoring system  using automatic  vehicle
     identification (AVI) technologies and covering all major facilities including freeways  and major
     arterials. Systems such as these are currently being deployed on tollways in many, parts of the U.S.
     as well as  abroad, and offer timely and accurate fee collection. In one  design motorists purchase
     debit cards which are displayed on their vehicles; fees are deducted from the cards electronically as
     the vehicles pass AVI  readers.   In .another design the  readers record  each  passing vehicle's
     identification code and transmit the data to a computerized system which accumulates the charges
     and periodically bills the vehicle owner.

     An alternative concept currently in prototype stage  would base the VMT fee on an at-the-pump
     reading  of an electronic  odometer  or  a special VMT-accumulating "smart  card;"   the
     corresponding fee would be calculated electronically and could be collected as part of the payment
     for fuel, or perhaps  recorded and billed  separately.   In one approach,  scanner  or microwave
     technologies would automatically read the odometer or another  on-board  electronic  device
     designed to monitor VMT. In another approach, the motorist would insert the vehicle's  " smart
     card" into a  special reader,  following  a sequence of actions much  like  those  used with the
     automatic credit card debiting devices now present in many fuel pumps.

     The availability of approaches, high tech or low, for collecting a VMT fee at or close to the time of
     road  use is important, because such immediate and visible prices are likely to be treated by
     travelers essentially as out-of-pocket costs similar to current fuel costs.  Here we treat the VMT fee
     as  a  pure  increase in the per-mile cost of driving, with no possibility of avoidance  and no
     " discounting" by drivers for delayed payment. In essence, the fee defined in this way would be the
     equivalent of a fuel tax increase that is indexed to vehicle fleet efficiency.

     VMT fees ranging from 1 to  10 cents per mile were analyzed for each metropolitan area (at the
     base case fleet fuel economy, this is equivalent to fuel price increases ranging from $0.22 to $2.20
     per gallon).   Results  for the 2 cents per  mile fee are reported here. 1?  In  keeping with the
     methodology described earlier, all elements of the STEP model were employed, from  residential
     location through supply response.  For Los Angeles and.the Bay Area, we further checked the
     results by assigning STEP-based peak trip patterns to the highway networks.  No differences were
     found that  would significantly alter the findings from STEP.
     16 Income and payroll tax collection methods are another possible model: frequent payments are made based
     on estimated amounts due and reconciliation of the amounts due is done via an annual report, subject to audit.
                 *  '                 '         ..''*'       ''                        \

     17  We calculated impacts on the basis of 250 times the average weekday rate plus 115 weekend and holiday
     days at 95 percent of the weekday rate.
B-22                                                                U- &• Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                                      to Reduce Transportation Emissions

      Note that because the results were produced at a regional level!, they are for within-region VMT
      only.  They do not include VMT generated outside each region being analyzed.  A VMT fee
      designed for revenue generation might, of course, be implemented on a statewide basis and could
      be analyzed in that fashion.

      A regional VMT fee based on-AVI monitoring of road use would be simple enough to implement.
      A regional fee based on odometer readings, on the other hand, would charge the motorist for inter-
      regional, interstate, and  international travel (Mexico, Canada) unless  some  mechanism  for
      excluding such travel were devised.  One can easily imagine ways to credit motorists for interstate
      and international travel; for example, motorists who want a credit for out-of-state travel could have
      their odometers read at stations along major entry and exit routes to the state, or a procedure might
      be established allowing a tax credit for documented out-of-state travel, much like the, one now used
      for fuel tax credits for exempt off-road vehicle use.  It would be much more difficult to devise a
      low-tech way to credit within state inter-regional travel without creating a major paperwork burden
      for all involved.  Since Caltrans periodically does statewide travel surveys which  include both
      within-region and  inter-regional travel, one approach might be to use the survey data to create a
      system of adjustments for each region to account for the average out-of-region component of VMT,
      perhaps by vehicle age.

      If the VMT fee were collected  infrequently, e.gl, once.a year  based on an odometer reading or
      report, its impacts  might  be somewhat less than we estimate here due to  discounting of future
      lump-sum payments in comparison to equivalent "out-of-pocket"  payments.  Hence  the  results
      reported here should be viewed as the high end of likely effectiveness.


      B.3.6    Emissions Fees

      Emissions fees represent a  means of reducing tailpipe emissions that could give .the  consumer
      somewhat more flexibility than the current system of mandated performance backed by vehicle
      inspection and maintenance. The basic concept is that the total pool of annual vehicular emissions
      in a region would be assigned a cost (presumably pollutant-by-pollutant), and each vehicle  would
      be charged a fee set to reflect its contribution to the total emissions burden.  Levying such a  fee on
      vehicular emissions arguably would  be  the most direct way to  instill  a  sense of personal
      responsibility for mobile source air pollution.

      While the concept may be simple to state, emissions-based vehicle fees are the most difficult of the
      pricing policies to define and analyze. Reasons for this are:

              •   the literature offers widely varying perspectives on the social costs of air pollution,
                  so an agreement on a monetary basis for the emissions fee is not easy to reach;

              •   estimates of cumulative  emissions from individual vehicles are imprecise and are
                  likely to remain so  unless and until vehicles are  equipped with accurate, tamper-
                  proof on-board emissions monitoring devices;                      "
              •   because  knowledge,  about how consumers would trade off emissions  fees,  repair
                  costs, insurance,  and  other auto-related expenditures is not wejl developed, the
                  change  in fleet composition resulting.from  a targeted emissions fee is difficult to
                  estimate.                                  .
U, S. Environmental Protection Agency                                                               B-23

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
     We carried out analyses of two prototypical emissions fee strategies, each using a different type of
     information about emissions, following the same line of argument as for the VMT fee, we assume
     that emissions fees can be collected on a "pay as you go"  basis, so that they are perceived by
     drivers as an out-of-pocket expense. This could be done with a technologically advanced system
     such as an on-board monitor, read and  billed, e.g., at the time of fuel purchase; or by combining
     some other method of fee calculation  with  a monthly  billing  system.  If the emissions fee is
     determined as part of vehicle registration or inspection/maintenance and  is billed  annually or
     biennially  as part of those programs, the fee may well  have less influence on day-to-day travel
     faehayior than we show (on the other hand, a large, infrequent fee might have a big influence on
     vehicle ownership levels, vehicle age and type, and vehicle maintenance).
                                           ,',,',,         j

     Al] non-arbitrary emissions fee concepts  rely on  some assumption about the social  costs of air
     pollution.  Accordingly, we searched through the literature for evidence that would support a
     specific emissions fee in each region, and sought the advice of experts in university research groups
     and air pollution control agencies.  We found  that the costs of air  pollution had not  been researched
     consistently for all the case study regions, and that the sources that do exist show a wide disparity
     in their damage estimates.  Credible cost estimates for mobile source pollutants range from about
     .25 cents per vehicle mile to about 8 cents  per vehicle mile (using regional damage estimates,
     reduced by the portion of emissions not attributable to mobile sources, divided by annual regional
     VMT).  The range reflects  differences in the severity of the pollution problems of the various
     regions and in the types of damage considered, as well as disagreements over specific costs in a
     given  region  (controversy is especially acute concerning the interpretation of epidemiological
     studies).

     Lacking more specific estimates of the social costs of emissions  in each of the California regions,
     we chose to set our emissions fee to average one cent per vehicle  mile. This represents a plausible,
     perhaps somewhat conservative estimate of current social costs  of mobile source air  pollution in
     these urban areas. Evidence suggests a much higher pollution cost in the Los Angeles region and
     perhaps a lower pollution cost in the Bay  Area.  The one cent per vehicle mile average fee would
     total about $1.15 million per day in the Bay Area and about $2.9 million per day in the Los Angeles
     Region, under base year (1991) VMT conditions.  While the  amounts sound high, annual receipts
     from such a fee would amount to about 0,3 percent  of the gross domestic product of each region. 18

     Clearly,  it would be inaccurate to  simply  charge each vehicle  the regional average per-mile
     emissions fee, since vehicles' emissions characteristics vary widely. We therefore  analyzed two
     possible methods for assigning a per-mile emissions fee to different vehicles.   Under  the  first
     method, the per-mile emissions fee would  vary by model  year and would be based  on data on each
     model year's average emissions  characteristics (i.e., using EMFAC  in California).  Under the
     second method, the per-mile emissions  fee would  vary with the actual emissions  performance of
     each vehicle, which might be determined through  emissions  testing, remote sensing, or on-board
     emissions monitoring.  The latter approach would  account for the differences in emissions among
     vehicles of the same model year.
     18   We used the same one cent per vehicle mile average fee for the year 2010 analyses, lacking more specific
     cost data.
B-24
                                                                   U. S. Environmental Protection Agency

-------
                                                           Technical (Methods for Analyzing 'Pricing Measures
                                               •  '                     to Reduce Transportation Emissions

      For each'household in the four regional travel survey samples,19 we  knew the make, model, and
      age (year) of the vehicle 'holdings for the base year, and we knew how each vehicle actually was'
      used on a representative weekday.  Thus, we were able to provide a well-grounded assessment of
      how vehicles of different ages and types are used and who would be impacted by emissions fees 20
      However, we did not have access to a model of how household vehicle holdings or vehicle usage
    -  patterns would change as a result of differential changes in the per-mile cost of vehicle operations,
      so we had to address these issues in terms of plausible scenarios rather than modeled estimates.

      Fees Based on Average Emissions by Model Year:

      For the average emissions by model year approach, we began by determining, for each region the
      average daily within-region VMT and emissions for every vehicle in  the regional  travel survey.
      We extracted from the survey data the vehicle trip sequences and their characteristics, and inferred
      whether the trip was a cold start, etc., based on the time between trips in the trip sequence. We also
      determined the  average trip speed and distance,  deriving these  data from the applicable highway
      networks.   We  then used EMFAC7F  data specific to each vehicle model year to compute the
      emissions for each vehicle trip.   ,                                               •

      From  the resulting samples  of vehicle trips and their  associated emissions,  average  weekday
      emissions and VMT were  calculated for each model year on  a  region-by-region basis. Annual
      emissions and VMT for each region were then estimated.  The annual VMT estimates were used to
      calculate total emissions costs for each region at the postulated one cent per mile average.

      For the year 2010  forecasts,  it was necessary to describe the likely vehicle age distribution and
      patterns of use for that future year.  We made the  simple assumption that the 2010 fleet would have
      the same general characteristics (age distribution,  usage profiles) as the current fleet does.  We then
      applied EMFAC7F 2010 emissions factors to this'hypothesized future fleet's trips to determine the
      future base case (total VMT and emissions, emissions by model year, etc.).

      For both 1991 and 2010, we used our calculations of emissions, by  vehicle model year to apportion
      the regional emissions cost estimates among model years.  The annual VMT calculations by model
      year then were used to determine an average emissions cost per mile for each model year. For
      example, from the 1991 data for Los Angeles, the average emissions fee per mile for a 1 year old
      vehicle would be about 0.4  cents, while the average emissions fee for a 17 year old vehicle (from
      the pre-catalyst era) would be about 7.0 cents.

     Note that the method we describe here should apply only to miles driven within each urban area,
      since emissions costs are calculated and apportioned on a regional  basis.  If the collection scheme
    ^used odometer readings as the basis for the VMT portion of the fee, some vehicle owners would be
     charged for miles driven in other regions or in other states.   To avoid this potential  inequity,
         The most recent regional survey for Los Angeles did not record vehicle make and model data. However,
     the Caltrans statewide survey of the same vintage included these data and had enough observations in the Los
     Angeles region to support ,the analyses described here. For this policy only, then, we extracted the Los
     Angeles data from the Caltrans survey and used it for our analyses.

         We calculated, impacts on the basis of 250 times the average weekday rate plus 115 weekend and holiday
     days at 95 percent of the weekday rate.     '           '
U. S. Environmental Protection Agency                                  •    .-  •                       B-25

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions

      methods could be developed to estimate in-region and out-of-region vehicle use and apportion the
      fee(s) accordingly, and credits could be given for documented out-of-state travel.

      We analyzed the effects of our per-mile emissions fees varying  by vehicle  age, assuming that
      households would not alter their vehicle holdings or pattern of use in response to the fees.  This
      assumption is not entirely realistic, since households could lower their fees by replacing their older
      cars with newer ones, and if AVI measurements or odometer readings are the basis for the VMT
      component of the fee, by using their newer cars in place of their older ones for some trips.21
      Nevertheless, the analysis results provide an indication of the maximum travel impact and the
      minimum emissions  impact that such an emissions fee could  be expected to  have; without  fleet
      changes the full impact of the fee would be passed through as  an out-of-pocket cost to the  driver,
      and the emissions reductions would come from reductions  in travel rather than from the  use of
      newer, presumably cleaner, cars.

      A more robust analysis would consider how vehicle holdings and usage patterns might change in
      response to an emissions fee.  The analysis would account for the determinants of household
      vehicle ownership and use and would estimate  the effects of an emissions fee on the number of
      Vehicles owned, the vehicle makes and model years, and VMT per vehicle.  Such a comprehensive
      model was not available to us, but we did have STEP'S internal  auto  ownership model,  which
      estimates whether a household will have  0,  1,  or 2+  vehicles as a  function of household
      characteristics, travel conditions, and vehicle ownership and operating costs.

      We used the STEP auto ownership model to partially account for the effects on the vehicle fleet as
      follows. For each region and analysis year, the base case household fleet was used to estimate the
      average annual cost of auto ownership for each  household.  Then, revised  annual ownership costs
      were computed to reflect the addition of emissions fees for each vehicle (based on model year and
      the actual daily  VMT revealed  in, the  survey).  New auto ownership probabilities then were
      calculated using STEP.22

      While this method is  an improvement over simply representing the  emissions fee as an increase in
      out-of-pocket costs, we feel that on balance it still is likely to overstate travel effects and understate
      emissions effects: For implementation scenarios involving AVI or  odometer readings, households
      with more than one vehicle could shift use among household vehicles to reduce their emissions fees
      without cutting back on travel.   Both  the revenues from  emissions fees  and their impact on
      households are therefore likely to be lower than' what we have estimated here.
     21    Alternatively, VMT could be estimated based on averages by model year taken from survey data.  This
     might be simpler to implement than an approach requiring odometer readings, but would remove much of the
     incentive for multi-car households to reduce "older car" use by substituting their newer, presumably cleaner
     vehicles for certain trips.

     22    Since STEP does not predict which-autos might be disposed of or what model years added when auto
     ownership levels change, we imposed a series of assumptions. We assumed that; since the per-mile emissions
     fee is higher for older vehicles, households that reduce their auto ownership  levels would get rid of their
     oldest car(s).  We assumed that households  maintaining their  current auto ownership levels would also
     maintain the age distribution of the vehicles they own. Households that added vehicles were assumed to add
     car(s) of the average age and fuel efficiency for that ownership level.  These assumptions allowed us  to
     estimate the effects on emissions, fuel use, etc.
B-26                                                             .'•  ' U. $. Environmental Protection Agency

-------
                                                           Technical Methods for Analyzing Pricing Measures
                                                                      to Reduce Transportation Emissions
      Fees Based on Measured Emissions:

      *To analyze an emissions fee based on measured emissions, we first needed an estimate of how
      emissions vary within each vehicle model year. One possible source of such information would be
      the data from .vehicle inspection and maintenance tests, but we did not have access to these data.
      Therefore we used an alternative  source, a database from  Professor Donald Stedman of the
      University of Denver, containing in-use measurements obtained passively with his remote sensing
      device at a location on Rosemead Blvd. in Southern California.23 Stedman expressed these data as
      frequency distributions of emissions by model year.

      We used the Stedman data to develop a frequency distribution of emissions fees per mile for each
      model year in each region. Taking the fleet age distribution and the VMT by model year estimated
      from the regional survey data, we used the Stedman emissions distributions both to estimate the
      aggregate emissions by model year and to apportion emissions responsibility within, model years.
      This approach allowed  us to assess a higher fee for high-emitting vehicles, and a  lower fee for
      relatively clean vehicles, within each model year.24

      To estimate the effects of a measurement-based emissions fee, we first made a special STEP run to
      create a base case with  emissions  derived from the high-emitter distributions rather than from the
      pure EMFAC data.  Since we did  not have actual emissions measurements for the vehicles in our
      samples, during this run we simulated the presence of high emitters in the fleet.  Each vehicle in the
      sample was randomly assigned an emissions level from the distribution for its model year (and
      tagged with that emissions level for use in the "after" analysis). Then,  the fee policy was tested
      using the same method  as for the  fee based on model year, averages, except that in this case the
      proposed fees were based on the emissions level,assigned during the "before" run.

      A fee based on measured emissions would probably require new technology of one sort or another.
      Tamper-resistant on-board monitoring and recording equipment would be the preferred approach;
      fees  based on multiple measurements using remote sensing equipment would be a second option.'
      A third approach would be to use the emissions measurements from I/M testing, though this would
      raise a  number of issues including whether the  fee should be prospective or retrospective and
      whether it should be based on before-repair or after-repair measurements.

      With an emissions fee  targeting  super-emitters,  households  could  be expected to  adroitly.
      manipulate their vehicle holdings and use to  minimize the impact of the fee.  This would tend to
      produce lower travel impacts  and higher emissions reductions than shown  here.
         There is some reason for concern that emissions distributions recorded for a single location and operating
     environment may not reflect the full spectrum of operating conditions,  and thus cannot be assumed to
     represent the "high-emitter" distribution for all regimes of urban travel. A similar criticism would apply,
     however, to vehicle inspection/maintenance test measurements, which are based on a single measurement and
     a specified operations sequence, or to any other data set based on single measurements and conditions.

          An alternative approach would be to use the EMFAC data as the estimate of the average emissions by
     model year, and to use the Stedman  data (or another source) to represent the underlying  distribution of
     emissions for that model year. Note that the overall approach does, riot produce different results if a higher or
     lower total emissions burden is assumed.                                               '
U. S. Environmental Protection Agency                 •                                               g_2y

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
     B.4      Impacts of Transportation Pricing Strategies

     B.4.1    Basic Analysis Results

     This section presents analysis results for the set of transportatiompricing measures we analyzed for
     our four case study regions. We present a series of 18 tables summarizing the basic findings of our
     analyses,  both by measure and by  region.  For each pricing measure, we present  the predicted
     percentage changes in VMT, trips made, and travel time; fuel consumed and CC>2 emitted; ROG,
     CO and NOx emissions; and annual gross revenues25, for the years 1991 (the base year) and 2010.

     The results for each analysis year represent the long-term effects of pricing measures,  i.e., the
     impacts resulting from the pricing measures as if they had been implemented several years earlier.
                                               -    • •  '         "         !•          ,
     The results show that carefully crafted and targeted transportation pricing strategies could do much
     to reduce travel times (hence congestion), cut energy use, and reduce emissions, at the same time
     increasing gross revenues substantially.  At the same time, it is clear that auto use and its impacts
     are quite inelastic with respect to most aspects of price.  This has two important implications: first,
     sizable increases in revenue can be obtained with relatively little effect on travel; conversely,  large
     price increases are necessary to obtain sizable reductions in travel and its externalities.

     The results  also provide an empirical dimension to the notion that the most efficient way to use
     price as a mechanism for reducing transportation externalities is to price each externality in a direct
     way. Thus,  as the tables  detail, the most  effective pricing strategy for emissions control (in the
     sense of emissions reductions per dollar charged) is to target high-emitting vehicles as precisely as
     possible; the most effective strategy for reducing fuel consumption (and CC>2 production) is to raise
     the price  of fuel; the most effective way to reduce congestion- is to  impose a toll at congested
     locations; and so on. Note that we refer to efficiency and effectiveness here in a purely technical
     sense. Other factors - ethical, institutional, political, and social - contribute to a broader assessment
     that may lead to different conclusions about policy effectiveness.

     Tables B1-B5 present the results organized by pricing measure for the year 1991.  Tables B6-B10
     present a subset of the year 1991  results, reorganized  by region. Each regional table includes
     analyses of the synergistic effects of groups of pricing measures, under two scenarios:

            1.  "Modest Pricing" - A relatively low set of prices from each category (e.g., $1.00 per
               day parking; $0.50 per gallon fuel tax), coupled with enough investment in transit only
               to maintain existing levels of service.

            2.  "Full Pricing" - A relatively high set of prices from each category (e.g., $3.00 per day
               parking;  $2.00  per gallon fuel tax), coupled with investment in transit corresponding to
               build-out of each region's long-range transit plan (as expressed  in future  network files
               made available by each  MPO).  Note that such a transit  expansion would  absorb a
               significant fraction of the pricing revenues.
     25      Net revenues depend  on the specific implementation strategy selected (public vs. private sector
     implementation and administration, technologies used, scope of implementation, timing of implementation,
     etc.).  In general, implementation designs costing a small fraction (5-15%) of gross revenues are feasible.
B-28                                                                U. S. Environmental Protection Agency

-------
                                                            Technical Methods for Analyzing Pricing Measures
                                                                        to Reduce Transportation Emissions
      Tables B11-B18 present the same ensemble of results for the year 2010.  The percent changes
      presented are changes from a year 2010 base case, created by using STEP as a forecasting tool. The
      regions' forecasts of households, household income, and household size (or population) were used
      to "factor" the 1991 household file to create a year 2010 household file for each region. The STEP
      models then were run to create a year  2010 "base case," using the 2010 household file plus the
      MPO network data for the year 2010.26   Finally, policy analyses were carried out to predict
      changes from the future base case.
          In.the Los Angeles region, some adjustments were made to SCAG's highway,travel times after analyses
     indicated that the SCAG models then in use showed far more trips and VMT than STEP'S more complex
     models would predict. Otherwise, MPO level-of-service projections were broadly consistent with  STEP
     internal calculations and were used as provided to form the basis of the 2010 base case.
U. S. Environmental Protection Agency
                                                                                              B-29

-------
  o
  0>
ls
i i
tyj >2i

I?

*t
••dr 3
II
11
  §
  I
  <3
  3
  <0




CO -R


CO
re
m
^
en
•<-

o
u.
01
e
a
O






3
z
o
o
M
O
o:
8

0)
3
U.
>,

(!)
D
CD
1-
Q.
>-
a.

£
£1
oj Q)
g o
CD *•
> CL


5
cn


uL
Description


5?

V-
CM
^-
?
^
V)

**

in


a?
m
o>

£

£
T

CO
T

o>
o
0


OS
I


OQ


i

g
O
55
*7
as
m
^
-9

,-:
1



.
'
CO
T-
§
O

o
9

rt
o
O

3
c
£
u
cn
co


1

s?
T
?-
CM
s?
CD
CN
0

d
1

s?
in
o

S
«?
o>
°

o
T-

u>
o
o


0)
o
Q
c
(0
(O


N.
CO

§
CM
g
to
~s
10
"
^


*

^
to
CM*
CM
§

g
'y

«
CM

e
o

s
o
u
a
o
CO
The congestion pricing strategy analyzed here assumes that prices
would be assessed on a per mile basis everywhere that congestion
appears In the highway network, including on arterials and collector
streets as necessary. A technology for electronic toll collection would
be required. Roadway message signs or In-vehicle readouts would
provide information about tolls on upcoming segments, likely as
part of a broader highway information system. Prices would not vary
minute-by-minute, but would be set to reflect average conditions on
each highway link during each period of the day, perhaps with
seasonal adjustments. The results shown here are based on a
reduction of congestion to level-of-servlce DIE, defined as a volume-
to-capacity ratio of .9. Note that travelers would continue to experience
some delay under this criterion, but that greater reductions in volume
might not be justifiable in economic terms.
                                                     §
                                                     I
                                                     s
                                                     I
                                                     I

-------
I B  g

    O
It

H
T  S
                                            a>
                                            o>
                                            i
                                           I
                                           I
                                            Q)
                                            42
                                            "3
                                            (4
                                            O
                                      .

                                       3

                                                  _  Q)
                                                   §  i
                                                   =  I
                                                  <  *

                                                      o
                                                   1|
                                                   i £
                                                  s
                                                      c
                                                     _o
                                                     '5>
                                                     S
Description
                                                           if
                                                            0 CM
                                                            9 <*
                                                           s«
                                                           o
                                                             So
                                                             •o
                                                           f- CO
                                                                     C3 OO

                                                                        s
                                                                     cVi o
                                                                     •7 *?
                                                                     CD T^
                                                                     CM v

                                                                     7 •?
                                                                                 S?
                                                                                 f"
                                                                                 ,
                                                                               T «?
                                                                               T- 

                                                                                         7 CM
                                                                                         m u>
                                                                                         CM' h:
                                                                                         T -r
                                                                                         •7-7
          O

          "3
          o
                                                                 -
                                                                                                           en*

-------
I
 I
 I,

   l

































*-
o>
0>
Fuel Tax Increases - 1
.2
1
i!

.Q TO
I— "^
ra a
3 c
C 0
<1






CO
co
ffl
T-
o>
o>
s
u_
CD
B>
nj
O








s
z
8
o
o
a:

1
O
3
U_
1"
a
Q

o
I
K

in
a.
*
Js
a.
>
P

>
""" "5
S3 "**
^ «
Hi
c
1 Tax
1 Increme


c
o


(£.





c
o
z:
£.
u
8
Q

sill
§gg.g
•???*
HP
ssssss'-s
i-- *°. *? Iq
*? r» ir- *?

^ SS 5S S?
fi eo to co


§000
•J CM tO CM
*? CM Csl •^>

„ ^5 ^ ^
§5 5 5
• T ^i* V

gsgg
•?T™^

511?


?ssa
9" 9 9 9


O O O O
to o o o
0£J M CJ


£
r
fl§5
-5S5^^
?•; **t ^1 r>
^ £2 £2 co

§ssi


gggg


0 0 0
gggg
* T *•* *r

§ggg
"^?v^

?!fl


CO CO in CM
 R **i N.
*? 5- *- hr

55 g g g

'
gggg
•T S CM ?

0 ^
uj r^ CM o


g£.g§
<7'?*1*

?ll?


rs co 10 CM
»- v O CM
0 ^) «S 0


o o o o
w o «=> o
S £>£> £

o
a
CD
•Q
E
»


(O en CM co
§T- O> CO
CM en o
g-.^g
'?'?°'!:
ssSi
^SSSS^
CO CM ^ QJ
cf «2 5; ic
' '
-? ^ ss ss
^ co iq to

' '
^5 &•- o~> &-
g 0 O ID
*? K CM ?!
'
000
^^ si
"? r^ C\l tj-

o °^ **t 5s
*? *^ T^ ^r

?Ils


CO CO 10 CM
^- V O CM
*?*?*?3ES™"2j-u>c;a»a>
sIitHl!|!!i!S!!i!!
llil|ll|s=il2|x-x-S|*|
- •o-oS«o-fex£l.££8SS-SS=||a
Nicfea-oi3'Es°-Sl'iS0«3
»!?^=°>£SSSS5'f^^J5|!gJ
sf5sai|is5g;ifR ?5|sf

||l^|||||l|| g f|l|f I*
t-a.cM4»m E « a 3S^ E £2o  S
"•- 1
is <° 3
I •= »,
g -2 =5
(0 0] *°
* a .2
2 . . x
1 £§
11 ?
I?!
ill
>i ^3 e
|i §
IS > 5
> •§ S
« ^e *-
— o 0
i*" O U)
* 5 £
> » S
? &*
11 -s
2 •" «!2
"So
_2 to O
1 £ if
S ° §
Hi
« 0 0
S •? 2
>t cj -n
5 o >.
S 2 *
» S .2
P> c
>> S
s •§ s>
Ions of dollars per year; VMTdenoti
weekday vehicle-trips; time is week
xide; ROG Is dally tons of reactive c
= £ °
S S f
s&|
o* 1
m » ™
I s •s
B-~ «
£ 6 g
IB » S
= = >.
Ill
* E m
* ° ~
« r o
Sou
o
•z.
                                                                          %
                                                                           a
                                                                           o
                                                                           I
                                                                          CO

                                                                          a
                                                                          CM

                                                                          *?
                                                                          «

-------
,o
•il
i
                   o>
                   o>
                   o>
                   I-

                   i
                   (0
                   42
                 0)  ,

                 •
                        i |
                        S*
                          I
                          5?
5?
o

••¥
                          CM


                          t
                          S?
                          o
                           ss

                           5
                          §
                              -15
                               §
                               «
                                "    -
                                 r
                              -
                            l-
        i!


-------
I

•
































L : » "! .,

I


































Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
Table B.5
Analysis Results for Emissions Fees - 1991
*
I Annual |








£

1
o
£
ca
en
j=
O









a
£









[Revenue!
O
o
o

CD
o
2

0
o

I
>
G)
a
CD
£
i=
.1
£
a.

5
^
E
^
0)
8
m


e
"3>
£

IDescription

O •*
CM CO
CO Ol
*9 I?
gtij

'
BS SS
u> **!
*? 3-
*
BSBS
?- o>
•f 
«? V
-5 --S
it S>

g§
T T *

BS g?
0 
«#«#
So «
*? T

B? BS
r- d
cji cvi


0 o

S J=

2
CD
£
U

expenditures Is not well developed. We looked at two broad strategies:
1) the emissions fee would be calculated based on EMFAC data for
average model year emissions per mile and actual VMT (so that the
per mile fee would vary only by model year); and 2) the emissions

CO T-
T CO
§»
BSg
"? S*
*
BS *
U) *t '
1^ ^
1
BSBS
O* CO
'f »r
B? BS
U) U3
*? V
0
c» co
~«?
gg
•? v

gs s?
"
7 T

« ^"
«# «?
01 a>


^ ^
w So
p CO O
•o >. >.
QJ JO iii
a ?. »
* ° o
S 2 o
E "o c
M CO n
0 ^ S
1-s i
> • S
^» *° 13
ctTo 5.
•o 2 J=
•" ffl 0
§ > c
Notes: Revenue expressed in millions of dollars per year; VMT denotes v,
of 10 microns or less; trips are weekday vehicle-trips; time is weekday
C02 is daily tons of carbon dioxide; ROG is daily tons of reactive orga

1
o
1 *
0
•3
1 .-
1
.1
?k
^
t§
""H
S

























V
m,


-------
&Q &a
| §
§ 'ES
3i
  "
•s:
1
a
                    OS
                      O)

                      '5
                      I
                      I
                      I
                      ,<2
                    •Q 2
                         21
                         Z I
                         0)

                         £? ffl
                         B O

                         Q) »-
                         > o.
                                  15
                                  §
                                          S o
                                         s g'i
                             Iillfliif|Sal
                            'IfiilliitlHl
                               «.t^4=«j





                              <=J£— 01  P in °-3 — S • - o
                                                 .2
                                                 o J2
                                                                    I

-------
CM



t
U

£
0)

I
42


£
      £
11
£ a.
S
Descriptio
          *;>
          BS s?
          co ^

          9 «?
          S? S?
          co n
            .
          CO 1O

          •7 •?"
                 21
                 9 «?
                 T «>
            gg
            «# 9-
                SS 2?
                10 a>
                T «?
                o o
                =! «!
                T- O
                       9
                       9 «*'
                   BS B?
                   o iq

                   T f

                   & B?
                   t »
                   T 7
                       B? S?
                       O CO
                       9 «#
                   Q

                   C
                              e esi
                           5?
                           (O
                         O) U)
                         cy w
                         5? S?
                         T- CO
                              o csi
O O
o o
*- CO
                         1

                            1
              : -a o oa re
                       to « .£ .2
                                      3 CD i«_
                                      O O) O
                                      « o a
                                      §.» •§
                                      » § §

                                      s a o
                                      ^ 01 w
                                      S ». w
                                      o- = =
                                     5
                                  s & s
                                  111
                                  ~° ™
                                  • jo o>
                                  ill
                                  2 «c o
                                  g i g
                                  &= E
                                  •S •§ g
                                  * i!
                                  §> ra ra
                                  ^S "
                                 > £ o
                                 •is >• £•
                                 O ^ r=
                                 !•§ «
                                 g „ «

                                 III
                                 * °
                                 •5 g £
                                 m ^ f?
                                  e i
                                  Hi'
                                  Hi
                                      = 2 I
                                      g n -a.
                                  S   s
                                  i ° =
                                  C k. *S
                                  o o 45

                                  11  -
                                  S o o
                                  O
                                  z

-------
11
11
f §
8?

I'
                             S
                              c
                              c
                                " T
                                 o

                                2 o
                                - •? CO
U3 CD tO

C^ CO  *^ *^ 1
                                        CO CO U> CM

                                        T- T- O CM
                                        ssss
                                                ,? cn u> eo
                                                ^ C* CM V-
                                                  : ^ ss
                                                  > m o
_r eo o to"

" 1"? T,
               ""S.«
                                                O O O O
                                                to o o o

                                                o oi oi CM
                                                        ^ T- S- °*
                                                        gg
                                                        CO ^ W rj
                        ?"•?
                                                                  o -5

                                                        cy « «2
        « ? g a
        o o o o,
         |,il,a.  .».








            •3ii-it!S*j!
               isllflf
                                               s 5 f | s 5 s s

                                  o
                                     ~
                                                       «e»

-------
I
                      .£

                      I
                      (0
                      .9
                      0)
S .12
OQ 5
     « I

       o>
     c S
                         o

                         O
                             I
                             g

                             ?
                              fr
                              CQ
                                   §
ffl


2
u
IB
CO
     8
     «o
                    s
                                        g
o
D>

£
O



I
     5

     CO
          §
          o
                                             IS
                                             o
                                             o
                         0
                         in
                                 — S — —
                                oroa))o


                               Ilflfl


            S = S * „ S i.^1
            • .-  ai ~ ". ^* ^r m
                              iiiilllllll
                              5-'£»'o'»
                              lllMllll

-------
I
ll
=n • M
a .£
S'«S
J?§
I'l
a, ?
§>!
II'
e K
13
I
1
s
I'.
                      2
                     •8
                      o
                      .8
                      8

                    * «
                    CQ 'a
                    fl) ^.

                    SI
§ i
c o









IT!
ro
m


o
CN

O

U.
a
Q









8
z
O
f T

C9
n
ce
o

••^
O
LL

fO

°
p
b
"~


H
EL
^

=i

to
 ~Z
"- S



o
O)
S

(Description .

CO CO
••t;

B?§

'i?T.
s?*5 '
•ef ^
*T

5? SS

«O 
-------
I
                        OJ
                        0>
                        CQ
                        O
                        O
                        I
                        i
                        (n
                        o
                     0) ^»

                     •
n =







10
n
m
en
T-

£
01
c
n
O







^
^N
z
O
o
0
i
1


D

P
1-
n

£
a
H;

|


Description




Strategy




^
^
evi
BS
1
s
S

Jg



B?
•?
^

V

jo
30



An Automatic Vehicle Identification (AVI) schem
would be used to price the regional freeway and
arterial system to maintain level-of-servlce (LOS
D/E

C 0
o o

O ffi O
Reglonwide Cong
Pricing (level-of-s
D/E) - Average $0
per Mile In Peak



o
^
CO
9
ss
9
i
*

B?
f>
CM

B*
T
BS

V

BS
9O

o
!»
All workers driving alone in the region would
experience a minimum $1.00 (1991) per day chai
for parking at the workplace.

*§


Reglonwide EmpI
Parking Charge o
Per Day



o>
T"
^
^
«?


g
g

^

T

BS
r-
co
^

Cjl

s?
CO
CM
0
O)
AH workers driving alone in the region would
experience a minimum $3.00 (1991) per day chat
for parkinq at the workplace.

ffl§
>\ 4»

Reglonwide EmpI
Parking Charge o
Per Day

CM

a
o>
BS
St



>.
•7
en

s«
a


$
•9
s?
m.


JjO
3>


CO
CM
s?

V
B?
CM'
IT
g
CM
B?
2

B?

CM
CM '

O

s?



ss

CM


d.
Q.
£
i
2
Q.
S
"5
M
0
O
U.




IFuel Tax Increase
by $0.50 (1991)

eo




IFuel Tax Increase
by $2.00 (1991)
£>
CO

I
^
^
»
BS
f
•f
V
i

BS
n


B?
CM
BS

T-

g5
3



§
5?

ID
T
3?
o>
£?
CO
en

-s

T

o»

S?

V

B?
10


C H>
Fees would average 1 cent per mile, and would
be collected frequently, as for the VMT fee.
In alternative 4a, the fee would be based on
annual mileage and average model year emissio
as reflected In EMFAC7F. In alternative 4b, the
fee would be based on actual odometer reading:
and in-use tailpipe measurements.

i I
~ <*

Mileage- and Emi!
Based Fee (Range
40-400/yr)


x
a> 2
~ 5*

Mileage- and Emi!
Based Fee (Range
10-1000/yr)



CO
B?
In


•f
g
B?
CM

^



CO

BS

'

B?
M

fe

Fees would be paid often, e.g., In the same man
as fuel taxes. This implies a potentially complex
collection scheme involving real-time reading
of the odometer, perhaps each time a vehicle Is
fueled. The simpler option of billing once a year
based on the odometer reading likely would hav
less effect on travel and emissions.

I

u.
3
S.
is
o •
£
r—

in

en
CO
n
CM
RS
C*
v
g
<«•
-14.4%
18.3%

B?

S

g
CO
35

I"?

;£

<9
m
r
1 , 2a, 3a, and 4a with maintenance of current tra
service.

1
0 f=

Example of Comb
Effects: Moderate



i
ss!

^^
c "w
1, 2b, 3b, and 4b with extensive transit investme
(Revenue not reduced to reflect cost of new tran




Example of Comb
Effects: High Imp:


              I
                                                                                                           I

IIS
                                                                                                           pa

-------

£f
fl
•S I

1-8

                 o>
                  I

                  I
                 §


                 1
                 (0
                 to
                 Q>
                 o
               «M Q;

               2
as
3
§
< £

«
c
c
I

m c
m =

E c,
s -
u. c


x: c
j :

(r


^
i

{Description

(strategy


o 3
> T-
g
5 g
:° ,
X to
3 ^
J B?
' T- .
m m

1

ss
CO
T

0


i


An Automatic Vehicle Identification (AVI) scheme
would be used to price the regional freeway and
arterial system to maintain level-of-servlce (LOS)
D/E • - •
§ S .
Regionwide Congest!
Pricing (level-of-serv
D/E) -Average $0.04
per Mile in Peak
^

en

S?

BS
V
BS
V
g
*
?


to
V

CM


*

o
§i workers driving alone in the region would
perience a minimum $1.00 (1991) per day charg
r parking at the workplace.
o
«ii O
Inegionwiae employe
Parking Charge of $1.
Per Day
•a ' •
M

O
cn
CM



B?
f .
^
5
9
1 -
D
to


r- '•
*t'

B?


?.


Ei workers driving alone In the region would
perience a minimum $3.00 (1991) per day charg
r parking at the workplace.
§
D O
4»
•8
: o
if
.c
0
= 1
,3 Q
In*


S

B?
•i
•^
?
,0
?
§
•
0

v»
CD
CO*

g
^

g


O
1*."

CO
CM
ifi

s?
!O
s

Q
d
1

CO
£;
'
3
T*
'
>
1


hees would be paid at the pump.

by $0.50 (1991)
e.,-1 T-., i 	
W - J


by $2.00 (1991)
?

r-

BS
1r
B?
CO
'
r^
rq
to .
i


?-
c*>

9
s|

B?



CO

1 '
M
a •
»
5 ,
\ '
I


BS
CM
B5
en

CO
g
6 }
n


eO

~*t
BS
•
'
«


i, ta, ja, ana 4a with maintenance of current trans
service.
•s
example or uomoinea
Effects: Moderate imps


CD
CO
T-
ft?
U5
fS
^
so"

7)
(O

gs
CM
O
•7



CM
9 '-'•

CJ
>
CM
-
i, ZD, ^D, ana 4D witn extensive transit investment.
Revenue not reduced to reflect cost of new transit

Effects: Hiqh Impact

                                                                              T
                                                                              CQ
                                                                       2 £
                                                                     > « S

                                                                              ^
                                                                     = ffl S '
                                                                              to

                                                                              t)

-------
1
1
                  i
                  I
                  eo

                  Q>
42

3
U)
Q>
o;
                o .*.

                II

5 S
c o>
C 9







in
S
^
en
o
£

c
en
O






0)
X
0
Z
o
u
to
o
u.
o
o
5
j
LL.
CO
0)
Q
i
~
0)
w
s
-


s
(Description


S
33




o


B?
" '
B?
«»
§
CM

91
CM


§
0
1
r-
1
i

j.

V

IAn Automatic Vehicle Identlflcatlon (AVI) scheme
would be used to price the regional freeway and
arterial system to maintain level-of-servlce (LOS)
D/E

c o
Reglonwide Congestio
Pricing (level-of-servlc
D/E) - Average $0.06
per Mile in Peak

T-


O
0)


g
?
BS
r-
§
^J~

§
T


eg

£
1
?

B?

•^

IAII workers driving alone In the region would
experience a minimum $1.00 (1991) per day charge
for parking at the workplace.

o
IRegionwIde Employee
Parking Charge of $1.0
Per Day
n
4


S
U)

£
CM
§

g
«?

g
Cy


§

§

n
c\i

B?
D
ri

IAII workers driving alone In the region would
experience a minimum $3.00 (1991) per day charge
for parking at the workplace.

o
IRegionwIde Employee
Parking Charge of $3.0
Per Day
A
CM


S>


lO
«?
ss
«?
ss
"

£
Cr>


0
'
£
•?
i

s?

7






§


"V
o


is



£
o

$
T
M

S
°!
CO

i
i
D
3
S.
D
3
to
0
0
u_


IFuel Tax Increase
by $0.50 (1991)

t


IFuel Tax Increase
by $2.00 (1991)




CO


§

BS
*
§


o>
7


£
*
§
*?
^
csi

B?

t



T-


»
i-
B?
o>
g
?

g
hJ


§
'
BS
C\l
g

ss

Csl

Fees would average 1 cent per mile, and would
be collected frequently, as for the VMT fee.
In alternative 4a, the fee would be based on
annual mileage and average model year emissions
as reflected in EMFAC7F. In alternative 4b, the
fee would be based on actual odometer readings
and in-use tailpipe measurements.
X
i S
Mileage- and Emission
Based Fee (Range App
40-400/yr)

•3-
X
A S
Mileage- and Emission
Based Fee (Range App
10-1000/yr)

^J-


Ol
«

BS
7
jg
*
B?
7

§
T


g
'
B?
o>
•p
BS
>i

^

7"

Fees would be paid often, e.g., in the same manner
as fuel taxes. This implies a potentially complex
collection scheme involving real-time reading
of the odometer, perhaps each time a vehicle is
fueled. The simpler option of billing once a year
based on the odometer reading likely would have
less effect on travel and emissions.


5
n.
s
o
»
5 •
L




CO


*
£
55
V
0)
CO

I0
<0


o>

0

s
!*•

~9

to

1, 2a, 3a, and 4a with maintenance of current transit I
service.

r,
Example of Combined
Effects: Moderate Impa




?
CM
CM
5?
CO
«
BS
CO
£
«

*
U)




£
to.
CM
!

B?
r>
3

1, 2b, 3b, and 4b with extensive transit investment.
(Revenue not reduced to reflect cost of new transit.) |


Example of Combined
Effects: High Impact


                                                                                *p
                                                                                p

                                                                                Q,
                                                                                I



-------
11
•^ tq
bo ~

11
II.
8

I
                 §
                 X-

                 4

                 s
                  (4
                 •2
                  a>
                  §
                  -j
                  Q>
                 I

                 42
               II
TO 3


Jj; lr
< s.







V)
&

o>
tn
E
t!
LL.
C
CO
U





8
z
o
U
0
n
a.
n
—

3
U.
m
"o
Q
0)
p
n

1-
i
I
(Description


£
S
w





T~

»
«*
5?
IO
•9
s?
«
IO

g

'

§

<*
»
<9
^

c\i
ro
"'
An Automatic Vehicle Identification (AVI) scheme
would be used to price the regional freeway and
arterial system to maintain level-of-servlce (LOS)
D/E

i 8
Regionwide Congesti
Pricing (level-of-servi
D/E) -Average $0.10
per Mile in Peak





en

§
T-
s?
T
s?

*7

85

1 ,

S?
"*• .
*
f
s«

*T
§

All workers driving alone in the region would
experience a minimum $1.00 (1991) per day charge
for parking at the workplace.

o
Regionwide Employe
Parking Charge of $1
Per Day



CO
CO


i5
151
SS
CM
ss
CO
CSI

g

1

5?
"2
1
1
3?
o
*?
ss
1
All workers driving alone in the region would
experience a minimum $3.00 (1991) per day charge
for parking at the workplace.

o

1, 2a, 3a, and 4a with maintenance of current-transit j
service. |

CO
Example of Combined
Effects: Moderate Imp





0>
-«-
cn
^
s?
CO
CO
S?
T-
o

CO
h-
T

s?
T
T
S
RO

?
s? '•
IO
«?
1, 2b, 3b, and 4b with extensive transit investment.
(Revenue not reduced to reflect cost of new transit.) I


Example of Combined
Effects: High Impact


                                                                             X * C
                                                                             > « 3
                                                                             >,*> v

                                                                             fl-
                                                                             IS i
IT ... oj
« « r-

« tt S
                                                                               L
-------

•58
I'i

tl
                I

                ra

                2
               CQ

               O
o


I


I
Q>
               0)
             «o Q;

             CQ -a
             o ^«


             II
~ea :
3 1
< —
11
o ;=
ir


E
a



c
_o
0
CO
Q)
Q






>s
a
1
CO


N
CM
Jj


to

55


S?

'
g.
V
5?
"


?d
°> > 1*°*
a ^g
tt S a


"

sS
9
BS


X o
3 >
Q
•O

3

^
?
s?
CO

5?

in
~S
m.

'a
•7

CM
^
»
^

3
M
C\l

5
CO
g
T
S«
I2
1
ss
f*-
s.
^
CO
«9
g
^
5?
CM
^
^f
~

s?
CD
1
i-ees would average 1 cent per mile, and would
be collected frequently, as for the VMT fee.
In alternative 4a, the fee would be based on
annual mileage and average model year emissions
as reflected in EMFAC7F. In alternative 4b, the
fee would be based on actual odometer readings
and In-use tailpipe measurements.
X
2
0.
Q.
<
3,
i
£.
Str
L. £.
I-O 0
0) C3
S3
m °

X
A> S.
c a.
o o.
•z <
CO Q)
E g
a re
o£
Mileage- ar
Based Fee
10-1000/yr)
D

CM
CM
""
g

s?
•7

~5


5?

*
o

S?
•9
~?

•7

-S
en
«?
t-ees would ue paid otten, e.g., in the same manner
as fuel taxes. This implies a potentially complex
collection scheme involving real-time reading
of the odometer, perhaps each time a vehicle Is
fueled. The simpler option of billing once a year
based on the odometer reading likely would have
less effect on travel and emissions.

_







"
^1
0
r~
s«
2
•
s«
0
in
s«
en
en
t>
D
s«
CM
CO

•^f


0


i, 'iz, 3a, and 4a with maintenance of current transit 1
service. . |

**
g,
E
0}
S
•8
i °



-------
I
to *J
II,
  _.

  I I

                   x:
3 -? §, §
• v> > """*
° = JL.E
p-Sa s
a: D- Q Q
,-


CM
""
g

B?
1-

j=

T
B?
?

S
BS

T
g




BS
o
T

IAII workers driving alone in the region would
experience a minimum $1.00 (1991) per day charge
for parking at the workplace.
o


0.0
LU H>
o «
g 0> >i
> « o
0:0.0.
CO


en
5 ,
g

5S
CO
CM

B?
DO
CM
BS
«?

S
BS

m
BS
CO




BS •
CO
•y

All workers driving alone in the region would
experience a minimum $3.00 (1991) per day charge
for parking at the workplace.
o

o> e->
§•£
L °
a» £
1 0) >
II?
J ec o
CM


^. *
t
g
"
BS
en


BS
3
T
S?
OT

b
3?
D
1O
r>




?
T

i
3
a. •
3
3
3
a
j
i
3
>
}



>
)
» T-
X 0
) **
. J3
CO


in
v
S?
10
ci
IT-
§
CM
*
BS

T
BS
S

BS
CM
CM
BS

T
g

1


BS
CM






O
*
£ m
u en
X 0
K S
o> w
£.§
eo


CD
T-
g
^
g
ui

BS

»
BS
?

BS
iq
BS
0
*
o
M



BS
t



CM
0
T-
ss '
CO
T
BS
A
1
BS

S
^ •' •
1-^

g
>i
^

t
g




S?
D
CM

Fees would average 1 cent per mile, and would
be collected frequently, as for the VMT fee.
In alternative 4a, the fee would be based on
annual.mileage and average model year emissions
as reflected in EMFAC7F. In alternative 4b, the
fee would be based on actual odometer readings
and In-use tailpipe measurements.
X

1 <
: o
" &
S g«
i £ 5-
o> -o o
is?
sm §
I
A?'

"to ^
^ o>
*° £•
? 8S
CD "- 0
c> -a o
CJ CD °
» S T
is?
ja.



S
«
CO
BS
^ .
'
-5

* . ' '
^
•f '

S '
-5

f
BS -




BS .
•?

Fees would be paid often, e.g., in the same manner
as fuel taxes. This implies a potentially complex
collection scheme involving real-time reading
of the odometer, perhaps each time a vehicle is
fueled. The simpler option of billing once a year
based on the odometer reading likely would have
less effect on travel and emissions.
£

I
CM .
*
O
O
I



to
o
BS.

5?
to
3

•7
BS
i

BS
S
T
BS

?
BS
M
a



3
s|

1 , 2b, 3b, and 4b with extensive transit investment.
(Revenue not reduced to reflect cost of new transit.) I


•o
a +j
38
U £
5*
!• S

                                                                                                  I
I





I







I


«§

-------
                                                         V»*r .'.-••  .v „
I?
                                  I

                                 §
I
 (0
to
                                 1
                                  V)
                                  Q>
                             •SJ ^*
                             •Q  2

o
3

<






m
m
o


3
u.
a>
c
m
f















3
C
Q>
>
X
z
0
o
o
n
X.
CM
O
0
0)
3
U.

re
o
Q


o
1

n
£
Q.
t-f
Ssi
jS
E

c
«
u
a

>,
05
a
1
m





>
CO
O

a
CO

4
^T

^
*9


^
CO


^a
•*
^
y


^
|^
•^


An Automatic Vehicle Identification (AVI) scheme
would be used to price the regional freeway and
arterial system to maintain level-of-servlce (LOS)
D/E

Reglonwide Congestion
Pricing (level-of-servlce
D/E) - Average $0.09 •
per Mile in Peak




r-


B
CO
»
o>
9
s
o

=.



to



S5
5
s?
3


*S
o>


o
All workers driving alone In the region would
experience a minimum $1.00 (1991) per day charg
for parking at the workplace.

Regionwide Employee
Parking Charge of $1.00
Per Day
CO
CM



D
CO
-«
«*
ib
CM
g
CM

U>
CM


ss
*r


iS
CO
CO
s?
o
CN

^0
-Cf
CM

cu
All workers driving alone In the region would
experience a minimum $3.00 (1991) per day charg
for parking at the workplace.

Regionwide Employee
Parking Charge of $3.00
Per Day
a
CM


1^


-S
n

CO
•?
CO
«?

i:
?


ss
o



^e
to
•9
s«

•7

*?
en
eo


Fees would be paid at the pump.

Fuel Tax Increase
bv $0.50 (1991)
n



rw

CM
H
T-
g
~
?5
T

CO

CM




Fuel Tax Increase
bv $2.00 (1991)
JQ
CO




N
(0
T
^
«?
>
>

^
t


o
U>



>0
CM
^

CM

X-
to
CM


Fees would average 1 cent per mile, and would
be collected frequently, as for the VMT fee.
In alternative 4a, the fee would be based on
X
Mileage- and Emissions-
Based Fee (Range Appro
40-400/vr)
n
Tt



n

N
CO
N
O> r
in
T


*!•


to



SS
cl
5?
r^


^
o>
T


annual mileage and average model year emission
as reflected in EMFAC7F. In alternative 4b,the
fee would be based on actual odometer readings
and In-use tailpipe measurements.
X
Mileage- and Emissions-
Based Fee (Range Appro
10-1000/yr)





M
CO
CO .
CO
O
-r
^
"*

4
t


lO
°P


s^O
•?
-?
3
•i

^o
CM



Fees would be paid often, e.g., in the same manne
as fuel taxes. This implies a potentially complex
collection scheme Involving real-time reading
of the odometer, perhaps each time a vehicle Is
fueled. The simpler option of billing once a year
based on the odometer reading likely would have
less effect on travel and emissions.

VMT fee of $0.02 per mill

to


o


9
O
4
eo
^
^
xO
en
•7


§
CM


3^
f
ss

*T

s^
in
«P

*j
1, 2a, 3a, and 4a with maintenance of current tran
service.

Example of Combined
Effects: Moderate Impact




o>

CO
3

o>

LJ **
1, 2b, 3b, and 4b with extensive transit investmen
(Revenue not reduced to reflect cost of new trans

Example of Combined
Effects: High Impact


                                                                                                                                                               I
                                                                                                                                                               s;

                                                                                                                                                               .1

                                                                                                                                                               I
                                                                                                                                                               a

-------
I
 -J
 I  §
I
 o o
I
E2;
                           •i

                           i
                           o
                          -J
                           o
                          «

                          5
                          42
                           3 ,
                           8
                       oo a:

                       2
"<5 3
ZJ I
& 0
<





y.
CO
m
o
0
.»
IRegionwide Empic
Parking Charge of
Per Day


T~
10
^-

§
*?
CM
-S
3

CM

s«
to
°?

S?
M
7 .
b
*?


i>.

IAII workers driving alone in the region would
experience a minimum $3.00 (1991) per day charge
for parking at the workplace.

o

>.£»
IKeglonwIde Empic
Parking Charge of
Per Day


^,
K
CO
ss
"
I
V-
•f

en

s?
u>
°?

5?
^_
^
5?
en
CO


«i
T
it-ees would De paid at the pump.




Ii-uei lax increase
by $0.50 (1991)


U>
CM

§
CM
g
N
s?
CO
N

*D
^

§
CM

s?
!*•
CO
S?


"
b -•






I rue i lax increase
by $2.00 (1991)
D

tf\
0
—
§
•i
i
•*
U)
•9

1

g
10

5?
CO

•»
CM


)
CM


CO

g
"
CO
CO
5?
c»
CO

s

0
*?

^


en



BS
CNI
hees would average 1 cent per mile, and would
be collected frequently, as for the VMT fee.
In alternative 4a, the fee would be based on
annual mileage and average model year emissions
as reflected in EMFAC7F. In alternative 4b, the
fee would be based on actual odometer readings
and in-use tailpipe measurements.
X
en 2
: Q-
T; <"
Mileage- and Emls
Based Fee (Range
40-400/yr)

X
CO 8
c Q.

mileage- ana tmiss
Based Fee (Range
10-1000/yr)
D ' '


3
CO
B«
CO
N

CM
-i

s?
CM
»?

52 - ,
£2

0
°fc
«?
^
T




hees would be paid often, e.g., in the same manner
as fuel taxes. This implies a potentially complex
collection scheme involving real-time reading
of the odometer, perhaps each time a vehicle Is
fueled. The simpler option of billing once a year
based on the odometer reading likely would have
less effect on travel and emissions. .








in
cs
CM
V
O
N
5?
01
£
0
i

6?
10
CM

5?
o
5

s?
o>
CO
S?

i , za, aa, ana 4a witn maintenance of current transit 1
service. • " |


CO
a.
c
txampie or uomoin
Effects: Moderate li



O

ss
«t
o>
.
i

1
•Cf

*
CO

^
^_
n
^
\|


CN
«J
1, 2b, 3b, and 4b with extensive transit investment. I
(Revenue not reduced to reflect cost of new transit.) I



o
0 *j
5 §
• E
> a
| X
if

                                                                                                                         s.^- s
                                                                                                                         is ^ ^
                                                                                                                                              30


                                                                                                                                              C.

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
     B.4.2    Equity
                   !  '                     ' !'    I      ','',. '    '     i
                   t                   ,    I1           ,                   j
     One of the biggest contentions raised  about strategies that increase the price of transportation is
     that, while some people would benefit, others would be unduly hurt.  Whether from an ethical or a
     pragmatic political perspective, these equity concerns, which stem from the possibility of unevenly
     distributed benefits and costs, are a central implementation issue for transportation pricing.  Price
     increases are  especially a worry for low income individuals who may not be able to afford the
     higher costs and hence might be priced out of certain travel options.  Higher transportation prices
     also are a concern for moderate income people who have little flexibility about when or where they
     travel and hence might have to devote a larger share of their income to transportation.

     One would  not want to overstate  equity issues.  First,  one might argue that there is  nothing
     inherently unfair about expecting people to pay for the services they consume, to cover the costs of
     damage they do to  the environment, and so on, regardless of their socioeconomic status.  In fact,
     this could be  seen as a more equitable result, since it removes undeserved burdens from others.
     Second,  it is important to note that for many pricing applications, and especially for congestion
     pricing, the dollar cost is higher for those who pay it, but time and other costs decline; many people
     should be better off despite the higher prices. Finally, for  any of the measures, use of the revenues
     to improve transportation services could result in net benefits for most. In short, simply noting that
     prices are higher does not mean that the result is necessarily less equitable.

     Nevertheless, it is important to have good information on the distribution of costs and benefits of
     various transportation pricing strategies, including the status quo, so that the social and political
     ramifications  can be  anticipated and dealt with and so that program  designs can be structured to
     achieve a satisfactory level of fairness. While full treatment of the equity  issues of transportation
     pricing would require a separate study, a portion of our effort was devoted to exploring the impacts
     of the various pricing strategies on different groups and interests.  Indeed, the analysis procedures
     described here  were  designed to  produce as  much  information  about  the  distributional
     consequences of pricing as possible.
                                              1                 • ' ' '      I
     The distribution of impact and equity can be thought of  along many dimensions  - income, class,
     race, ethnicity, age, sex, and geography are among those commonly considered. For the illustrative
     purposes of this section, however, we have chosen to focus our attention primarily on differences
     by income level. We split the households of California into five equal household income groups,
     and used the resulting quintile boundaries to categorize  our findings throughout the analysis of
     pricing policies.  The five quintiles are:

               Quintile              Income Range (1994$)

               1                             <= $18,700
               2                             $18,701-$36,500
               3                 "           $36,501-852,100
               4                             $52,101-$71,300
               5               '              >=$71,301
 B-48                                                                U. S. Environmental Protection Agency

-------
                                                            Technical Methods far Analyzing Pricing Measures
                                   '              ,                       to Reduce Transportation Emissions

     Tables B19 and B20 present a distillation of quintile data based on the 1990 U.S. Census Public
     Use Microdata Sample for California.  It may be helpful to begin a discussion of equity by first
     looking at some basic facts about the distribution of income in. California, as shown in these tables.
U. S. Environmental Protection Agency
B-49

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
  table B.19: 1990 US Census California Statewide Summary
             Public Use Microdata Sample

  Share of Households in Each State Income Quintile

Region
Sacramento Region
San Diego Region
San Francisco Bay Area
South Coast
Balance of State
California Combined Total
Share by State Income Quintile
1
0.22
0.19
0.16 .
0.19
0.27
0.20
2
0:22 -
0.21
0.17
0.19
0.24
0.20
3
0.21
0.21
0.19
0.20
0.20
0.20
4
0.20
0.20
0.22
0.20
0.17
0.20
5
0,15
0.18
0.26
0.22
0.12
0.20
Total
Households
598405
885574
2242554
4560620
1744923
10032076
 Share of Population in Each State Income Quintile

Region

San Diego Region
San Frandsco Bay Area
South Coast
Balance of State
California Combined Total
Share by State Income Quintile
1
0.17
0.15
0.11
0.15
0.21
0.15
2
0.21
0.20
0.15
0.18
0.24
0.19
3
0.22
0.22
0.19
0.20
0.22
0.20
4
0.23
0.23
0.25
0.22
0.19
0.22
5
0.17
0.21
0.31
0.25
0.14
0.23
Total
Population
1560521
2386031
5852335
13233643
4930037
279962567
 Share of Autos in Each State Income Quintile

Region
Sacramento Region
San Diego Region
San Francisco Bay Area
South Coast
Balance of State
California Combined Total
Share by State Income Quintile
1
0.13
0.10
0.08
0.10
0.17
0.11
2
0.19
0.17
0.13
0.15
0.22
0.1 e
3
0.22
0.22
0.18.
0.19
0.23
0.20
4
0.25
0.25
0.26
0.24
0.22
0.24
5
0.21
0.26
0.36
0.31
0.18
0.29
Total
Autos
1080383
, 1577796
3941140
8077199
3163821
17840339
 Share of Resident Workers in Each State Income Quintile

Region
Sacramento Region
San Diego Region
San Frandsco Bay Area
South Coast
Balance of State
California Combined Total
Share by State Income Quintile
1
0.09
0.08
o:os
0.08
0.11
0.08
2
0.18
0.17
0.12
0.15
0.21
0.16
3
0.23
0.22
0.18
0.20
0.24
0.21
4 5
0.27 0.23
0.26 0.26
0.27 0.37
0.26 0.32
0.25 0.19
0.26 0.30
Total
Workers
715029
1145517
2993791
6237629
2010851
13102817
B-50
                                                             U. S. Environmental'Protection Agency

-------
                                                     Technical Methods for Analyzing Pricing Measures
                                                               to Reduce Transportation Emissions
 Table B.20: 1990 US Census California Statewide Summary
             Public Use Microdata Sample

 Autos per Worker in Each State Income Quintile

Region •
Sacramento Region
San Diego Region
San Francisco Bay Area
South Coast
Balance of State
California Combined Total
Autos per Worker by Income Quintile
1
2.16
1.75
1.89
1.75,
2.37
1.93
2
1.62
1.37
1.38
1.28.
1.66
1.40
3
1.48
1.33
- 1.30
1.24
1.47
1.32
4 5
1.38 1.36
1 1.33 1.36
1.25 1.27
1.23 1.28
1.37 1.42
1.27 1.30
Regional
Average
1.51
1.38
1.32
1.29
1.57
1.36
 Work-at-Home Share in Each State Income Quintile
Region
Sacramento Region
San Diego Region
San Francisco Bay Area
South Coast
Balance of State
California Combined Total
Share by State Income Quintile
1 2 3 4 5
0.041 0.029 0.030 0.026 0.037
0.042 0.033 D.026 0.032 0.039
0.052 0.034 0.030 0.027 0.034
0.035 '0.023 0.022 0.022 0.032
0.044 0.038 0.032 0.031 0.047
0.041 -0.030 0.026 0.026 0.035
Regional
Average
0.031
0.034
0.032
0.027
0.037
0.03
 Commute Time per Worker in Each State Income Quintile

Region
Sacramento Region
San Diego Region
San Francisco Bay Area
South Coast
Balance of State ,
California Combined Total
Minutes per Worker by Income Quintile
1 ' . 2 3 4 5
19.17 20.28 21.55 22.89 - 22.53
21.72, 21.84 22.28 23.23 23.20
23.24 23.64 25.35 26.20 26.37
25.65 25.30 25.90 26.81 27.28
18.00 18.70 19.45 20.47 20.26
22.84 23.04 24.03 25.20 25.83
Regional
Average
21.71
22.65
25.65
26.46
19.55
24.63
Drive Alone Share for Workers in Each State Income Quintile

Region
Sacramento Region
San Diego Region
San Francisco Bay Area
South Coast
Balance of State
California Combined Total
Drive Alone Share by Income Quintile
1
. 0.65
0.62
0:54
0.58
0.64
0.59
2
0.71
0.67
0.61
0.64
0.70
0.65
3
0.76
0.73
0.67
0.70
0.75
0.71
4
0.79
0.77
6.71
0.75
0.77
0.75
5
0.77
0.80
0.73
0.80
0.79
0.78
Regional
Average
0.75
0.74
0.69
0.72
0.74
0.72
U. S: Environmental Protection Agency
                                                                                    B-51

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions

                  "                         :         I           '            '
     By definition, each income quintile contains one fifth of the total number of households in the state.
      But the distribution of household income within the state is uneven; there are notable differences
     among regions. For example, the San Francisco Bay Area is relatively well off, with 48 percent of
     its households in the top two quintiles and only 33 percent of its households in the bottom two
     quintiles. In contrast, the small urban and non-metropolitan areas of the state have just 29 percent
     of their households in the top two quintiles and 51 percent in the bottom two quintiles.   While
     housing prices and other cost-of-living factors may cloud the comparison somewhat, it seems clear
     that the ability to pay higher transportation prices is not distributed evenly around the state, but is
     higher in its metropolitan areas.

     Other important points can be observed be examining the income quintile data.  In particular:

              •   Population is not distributed evenly among  the quintiles. Higher income  households
                  tend to be larger, such that 23 percent of the population is in the highest quintile and
                  15 percent in the lowest quintile.

              •   Auto ownership increases with income. 53 percent of the vehicles for personal use in
                  California are owned by the top two quintiles, while only 27 percent are owned by
                  the bottom two quintiles.  This suggests that policies which cause a general increase
                  in the cost of auto ownership may apply disproportionately to upper income groups.

              •   Households with workers tend to have higher incomes than those which do not. 56
                  percent of the workers  statewide are in the top two quintiles, while only 24 percent
                  are in the bottom two quintiles.. This suggests that policies which cause a general
                  increase in the cost of commuting  may apply disproportionately to  upper income
                  groups.

              •   Autos per worker  is consistently high in all  income groups. Table 20  shows that
                  quintile 1 - the lowest  income group - has  the highest auto ownership per worker.
                  This counter-intuitive result is due to the large  group of retirees falling into that
                  quintile. Removing the retirees from the  data base produces a  ratio of autos to
                  workers of 1.25:1  for each of the five quintiles.  While this does not have  direct
                  implications for pricing policy, it does suggest that access to an automobile for the
                  commute is widely distributed in California.

          •   Drive-alone share for commute .travel rises with  income. The drive-alone share statewide
              is about .59 in the lowest quintile and .78 in the highest quintile, with similar variation in
              each region.  Putting the mode shares (including  the shared ride data not shown here)
              together with the proportion of workers in each quintile, it becomes clear that only about
              6 percent of the commute  vehicles statewide will have drivers in the lowest quintile,
              while about 35 percent will  have drivers in the highest quintile.

          •   Commute time per worker  rises with income.  The average self-reported commute trip
             time statewide is about 22.8 minutes for workers in the lowest quintile  and 25.8 minutes
              for workers in the highest quintile, with similar  variation  in each region. Because  many
              of the low income workers' miles are made by  transit (or by foot) at  speeds far below
              auto speeds, even on congested networks, it is  clear that higher income  workers' trips
                                                                  (J, S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                                     to Reduce Transportation Emissions

               must be considerably longer (in VMT) than those of their lower income counterparts.
               This illustrates a crucial point for pricing studies: higher income workers are the largest
               contributors to work trip VMT, partly because high income jobs  and high-end housing
               are relatively sparsely distributed around each region.

           •   Both low and high income workers are more likely to work at home. About 3.5 percent
               of workers in the highest quintile and 4.1 percent of workers in the lowest quintile  listed
               home as the primary place of work in 1990, compared to 3 percent of workers overall.
               While these phenomena are not well understood, it is said that  participation rates by
               upper income households have been increasing in recent years. This may indicate that
               upper income  households have an  important way to blunt the  effect of large  price
               increases, namely by choosing to work at home some of the time.

     .The PUMS  data demonstrate one of the  most  important  facts about equity  of the current
     transportation system.  Truly poor people make  relatively little use of the  highway system as it
     operates today and, consequently, would pay  comparatively little under most transportation pricing
     scenarios (in absolute terms, not necessarily as a share of income).

     Equity Analyses Using PUMS and STEP

     An unstated implication of the PUMS analysis is that the lower middle class - say, quintiles 2 and 3
     - would sustain much of the impact of pricing policies. This hypothesis was explored in a range of
     analyses using STEP, examples of which are shown in Tables B21 and B22.  The STEP analysis
     framework allows us to examine equity issues in detail because it utilizes  specific demographic
     information, at the individual household level, that can be associated directly with the effects of
     each pricing policy.
U. S. Environmental Protection Agency
                                                                                           B-53

-------
Technical Methods for Analyzing Pricing Measures
to Reduce Transportation Emissions
 Table B.21
 Equity Implications of a VMT Fee in the Los Angeles Region -1991
VMT Fee
(cents/mile)
1
2
3
4
5
6
7
8
9
10
Absolute Change in Daily VMT by Income Quintile
Q1
-1.8
-3.4
-4.9
-6.2
-7.4
-8.5
-9.5
-10.5
-11.3
-12.0
Base VMT (millions) 25.5
Per Capita Daily VMT 11.7
Q2
-1.9
-3.7
-5.4
-7.0
-8.6
-10.1
-11.5
-12.9
-14.2
-15.4
45.0
17.3
Q3
-1.4
-2.8
-4.1
-5.5
-6.8
-8.1
-9.3
-10.5
-11.7
-12.9
54.8
19.1
Q4
-1.1
-2.2
-3.3
-4.4
-5.6
-6.7
-7.8
-8.9
-10.0
-11.1
71.9
22.0
Q5
-0.5
-0.9
-1.5
-2.0
-2.6
-3.2
-3.8
-4.5
-5.1
-5.8
92.8
25.8

Total
-6.6
-13.0
-19.2
-25.2
-31.0
-36.6
-42.0
-47.2
-52.3
-57.3
290.0
20.0
VMT Fee
(cents/mile)
1
2
3
4
5
6
7
8
9
10
Percent Change in Daily VMT by Income Quintile
Q1
-7.0%
-13.3%
-19.1%
-24.3%
-29.1%
-33.5%
-37.4%
-41.0%
-44.2%
-47.2%
Q2
-4.2%
-8.2%
-12.0%
-15.6%
-19.1%
-22.4%
-25.6%
-28.7%
-31.5%
-34.3%
Q3
-2.6%
-5.1%
-7.5%
-10.0%
-12.4%
-14.7%
-17.0%
-19.2%
-21.4%
-23.5%
Q4
-1.5%
-3.1%
-4.6%
-6.2%
-7.7%
-9.3%
-10.8%
-12.4%
-13.9%
-15.4%
Q5
-0.5%
-1.0%
-1.6%
-2.2%
-2.8%
-3.5%
-4.1%
-4.8%
-5.5%
-6.3%
Total
-2.3%
-4.5%
-6.6%
-8.7%
-10.7%
-12.6%
-14.5%
-16.3%
-18.0%
-19.7%
 Note: Quintiles defined in terms of 1989 Census household incomes.
   VMT is vehicle-miles traveled in millions per day. Sales tax relief,
   Improved transit, and other potential expenditures to mitigate
   impacts on lower income households are not reflected here.
B-54
                                                                U. S. Environmental Protection Agency

-------
                                                        Technical Methods for Analyzing Pricing Measures
                                                                  to Reduce Transportation Emissions
 Table B.22
 Equity Implications of Congestion Pricing in the
 San Francisco. Bay Region -1991
Average
Peak Fee
(cents/mile)
1
2
3
4
5
6
7
8
, ; 9
10

Absolute Change in Daily VMT by Income Quintile
Q1
-0.2
-0.3
-0,4
-0."5
-0.6
-0.7
-0.8
-0.8
-0.9
-0.9
Base VMT (millions) 7.2
Per Capita Daily VMT 10.0
Q2
' -0.2
-0.3
-0.5
-0.6
-0.7
-0.8
-0.9
-1.0
-1.1
-1.1
14.0
15.3
Q3
-0:1
-0.2
-0.3
-0.4
-0.5
-0.6
, -0.7
-0.7
-0.8
-0.9
19.6
16.8
• Q4
-0.1
-0.1
-0.2
-0.3
-0.4
-0.4
- -0.5
-0.6
-0.7
-0.8
30.3
19.5
Q5
0.0
0.1
0.1
0.1
0.2
0.2
0.3
0.3
0.3
0.3
44.0
22.6


Total
-0.6
-1.1
-1.6
-2.1
-2.5
-2.9
-3.3
-3.6
-3.9
-4.2
115.0
18.3
Average
Peak Fee
. (cents/mile)
1
2
3
4
5
6
7
8
9
10
Percent Change in Daily VMT by Income Quintile
Q1
-2.2%
. -4.2%
-6.0%
-7.5%
-8.8%
-10.0%
-11.0%
-11.8%
-12.4%
-12.9%
Q2
-1.2%
-2.3%
-3.3%
-4.2%
-5.0%
-5.7%
-6.4%
-7.0%
-7.5%
-8.0%
Q3
-0.6%
-1.1%
'-1.7%
-2.2%
-2.6%
-3.0%
-3.4%
-3.8%
-4.2%
-4.5%
Q4
;-0.2%
-0.5%
-0.7%
-1.0%
-1.2%
-1.5%
-1.8%
-2.0%
. -2.3%
-2.6%
Q5
0.1%
0.2%
0.3%
0.4%
0.5%
0.5%
: 0.6%
0.6%
0.6%
0.6%
, Total
-0.5%
-1.0%
-1.4%
-1.8%
-2.2%
-2.5%
. -2.9%
-3.2%
-3.4%
-3.7%
Note: Quintiles defined in terms of 1989 Census household incomes.
  VMT is vehicle-miles traveled in millions per day! Sales tax relief,
  improved transit, and other potential expenditures to mitigate
  impacts on lower income households are not reflected here.
U. S. Environmental Protection Agency
                                                                                        B-55

-------
 Technical Mefhods for Analyzing Pricing Measures
 to Reduce Transportation Emissions
      Table B2I presents results for VMT fees in the Los Angeles region at levels ranging between 1
      cent and 10 cents per mile. The STEP analysis shows that daily VMT is skewed heavily toward the
      upper income quintiles - the highest income quintile accounts for about one-third of total VMT,
      while the lowest quintile accounts for less than 10 percent. Nevertheless, the absolute drop in VMT
      resulting from a VMT fee is largest in quintile 2 (the second lowest income level) and smallest in
      quintile 5 (the highest income level). The absolute drop in VMT is of the same basic magnitude in
      each of the first four quintiles, and the percentage drop is progressively larger the lower the income.
      level. (Percentages are shown in the second part of the table).

      Table B22 presents results for congestion prices ranging from one cent to  ten cents  per mile, on
      average, for the San Francisco Bay Area. Here we find that absolute VMT decreases are roughly
      the same among the lowest four quintiles, while VMT for the highest quintile actually rises (as one
      would expect for high-value-of-time travelers).

      Comparable analyses for parking fees in the San Diego region and fuel  taxes in the Sacramento
      region (details not shown here) yielded similar results:  The largest VMT decreases per capita are
      concentrated in the four lowest income categories. While the VMT drop should not be read as a
      pure decline in mobility, since some trips are still made by other modes, shortened, etc., it does
      show how the mobility impacts would be distributed in the absence of efforts to improve modal
      alternatives for impacted individuals.
                            A
      Another way to think about equity is in terms of the per capita daily  payment by each quintile.
      Based on Table B21, the quintile total payments for a 5 cent VMT fee in the Los Angeles region
      would be:
               Quintile
              '  1
                2"'":
                3
               . 4' '
                5
Daily Payment (million
   ,    0.9
       1.8
       2.4
       3,3   •
       4.5
     Out of a daily total of $12.9 million, 35 percent is paid by the top quintile and 61 percent is paid by
     the top two quintiles. Similarly, only about six percent of current fuel taxes are paid by members
     of  the  lowest income quintile  and  10 percent by  the  second quintile.  Thus, while  the
     travel/mobility impact falls disproportionately on the lower income quintiles, the  financial burden
     falls squarely on the upper income quintiles.

     It is harder to say how VMT fees and vehicle emissions fees would affect different income groups;
     we can estimate impacts on trip making and location clroice, and can forecast auto ownership levels
     by income group, but we have no direct evidence on how the various groups would change the type
     and age of the vehicles they own in response to new fees (our analyses on vehicle type and  age
     changes were based  on assumptions provided to the  models rather than computed outputs of the
     models).  Nevertheless, available data do provide some  insights into equity impacts.  Using data
     collected  by  Caltrans as part of a statewide travel survey, we find that about 55 percent of the
     vehicles over eight years old are owned by the top three income quintiles, mostly  as second, third,
     or even fourth or fifth cars. The remaining 45 percent of the older cars are owned by the two-fifths
B-56
                                                                  U. S. Environmental Protection Agency

-------
                                                          Technical Methods for Analyzing Pricing Measures
                                                                     to Reduce Transportation Emissions

      of the households with low or moderate incomes.  To the extent that vehicle registration fees fall
      most heavily on these older vehicles, they also would fall somewhat more heavily than proportional
      on low and moderate income households.
                                                                          \
      Implications

      As with any change in tax policy, the distributional consequences of the proposed change should be
      carefully examined:  The distribution of the burden of the proposed tax among income groups
      should be compared with the distributional consequences of tax alternatives, including the current
      tax system.   If adverse equity consequences are deemed to be significant, policies to ameliorate
      those burdens should be examined, including the use of tax revenues to benefit those who might be
      disproportionately affected by the tax change (such as increased funding for public transportation
      services, use of revenues to provide tax exemptions (life-line rates) for low-income users, or use of
      tax revenues to replace existing taxes that disproportionately burden low-income taxpayers).
U. S. Environmental Protection Agency
                                                                                            B-57

-------
 Technical Methods for Analyzing Pricing Measures
 to Reduce Transportation Emissions

      References
      1.
     2.
     3.
     4.
     5.
     6.
                                                          i •
Greig Harvey, STEP: Short-Range Transportation Evaluation Program, prepared for the
Metropolitan Transportation Commission, Oakland, California, 1978.

Greig  Harvey  "Methodology  for Incorporating  Transportation System  Effects  into
Regional Transportation Energy Demand Forecasts,"  report prepared for the California
Energy Commission under Contract No. 400-82-023,  September 1983.
                                                          I
D. Pickrell, "What Gasoline Taxes Can - and Can't - Do," paper presented at the Graduate
School of Architecture and Urban Planning, UCLA, Los Angeles, March 16, 1993.
                  ;               •'        •       '         f     '
D.  Pickrell, "VMT  and  Vehicle  Fuel  Economy  Elasticities,"  Memorandum,  Volpe
National Transportation Systems Center, Cambridge, Massachusetts, February 14, 1995.

G. Orcutt, Policy Evaluation through Discrete Microsimulation, Second Edition, Brookings
Institute, Washington, D.C., 1976.

J.D. Wardrop, "Some Theoretical Aspects of Road Traffic Research," proceedings of the
Institute of Civil Engineers 2(1), 1952, pp. 325-378.
B-58
                                                                  U. S. Environmental Protection Agency

-------