EPA-450/3-74-041
April 1974
STUDY AND EVALUATION
OF COMPUTER
CARPOOL PROGRAMS
IN CERTAIN
METROPOLITAN AREAS
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Waste Management
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
-------
-------
EPA-450/3-74-041
STUDY AND EVALUATION
OF COMPUTER
CARPOOL PROGRAMS
IN CERTAIN
METROPOLITAN AREAS
By
David A. Bryant
GCA Corporation
GCA/Technology Division
Bedford, Massachusetts 01730
Contract No. 68-02-1337
Project 300927
Task Order No.3
Project Officer: Lewis D. Tamny
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Waste Management
Office of Air Quality Planning and Standards
Research Trinagle Park, N. C. 27711
-------
This report is issued by the Environmental Protection Agency to report
technical data of interest to a limited number of readers. Copies are
available free of charge to Federal employees, current contractors and
grantees, and nonprofit organizations as supplies permit from the Air
Pollution Technical Information Center, Environmental Protection Agency,
Research Triangle Park, North Carolina 27711; or, for a fee, from the
National Technical Information Service, 5285 Port Royal Road, Springfield,
Virginia 22151.
This report was furnished to the Environmental Protection Agency by
the GCA Corporation, in fulfillment of Contract No. 68-02-1337. The
contents of this report are reproduced herein as received from the GCA
Corporation. The opinions, findings, and conclusions expressed are
those of the author and not necessarily those of the Environmental Protec-
tion Agency. Mention of company or product names is not to be considered
as an endorsement by the Environmental Protection Agency.
Publication No. EPA-450/3-74-041
11
-------
ABSTRACT
This report presents the results of a survey and evaluation of computer-
assisted systems designed to increase the amount of carpooling in several
metropolitan areas. The report considers most of the ma.jor carpool
systems which serve either an entire metropolitan region or a large
number of employers in a region and which were well underway by early
1974.
In describing existing systems, the report treats not only the actual
computer programs used for matching prospective carpool partners but
also discusses organization, history, marketing, incentives and dis-
incentives, and results to date. In addition, the report contains
general discussions of related elements such as legal questions,
insurance, benefits of carpools and of carpool systems, and the theoret-
ical potential for carpooling. Theoretical and general considerations
are compared with actual practice throughout the report.
From the review of experience to date it is concluded that certain types
of carpool systems have the potential for increasing the amount of
carpooling and reducing the amount of air pollution from automobiles,
throughout a metropolitan area. Additional study and research on car-
pool systems are recommended, however, because computer-aided carpool
systems are relatively new and were found to be rapidly evolving.
Included in the report are recommendations for organizing and operating
a carpool system to serve a metropolitan area. These recommendations
in
-------
were developed largely from analysis of the actual carpool system
experience reviewed for the report.
In addition, an annotated bibliography and list of information
sources are provided to aid those who wish to organize, design,
operate, or study carpool systems.
This report was submitted in fulfillment of Task Order Number 3,
Contract Number 68-02-1337, by the GCA Corporation, GCA/Technology
Division, under the sponsorship of the U.S. Environmental Protection
Agency. Work was completed in April 1974.
IV
-------
CONTENTS
Page
List of Figures vi
List of Tables vii
Acknowledgments viii
Sections
I Background on Carpool Systems 1
II Benefits of Carpool Systems 7
III Matching Method and Data Base 11
IV Matching Software 58
V Organization of Carpool Systems 91
VI Marketing 100
VII Incentives and Disincentives 116
VIII Review of Other Issues 128
IX Estimating the Potential for Carpool Systems 134
X Summary and Recommendations 148
XI References 163
XII Annotated Bibliography 167
-------
LIST OF FIGURES
No. Page
1 Geographic Coding Concepts 61
2 Cluster Search Method in Denver, Colorado,
Carpool Matching Program 83
3 Vector Search Method in Denver, Colorado,
Carpool Matching Program 84
VI
-------
LIST OF TABLES
No. Page
1 Air Quality Control Regions with Carpool Systems in
State Implementation Plans as of 30 November 1973 4
2 Characteristics of Selected Boston Area Employer-
Operated Destination-Based Carpool Systems 20
3 Summary of Incentives 127
4 Differences in Travel Behavior between Carpools and
Non-Poolers on Hollywood Freeway, 1972 140
Vll
-------
ACKNOWLEDGMENTS
Many individuals and several organizations have provided valuable
assistance and information to this study; for these contributions
GCA/Technology Division extends its sincere gratitude.
Project direction and guidance were provided by Mr. Dave Tamny of the
Land Use Planning Branch, EPA, Durham, North Carolina.
Vlll
-------
SECTION I
BACKGROUND ON CAKPOOL SYSTEMS
BASES FOR THE INTEREST IN CARPOOL SYSTEMS
The Clean Air Amendments of 1970
The present interest of the U.S. Environmental Protection Agency (EPA)
&
in carpool systems originates from the Clean Air Amendments of 1970
(Public Law 91-604). Section 109 of the amended Clean Air Act requires
the Administrator of EPA to prescribe ambient air quality standards
for air pollutants. Section 110 of the Act requires that every state
prepare and submit to EPA an implementation plan by which the air
quality standards will be met. Section 110 (a)(2)(B) specifically
allows the state implementation plans to include controls on trans-
portation and land use.
As the state implementation plans were developed and submitted to EPA,
the Administrator of EPA designated the regions where transportation
control plans would be needed to meet air quality standards for carbon
monoxide and oxidants. As the development of these plans has pro-
gressed, the list of regions needing transportation control plans
has steadily lengthened. At the present time more than 40 air quality
control regions will require transportation control plans; many of
these plans have been prepared, submitted to EPA, and approved.
The focus of this report is carpools, but much of the information
also would apply to vanpools or buspools. Refer to page 120 for
further comment on vanpools and buspools.
-------
Fuel Shortages
In addition to the interest in carpooling provided by the Clean Air
Amendments of 1970, the shortages of automotive fuels in some areas
of the U.S. during 1973 and early 1974 provided additional impetus
to organized carpool systems. In several cities, new carpool systems
were organized by radio and television stations, major employers,
civic groups, or government agencies, as part of the effort to cope
with the fuel shortages. Some of these efforts were directed towards
the general public, whereas others were intended for particular
groups, such as employees of the sponsoring employer. Several of
these programs have enlisted thousands of participants.
Also in response to the fuel shortages, the Congress passed the
Emergency Highway Energy Conservation Act. Among other steps, this
Act authorized the U.S. Secretary of Transportation to (1) provide
financial support to "demonstration projects designed to encourage
the use of carpools in urban areas" and (2) study (in cooperation
with other agencies including EPA) what measures "might lead to
2
significant increases in carpool ridership in urban areas..."
As a result of this Act, the U.S. Department of Transportation (DOT)
organized and conducted a series of carpool seminars throughout the
U.S., in which DOT presented guidance to assist state and local
agencies to establish carpool systems for the purpose of conserving
fuels. The Highway Users Federation for Safety and Mobility (HUFSAM)
is conducting a similar series of seminars aimed at industrial
organizations.
As part of the U.S. Government's effort, the Secretary of Transporta-
tion asked that each regional transportation planning agency submit
an emergency transportation program by May 1, 1974, in which the
agencies specify low-cost, short-term projects which will help to
-------
reduce transportation fuel consumption. Such projects may include
carpool matching systems, exclusive lanes for carpools and/or buses,
carpool parking lots, and similar actions. Federal funds were made
available for some such emergency projects, including carpool match-
3
ing systems as mentioned above.
As a result of all these efforts, there is a great deal of activity
related to organizing carpool systems at the present time throughout
the U.S. Thus it can be seen that, as a result of (1) the efforts
to comply with Clean Air Amendments of 1970 and (2) the efforts to
cope with present fuel shortages, there are dozens of cities in the
nation which either have or shortly will have organized carpool
matching systems and campaigns to organize carpools. This report
will examine the experience to date with such carpool programs and
identify the characteristics of an effective program.
STATUS OF CARPOOL SYSTEMS
Many of the transportation control plans (TCP's) proposed or
approved to date have included carpool matching systems as one of
the strategies to reduce the amount of pollutant emissions from motor
vehicles. Some of the Air Quality Control Regions (AQCR's) with car-
pool systems in the implementation plans are listed in Table 1.
Most of the promulgated plans do not, however, include detailed re-
quirements for establishing the carpool matching systems. For ex-
ample, the published regulations for the Boston AQCR state:
§ 52.1138 Regulation for computer carpool
matching.
(a) "Carpool matching" means assembling
lists of commuters with similar daily travel
patterns and providing a mechanism by which
persons on such lists may be put in contact
with each other the purpose of forming carpools.
-------
Table 1. AIR QUALITY CONTROL REGIONS WITH CARPOOL SYSTEMS
IN STATE IMPLEMENTATION PLANS AS OF 30 NOVEMBER 1973
State
AQCR
Ari zona
California
District of Columbia
Indiana
Maryland
Massachusetts
New Jersey
Pennsylvania
Texas
Washington
Phoenix-Tucson
San Francisco Bay Area
Metropolitan Los Angeles
San Diego
Sacramento Valley
San Joaquin Valley
National Capital
Metropolitan Indianapolis
Metropolitan Baltimore
Hartford-New Haven-Springfield
Boston
New York, New Jersey, Connecti-
cut, Metropolitan Philadelphia
Metropolitan Philadelphia
Southwestern Pennsylvania
Houston-Calveston
Dallas-Fort Worth
San Antonio
Puget Sound
Eastern Washington
Northern Idaho
Source: U.S. Environmental Protection Agency
-------
(b) This section is applicable in the
Boston Intrastate Region if the Automobile
Legal Association/WBZ "Commuter Computer
Club Car" is discontinued.
(c) The Commonwealth of Massachusetts
shall, unless otherwise exempted by the
Administrator on the basis of a finding
of the continued existence of equivalent
(private) service, establish a computer-
aided carpool matching system that is
conveniently available to the general
public and to all employees of employers
having more than 50 employees within the
Intrastate Region who operate light-duty
vehicles on streets and highways over
which the Commonwealth has ownership or
control. No later than 3 months after
discontinuation of the "Commuter Computer
Club Car," the Commonwealth shall submit
legally adopted regulations to the Admin-
istrator establishing such a system. No
provisions of such regulations shall have
an effective date later than 3 months from
the date of adoption. The regulations
shall include:
(1) A method of collecting information
that will include the following as a
minimum:
(i) Provisions for each affected em-
ployee to receive an application form
with a cover letter describing the match-
ing program.
(ii) Provision on each application for
applicant identification of commuting
time, origin, and destination, and the
applicant's desire to ride only, drive
only, or share driving.
(iii) A computer method of matching
information that will have provisions
for locating each applicant's origin
and destination within the Boston Intra-
state and the Interstate Regions and
matching applicants with similar origins
and destinations and travel schedules
and enabling the persons so matched to
make contact with each other at the re-
quest of any one of them.
(iv) A method of providing continuing
service such that the matched lists of
-------
all applicants are retained and made
available for use by new applicants,
application forms are currently avail-
able, and the master lists are period-
ically updated to remove applicants who
no longer meet the governing criteria
and add new applicants who do.
(v) Designation of an agency or agen-
cies responsible for operating, over-
seeing, and maintaining the computer
carpool matching system.
In short, the regulations require the Commonwealth of Massachusetts
to make available a carpool matching system to all employers in the
AQCR with 50 or more employees. The regulations specify the general
types of information on which matches should be based, but otherwise
are not very detailed regarding how such a system should be organized
and run. Regulations for carpool matching systems in other AQCR's
are similar.
Because such computer-based carpool matching systems are relatively
new, and because the widespread use of organized carpooling has never
before been attempted on this scale, there is a need for in-depth
development of detailed guidelines for organizing carpool matching
systems. This report is addressed to that need.
-------
SECTION II
BENEFITS OF CARPOOL SYSTEMS
BENEFITS RELATED TO AIR QUALITY
There are a number of potential benefits associated with widespread
use of carpools. Some of these are public benefits--that is, they
accrue to society as a whole; and others are private benefits--that
is, they accrue to the individuals involved.
Chief among the expected public benefits of widespread use of car-
pools would be benefits related to air quality. It is generally
acknowledged that private passenger cars are the predominant source
of several air pollutants in many urban areas. Consequently, the
use of carpools, or any action which reduces the number of vehicles
being used, would reduce pollutant emissions and improve air quality.
More specifically, there are two ways in which the increased use of
carpools would improve air quality. First, the use of carpools would
result in fewer vehicles using the road system at any given time
(during rush hours, for example). With fewer cars on the road at a
given time there would be less congestion and improved traffic flow.
Improving traffic flow, by increasing speeds and reducing the number
of starts and stops, reduces the amount of pollutant emissions per
vehicle-mile. Second, with fewer vehicles being used there will be
less total pollution in a given geographical area.
-------
Both these effects would be most pronounced in peak traffic periods,
inasmuch as most carpool programs are intended for commuting to work.
Total daily automotive pollution output could be reduced still further
by carpools for non-work purposes, such as shopping, school, and
recreation.
Another public benefit could be the use of widespread carpoolirig
instead of an alternative transportation control strategy in those
Air Quality Control Regions (AQCR's) which need transportation con-
trol plans in order to improve air quality enough to satisfy Federal
standards. Carpooling may be more attractive to the public and more
cost-effective than alternatives such as substantial and costly
additions to mass transit systems. Furthermore, the use of carpools
can be rapidly and readily instituted, whereas alternatives such as
new transit systems require long periods of time to implement.
As an example, consider a hypothetical city in which 120,000 people
now commute to work by car. Typically, such a city would have an
average of about 1.2 persons in each passenger car used for commuting
6
to work. Thus in the hypothetical city there are 100,000 passenger
cars used for travelling to work each day. Suppose further that in
order to meet air quality criteria it is necessary to reduce the
total amount of work-related travel by passenger cars in the metro-
politan area by 10 percent, as one part of the transportation control
plan. To achieve this reduction by shifting people to buses would
require providing enough buses for 10 percent of the commuters, or
12,000 people. To carry 12,000 commuters each way to work in the
peak hours would require 120 buses, assuming each bus could make two
runs in the peak period and could carry 50 passengers. The capital
cost of 120 buses would be approximately $5,000,000. Operating costs
would include the wages for the drivers and dispatchers, plus fuel,
maintenance, storage, licensing, and insurance for the buses.
-------
As the alternative, the 12,000 commuters could be carried in carpools.
On the average, the number of persons in each car would increase from
1.2 to 1.33, so on the average one car in three would have passengers.
It would be necessary to match 12,000 riders with drivers. The public
cost would be the cost of operating the carpool matching program and
advertising campaign; in many cities this is being done at no charge
as a public service by auto clubs, radio stations, or employers; else-
where the cost is estimated to be less than $100,000 for a two-year
initial effort.
This example, while simplified, demonstrates the attractiveness of
carpooling as an element in the transportation control plan.
OTHER ADVANTAGES
Besides the air quality advantages of increased carpooling, there
are other benefits as well, both public and private.
An important public benefit of carpooling could be reduced energy
consumption, provided that the vehicles no longer being used for
driving to work are not used for other, counter-productive trips.
If leaving cars at home should lead to increased travel by members
of the family besides those in carpools, total travel may not be
reduced as much and air quality would not be improved very much.
Net energy reduction from carpools will be greatest if such addi-
tional trips are discouraged, and if carpooling can actually result
in reducing the number of cars in use (i.e., if some people reduce
the number of cars they own). The Highway Users Federation has
estimated that a 25 percent increase in carpooling would reduce the
consumption of petroleum by nearly 2 percent.
A private benefit of carpooling is the direct monetary savings for
the commuter. Obviously the amount of savings would depend on many
-------
elements, such as car size, parking costs, number of people in car-
pool, distance driven, and so on. Compared to driving alone, car-
pooling reduces costs in direct proportion to the number in the pool.
Carpooling is less expensive than transit in many cases, and also
may be more convenient as well.
Additional benefits for the participants in carpools may occur as a
result of campaigns to encourage carpools. Some employers provide
incentives to carpoolers, such as cash payments, preferential park-
ing, or flexible working hours. Government agencies can and do pro-
vide additional incentives, such as the use of exclusive highway
express lanes, preferential treatment at interchange ramps, prefer-
ential parking in public facilities, and monetary incentives such as
reduced tolls on turnpikes.
SUMMARY
Widespread use of carpools can provide substantial benefits both to
the public and to the individuals involved. Carpooling systems offer
an attractive and inexpensive alternative to other means of reducing
air pollution through reductions in travel. The attractiveness of
the use of carpools results in part from the ease with which they can
be implemented, compared to alternatives such as enlarged transit
systems.
10
-------
SECTION III
MATCHING METHOD AND DATA BASE
INTRODUCTION TO MATCHING
General
Central to the discussion of a carpool system is the structure of the
process by which prospective carpool members are identified to each
other, or matched. In this matching process, information about people
who have expressed an interest in carpools is manipulated in a way to
identify likely candidates for carpools. Typically, the matching pro-
cess uses information such as employee work schedules (starting and
quitting times), home and work locations, and perhaps preferences re-
&
garding smoking and sharing driving responsibility.
The success of a campaign to encourage carpooling is strongly dependent
upon the number of potential carpools which can be identified by the
matching process. The chance of identifying a potential match in turn
depends upon how large the data base is (the data base is the collection
of potential carpoolers) and upon the amount of variability in home
addresses, work locations, work schedules, and preferences. For a given
size data base, the chances of a match decrease rapidly with increases
in the number of combinations of work schedules, home addresses, work
locations, and preferences. For a fixed number of home addresses, work
locations, schedules, and preferences, the chance of matching increases
*
Throughout most of this report, the carpool systems being discussed
are intended to serve the journey to work. Carpool systems for other
trip purposes are briefly discussed later in this Section (page 52).
11
-------
rapidly with the size of the data base. (These relationships are
discussed in greater detail in Section IX.)
Destination-Based Matching
In the past, most carpool systems were used for only one destination,
and often for employees on only one or two work schedules. For
example, a single employer would provide a system for his employees.
With such a destination-based system the'matching problem typically
is reduced to that of identifying people who live near each other and
desire to form a carpool. Relatively high chances of matching could
result from even a few hundred employees at one site.
Regional Matching
With the current interest in encouragement of carpooling throughout
an entire metropolitan area in order to achieve substantial reductions
in pollutant emissions (and in fuel consumption), attention has
focused on matching commuters throughout the area. This type of
matching potentially could have a large data base but also is more
complex than single-destination matching because now multiple desti-
nations and multiple work schedules are added variables.
For example, a metropolitan area could be divided into a grid for the
purposes of matching. Such a grid would typically have several
hundred or even a few thousand elements. There may be 50 or more
destinations (work locations). As many as 10 separate work schedules
could be involved. Altogether, the number of combinations of home
grids, destinations, and work schedules could be on the order of
hundreds of thousands. Hence enormous additional complexity accom-
panies the expanded data base.
12
-------
Combination Matching
A third type of matching system combines the features of the single-
destination and regional systems. In a combination system matching
is done for specific organizations or destinations and also for the
region as a whole. The data on individual employers may or may not
be part of the data base for regional matching. Such systems attempt
to capitalize on the simplicity of destination-based matching while
at the same time serving the smaller employers who require a regional
data base for effective matching.
Because regional and combination matching systems are relatively new
and involve potential complexity, this section will separately
review the experience with destination-based, regional, and combina-
tion matching systems. In addition this section will discuss special-
purpose matching for non-work purposes.
EXPERIENCE WITH DESTINATION-BASED SYSTEMS
General
Destination-based carpool matching systems have been in use for
several years,. Many systems have been manual, using self-service
matching techniques based upon maps with pins and cards, or personal-
ized matching through organization employee address files. In recent
years, several employer-operated systems have been computerized for
efficiency. The use of such systems was prompted by various circum-
stances. Probably the most common reason for initiating a destination-
based carpool matching system in the past was parking limitations.
For example, in Washington, D.C., carpool matching map systems have
been common for many years because of the large numbers of people with
common work locations combined with the lack of sufficient parking
for all employees. Many Washington agencies have had a long-standing
practice of preferential parking for carpools.
13
-------
Destination-based systems are most commonly managed by an employer.
Another arrangement is for several organizations in one building., or
perhaps in a few buildings at one location, to pool their data and
match together. Such systems can be operated by outside agencies as
a service, also, as is the case with at least one of the broadcasting
station systems currently in operation.
Destination-based systems have been proven in concept, are relatively
simple to operate, and offer efficient matching with modest databases.
To illustrate the use of such systems, several cases will be reviewed.
Case Histories of Destination-Based Systems
General - This section will briefly summarize the experience with
several destination-based systems. Because of the current surge of
interest in carpooling, several recent reports have described numerous
such systems in detail (see Bibliography). This report will not
attempt to duplicate existing literature but instead will only high-
light illustrative examples, and will concentrate on recent develop-
ments and on comparisons between destination-based and regional
systems.
Federal Highway Administration, Washington, D.C. - One of the most.
widely used computer programs for carpool matching is one prepared by
the U.S. Federal Highway Administration (FHWA). One of the first
applications of this computer program was at the FHWA headquarters in
Washington, D.C. In 1972 the FHWA used the program to provide lists
of potential carpool partners to employees who expressed an interest
in forming carpools. Shortly thereafter, a follow-up survey was done
to assess the impact of the matching service.
The initial carpool questionnaire was sent to approximately 1,200
employees in one building. The forms were returned by 780 (65 per-
cent) of the employees, of whom 550 (46 percent of the total) asked
14
-------
to be included in the matching process. The FHWA computer program
was then used to prepare lists of potential carpool partners for
each interested person. Reportedly, the majority of the participants
received a substantial number of potential matches.
From the follow-up questionnaire (distributed two months after the
carpool matching lists were distributed) the following results were
reported by FHWA:
" A. About 65 percent of FHWA employees commute in carpools.
B. FHWA employees averaged 2.34 persons per auto before
the program, 2.45 after. This small increase (5 percent)
is thought to be due to the fact that the occupancy
rate was so high to begin with that 'room for improvement'
was slight....
C. Bus ridership among the 980 respondents decreased from
197 (20 percent) before to 169 (17 percent) after. All
of these 28 persons joined carpools. Only 18 of them,
however, indicated that the matching program helped them.
It is not known how many of these 18 would have joined a
carpool without the help of the matching program....
E. Of the 980 responding, 360 (35 percent) said they had bene-
fited in one way or another from the program....
F. Of the 360 who said they had benefited, 115 (30 percent)
actually joined a new carpool, increased their existing
carpool or changed members in their carpool due to the
program."8
In this instance, carpooling was already so prevalent before the
matching process began that little increase was observed in the
amount of carpooling. Many people rearranged their carpools to be
more convenient or otherwise suit them better, however, and others
retained the matching lists for future reference for adjusting car-
pools .
Carpooling was very popular at FHWA headquarters before the matching
process began; because of a shortage of parking spaces, carpools have
15
-------
been given preference in the FHWA parking garage. Consequently the
matching process was not widely requested. Initially one might be
surprised that only 46 percent of the employees expressed a desire to
be matched, inasmuch as parking is in short supply; furthermore one
would expect a public transportation agency to have a vested interest
in demonstrating the usefulness of its own matching system. Many
people may have declined to participate because they already were in
an acceptable carpool. Consequently one might expect a higher rate
of return under conditions where carpooling is not well established.
Another way to look at this case is to conclude that carpooling will
occur even without a sophisticated matching system if the incentives
are adequate; in this case parking was substantially easier for car-
pools .
The same FHWA computer program has subsequently been distributed to
many other agencies and firms. (This will be further discussed in
Section IV.)
Burroughs Corporation - Another computer program now widely used is
the "Operation Energy" program first used at the Medium Systems Plant
of the Burroughs Corporation in Pasadena, California. Full details
on the response rate from employees have not been published, but
Burroughs reported that parking demand at the Pasadena plant dropped
from 659 cars to 427 cars, a 35 percent decrease. This substantial
reduction occurred as a result of a comprehensive campaign which
9 10
included incentives such as preferential parking. ' This reduc-
tion in parking demand is among the most dramatic observed changes
in travel patterns reported as a result of carpool matching programs
in the U.S.
Burroughs has made available their computer program and a description
of their carpool campaign; copies have been distributed to numerous
other firms and government agencies. The computer program, as well
16
-------
as other aspects of their campaign, will be further discussed in
subsequent sections of this report.
Station WBBM, Chicago - Radio station WBBM in Chicago, a CBS network
affiliate, has organized a different approach to destination-based
carpool systems. Their carpool campaign, known as the "News Radio 78
Work Together/Ride Together" carpool system, operates solely on
destination-based matching. Most of the other radio and television
carpool campaigns have been based at least in part upon regional car-
pool matching. The station has arranged to perform the carpool
matching service for a substantial number of companies and govern-
ment agencies, with a potential data base of tens of thousands of
Chicago area employees. The campaign was begun in the winter of
1973-1974, and has not been in operation long enough to permit an
evaluation of its response rate. Apparently, however, the station
has attracted firms with a significant fraction of Chicago area
employment; if the matching process is widely used by the employees
the campaign could be more effective than most regional carpool
systems have been.
Each employer is separately matched in the WBBM system. All distri-
bution and collection of questionnaires and publicity within the
firm are handled by the employer, so in effect the campaign operates
as a collection of separate destination-based systems. As an
incentive to the companies, WBBM compliments participating organiza-
tions on the air.
A unique feature of this is that the computer matching is performed
by the station itself, whereas most other broadcasting stations use
an outside source for computer aspects of their matching systems.
Manual Systems - Besides the relatively recent destination-based
systems using computers for carpool matching, other destination-
based systems based upon manual matching devices have been in
17
-------
operation for at least several years. A recent report to the
Federal Highway Administration entitled Manual Carpool Matching
Methods, reviewed in the Bibliography of this report, describes
several systems.
Most manual matching was based upon some sort of map and card system.
Matching was actually done by the participants themselves, in most
cases. In other cases, a central office performed the matching
function, often in an informal way such as with lists of employees
arranged by home town or zip codes.
FHWA considers that such manual systems are adequate for firms wLth
up to 1,000 employees. Manual systems have been used for larger
numbers of people, however. For example, Hallmark Card Company used
a manual system to match 2,500 of their 4,500 employees. The systems
serving the most people are probably those at the Pentagon in
Arlington, Virginia (upwards of 25,000 people) and at McDonnell-
Douglas in St. Louis (up to 47,000 employees), but it is unlikely
that a large fraction of these employees actually used the system.
At McDonnell-Douglas, the use of the carpool matching system
reportedly provided an average vehicle occupancy of 2.8 when employ-
ment was at its peak, but occupancy has declined to 1.8 in the
12
recent past as employment dropped to 25,000. " This may show the
effect of incentives, inasmuch as the preferential parking provided
for carpools at McDonnell-Douglas would obviously be more desirable
with a greater amount of competition for parking space.
Destination-Based Carpool Systems in Boston Area - As part of the
effort in this study to investigate the WBZ-ALA carpool matching
system in Boston (described later in this section), several major
employers were contacted to determine their opinion of the WBZ-ALA
system. It was discovered that a number of large employers had
decided to conduct their own internal carpool matching service
18
-------
either instead of or in addition to encouraging use of the WBZ-ALA
regional system.
Several large firms in the Boston area were surveyed and were found
to be using a manual or semi-manual matching system, rather than an
automated matching system. The firms investigated are using the
computer only to generate lists of potential carpool candidates by
town of residence, not to actually identify potential matches for a
particular person.
Two basic approaches are being used by the Boston area firms which
were studied. In one approach, lists of people who have actually
expressed an interest in carpools are prepared and either made avail-
able on request or distributed to others who are interested. In the
other approach, lists of all employees, arranged by home address,
are made available for reference by anyone interested in forming a
carpool.
It will be useful to discuss several of these employer-based carpool
programs as an illustration of what is currently being done without
the use of matching computer programs and for comparison with regional
matching systems. Five Boston area employer carpool systems were
reviewed. Four are operated by companies only for their own employees;
the fifth is a joint venture of several companies in one neighborhood.
Table 2 summarizes the carpooling characteristic of these five
systems.
*
Company A, a large manufacturing firm with several plants, used a
survey in January 1974 to determine the degree of interest in car-
pooling among its employees. At present, over 90 percent of the
employees come to work by car, and parking is abundant and free of
To preserve the confidential nature of some of these data, companies
are referred to in the text and tables only by pseudonyms.
19
-------
CM
(U
r-l
B
O
H
U
5>
H
n
01
a
H
X
o
I
w
0.
^4
X
d
a
f
0
o
d
«
Q.
8
U
O
g1
a
0.
o
o
«
d
o.
1
o
*
tn
01 X >>
IW A JQ
a.
CD "O
0) 01 O
-I ^ CJ
a. cu
o o *J a
\D 0) d -H
r- 04 *H N
0>
"O
O
CJ
00
d a.
H *H
CO N
CO
0) X
l-l .0
a
a> u *j
01 0) i-l
< ' *-* .-C
CL CU CD
B-2 O 4J
in a> C X
^"»
XI
*
n
V
0)
o
t-i
Q.
g .
C
< S 0
Z
ai r-( d
0>
u
0)
CU
M-t
£
0) O
4J
3 "
5 5* 0
j-i ^ CM
4J « c»
to o oi oi
0 4J J3 C
g ^j at
"i 2 1 ?
t-J
% in
ID CO
"4 4-1
O CD
v^- (U
r-. C
01 4 O CO
0) -H CJ 0)
t-l JJ N
iy M J-i
w C a.
0) iu <-<
0 iw qj -H
*
>.
* -Q
w
CU 13
SO) U
4J a)
WOO
O r-l li
O » Q, d
O r-l Q O
m « 5 K
o
«
> *
n oi
01 01
« O
s
o S u
o d
-i XI Z
.
01
2
m
01
3. S
4 £
"U
0)
l-(
^ 0
f\, ,tJ
Q-
3 CO
» g
> CO
bO -H T-i
C u 0
H do
** S
« d S
0- MO
(U
u
M
n
AJ
^H
d
a
o
z
4J
4J O
P IW
4J m 1-1
CD CJ
3 4J
0 o «
6-S O M
rg ^ *w
^
0 X
n d
"I
00 U
^ 5 S
* CJ
ftO w te
? « o*
- m
ti » eg *
4) 73 > CD
m -
H oi o) > d
(0 CL O
CO 0)
S 4)
O CD
d
^ S
d £
«
w y-v
O 0)
SO)
0)
5 a
S S
e
S
a
s
o
«J
(0
ej
o
0)
CJ 4^
M n
20
-------
charge. Interest in carpooling was expressed by 26 percent of the
employees. Lists were generated of the employees who expressed an
interest. These lists are arranged by home address zip code; there
are separate lists for each employment location. These lists are
available from a carpool coordinator at each site (only the list for
the particular site is available). No incentives have been provided,
but the program has been widely publicized through the company news-
paper, bulletin board displays, and special letters from the manage-
ment to the employees. The WBZ-ALA regional carpool program was also
publicized. It is not yet known how many employees have formed car-
pools, but a follow-up survey is now being planned.
Company B is a nort-manufacturing organization with 6000 employees at
one site. Carpool information has been distributed in the company
newpaper for more than fifteen years. In the recent past this ser-
vice has been expanded. Notices of carpool vacancies or interest in
carpools are exchanged among nearby firms (including company C), with
a total employment of over 10,000 people. Notices from all the
firms are published in each company newsletter. In addition, a com-
muter advisor service office has been established which dispenses
information and assistance regarding carpools (and transit). This
office maintains lists of employees by home address zip code and by
work schedule for the use of persons who wish to contact others from
the same town or adjacent towns. The WBZ-ALA program has also been
publicized and questionnaires were distributed. Recent surveys have
shown that nearly half of the employees commute by car and about 40
percent by transit. Parking is available, at a moderate cost, for
1,100 cars in company-owned facilities. Average vehicle occupancy
is reportedly about 2.5 persons per automobile, with less than 500
persons driving alone. No explicit incentives are provided for car-
pooling, but the limited parking space available operates as an
incentive. Data are not yet available on whether the amount of car-
pooling has increased with the recently increased publicity and
effort, but additional surveys are planned.
21
-------
Company C is another non-manufacturer, with approximately 2000
employees at one location. This firm has also carried carpool
notices in its company newspaper for several years, and in the winter
of 1973-74 expanded this to include reciprocal listings with Company
B and a third firm. An administrative office maintains lists of all
employees, arranged by hometown, for the use of persons who want to
contact others in order to form carpools. The availability of this
list has been publicized, and the WBZ-ALA matching service was pub-
licized and the questionnaires were distributed. Parking is provided
in a company facility for 500 cars. A survey in the fall of 1973
showed an average of approximately 1.8 persons per car. No explicit
incentives are provided, and parking spaces are allocated by company
management.
Company D is a manufacturing firm with over 3000 employees at one
location, working on a three-shift basis. Ample free parking is
available, and an October 1973 survey showed that more than 70 per-
cent of the employees commuted by automobile, and 26 percent via
transit. The WBZ-ALA program was not publicized; instead, the
October 1973 survey asked people to express their interest in car-
pools. Approximately 25 percent expressed an interest in carpooling.
Lists were prepared of those who expressed an interest (only) and
distributed. Each person received a list of other people living in
the same zip code area, and people were advised they could also
obtain lists for adjacent zip codes or areas on their route to and
from work. Preferential parking (i.e., parking substantially nearer
the building than most) was instituted for carpools. Substantial
publicity was used to encourage carpooling, including contests,
prizes, advertising, and special carpool registration periods in the
company cafeteria. Only about a dozen carpools existed before the
campaign, but by February 1974 there were over 100 carpools with a
total of 275 people. Most carpools are among day shift personnel.
Average vehicle occupancy is now approximately 1.1 whereas before
the carpool program it was barely in excess of 1.0.
22
-------
Company Group E is a group of firms located in a suburban industrial
park area who collectively organized a carpool system. Ten manu-
facturing and research firms originally discussed participating; so far
six have submitted completed questionnaires and been processed.
The participating firms have a combined employment of 3,700; of these,
20 percent expressed an interest in carpooling and were processed.
The processing consisted of (1) combining the replies from all the
participating firms and (2) printing lists of people who expressed
an interest in carpooling, arranged by hometown and zip code. The
lists were sent to the personnel officers at each firm; there, the
lists are available for examination by anyone interested in carpool-
ing. The lists show the employee name, employer name, work schedule,
and preferences regarding smoking, ride or drive, and frequency of
carpooling. Promotion of this effort varied among the firms but
appears to have been modest; ample free parking is available and no
incentives have been provided. Matching lists were distributed in
mid-March so no data are available concerning the number of carpools
formed. Some of the firms had earlier publicized or distributed the
WBZ-ALA questionnaire. The area is not served by transit so nearly
all employees now commute by car.
There are several common elements in all five of these Boston area
destination-based systems. These elements are:
The decision was made to not rely upon the WBZ-ALA
system. All but one of these firms made the WBZ-ALA
questionnaires available but all decided that more
effective results could be obtained by in-house
efforts. The most commonly voiced reasons for this
preference were:
a. The in-house system would eliminate the
problems of disclosing employee data (such
as home address and telephone number) to
people - strangers - outside the firm.
Home addresses were used for matching only
by the administrative personnel who
already possess this information, and
23
-------
generally only the town or zip code and
office telephone are disclosed to poten-
tial carpool partners.
b. The in-house, personal approach would be
more effective than the anonymous and
impersonal WBZ-AIA system. In-house
systems would provide face-to-face contact
with the administrators of the system and
with potential carpool partners.
c. The WBZ-ALA system had not proven to pro-
duce productive matches; the matching
rate is only 25 percent (see the separate
discussion later in this section). Using
a home address list allows people to check
the lists of other towns or other zip codes
within their towns, thus increasing the
chance of a match.
2. Matching is performed in an informal, flexible, unsophis-
ticated manner. Using lists of employees by towns or
zip codes amounts to customized matching in that each
person can use his own criteria for selecting potential
partners. Partners can be chosen from nearby locations
or from locations on route to work. Little or no com-
puter programming was required.
3. Matching is performed on the basis of hometown or zip
codes, not on a grid or other system of home address
coding. This system is perhaps more logical in the
Boston area than it would be outside of the north-
eastern U.S., because of the local government structure.
In Massachusetts, for example, there is no unincorporated
territory; all the state's land area is in organized
towns or cities. People identify strongly with their
municipality. Furthermore, these towns and cities are
fairly small, physically. For example, the Boston Standard
Metropolitan Statistical Area, in which most Boston area
employees live, had a population in 1970 of 2,730,228 but
consisted of 78 municipalities. In effect this means
that even in a large firm there would not be an awkwardly
long list of employees for any one community, except
perhaps for Boston and one or two other nearby cities.
Most of these larger communities are, however, well
served by mass transit so carpooling interest is small.
Furthermore addresses in the larger communities can be
sorted by zip codes. The zip codes generally follow
neighborhood demarcations in these larger communities
and are thus perhaps more logical for coding addresses
24
-------
than they would be elsewhere. In addition, it should
be noted that there is a one-to-one relationship between
zip codes and municipalities. Each city or town is
uniquely identified by a zip code. There are only nine
larger communities with multiple zip codes in the Boston
area. This observation will be important when the WBZ-ALA
program is discussed. Thus, sorting by zip code or home-
town produces matching lists with reasonable numbers of
names.
4. Where surveys have been taken among these firms, only
a minority (roughly one-fourth) of the employees have
expressed an interest in carpools.
5. With one exception, firms have provided no explicit
incentives for their employees to form carpools.
6. Vehicle occupancy is measurably higher where parking
space is restricted in availability or not provided free.
7. Firms A, B, C, and D all are now or have been involved
with other commuter assistance functions. All had
been informally providing carpool information, or provid-
ing information on transit service. Two firms have also
organized inter-plant company shuttle bus service as a
result of the same economic forces which have encouraged
carpools. Also, all have initiated carpool efforts in
recognition of the requirement in the Boston area
Transportation Control Plan that major employers must
reduce their automobile parking space supply by 25 percent
in the near future; some of these firms explicitly expect
to accomplish this through their carpool programs.
8. Companies A, B, C, and D all have extensively publicized
their carpool programs, and all but Company D also pub-
licized the WBZ-ALA program.
It is reasonable to conclude that these Boston area carpool systems
are all operating successfully in the sense that they are apparently
able to cope with the present demand for carpooling information. All
provide information systems which are flexible and readily available.
But it is not possible to conclude that these systems could provide
enough matching information if most people in such large firms
simultaneously wanted carpool information, if gasoline should
suddenly become less available or when the mandated parking space
25
-------
supply reductions become a reality. Furthermore, it is not clear
whether the overall company carpool campaigns by themselves are
adequate to encourage large-scale carpooling. What is clear is that
both preferential treatment of carpools (as at Company D) and
restricted parking supply (as at Company B and to a lesser extent at
Company C) can substantially increase vehicle occupancy above the
urban average, even without elaborate carpool matching systems, at
least among employees of larger firms.
Summary of Experience with Destination-Based Systems
Systems designed to facilitate the formation of carpools among
employees working at one location have been in use for many years.
In the recent past the need for carpool matching systems has
increased, because of declining transit service, dispersal of employ-
ment throughout metropolitan areas, and because of the rapidly rising
costs and problems associated with automobile travel. The destination-
based carpool matching systems have simultaneously become more
sophisticated, using computers to perform the matching function.
Experience with several modern matching systems shows they can be
effective if combined with suitable promotion and incentives.
Experience in the Boston area suggests that relatively simple,
customized destination-based carpool matching can still provide
effective matching services even for large numbers of employees (up
to a few thousand at one location). But experience has not yet shown
whether such systems can cope with the amount of carpool matching
needed to substantially reduce vehicle use by commuters. Nonetheless,
experience with selected firms shows the potential for increased
vehicle occupancy if incentives and adequate matching services are
both provided.
26
-------
EXPERIENCE WITH REGIONAL SYSTEMS
General
Systems to match potential carpool partners throughout a metropolitan
region are relatively new. The concept is to establish an informa-
tion system to identify people who (1) live near one another, (2)
work near one another, and (3) work similar schedules.
9
The potential advantage of "regional matching is that it can provide
more people with potential carpool partners than destination-based
systems usually can. For example, two people who live near each
other and work near each other would never be matched if they worked for
two firms with individual destination-based carpool matching systems.
This is particularly important for employees of small firms and
residents distant from their work place; such people have low proba-
bility of matches on destination-based systems. Furthermore,
regional systems can provide everyone equal opportunity to participate
in a matching system, regardless of whether his employer provides
carpool matching services. In theory, a regional system could provide
the same number of matches to employees of large firms as would a
destination-based system, if the same people participate, because the
data base would be the same. And in theory, the regional system
could provide much better service to employees of small firms than
would a destination-based system, because the data base would be
larger and the chances of matching would be higher.
The operation of such a system involves several functional steps
(leaving aside for the moment such administrative aspects as promo-
tion and incentives):
1. Collection of data on the home address, work location,
and work schedule for each potential carpool candidate.
Data on personal preferences, such as smoking, may also
be collected.
27
-------
2. Translating the home addresses and work locations into
a consistent code with which the computer can operate
in order to match the candidates. This step is
referred to in some literature as geo-coding. Home
addresses are usually coded according to town, zip
code, or location on a grid numbering scheme on a map
of the area. Work locations are usually coded according
to either a grid numbering system or a list of specific
locations or landmarks.
3. Matching the carpool candidates. The computer sorts the
data to identify people who (1) work in the same coded
area, (2) live in the same coded area, and (3) work the
same schedule. Some systems provide for the computer to
also match people who live in adjacent coded areas, work
in adjacent areas, or work slightly different hours. In
addition, candidates may be further sorted according to
personal preferences (such as smoking). Systems must
deal not only with original matches but changes after the
system is used for a while, as a result of address and
employment changes.
4. Distribution of lists of the potential matches to each
of the carpool candidates.
In performing the above steps, systems can differ from each other in
many ways. The data collected varies; some systems ask many questions,
others ask only a few. The geo-coding can vary; many forms of map
grids and other coding systems can be used. The matching process can
vary; systems vary in the criteria and methods for matching. And
distribution varies; systems differ in the manner of distribution and
in the kind and amount of data distributed. Hence there1, is a great
variety of possible systems for performing regional matches.
Regional carpool matching systems carr»also involve employees directly.
All but the third functional step - the actual matching process -
can be done entirely within the employer organizations rather than
through a public medium. What distinguishes regional systems is that
the matching is done on a regional basis, Hence, employers could
collect data, publicize the program, and distribute the output,
while at the same time all the data from a region is actually pooled
together.
28
-------
In the face of all this potential variety in operation, however,
there is only a limited amount of experience. Regional carpool cam-
paigns are all less than one year old, and none has matched more
than 10,000 to 15,000 candidates. There has therefore been little
opportunity to thoroughly examine the results of the many regional
systems in operation. This report will review the experience with
some of the regional systems which have been in operation the longest.
Besides the pragmatic approach of examining experience in this section,
Section IX will examine some theoretical aspects of such systems.
Case Histories of Regional Systems
General - This report will review the current information on a few
regional carpool matching systems. Again, the emphasis will be on
highlights, recent developments, and contrasts with other kinds of
systems. Additional case history material is available in the litera-
ture published by the Federal Highway Administration and described
in the Bibliography of this report.
WBZ-ALA, Boston - Station WBZ is a major AM and FM radio and televi-
sion broadcasting outlet in Boston, operated by Westinghouse
Broadcasting Company. In the past several years WBZ has conducted
several major public service campaigns. For example, in 1970 the
station conducted an anti-smoking effort; in 1972, several special
programs and related public service functions were devoted to drug
abuse education and control; in the summer of 1972 a physical fitness
campaign was conducted. These campaigns have been characterized by
intensive publicity, special programming on the air, and community
involvement through such techniques as WBZ's provision of a physical
fitness course on Boston Common.
In September 1971, WBZ joined forces with a Boston-based automobile
club, the ALA Auto and Travel Club, to operate a free towing and
assistance service on the major commuter road into Boston from the
29
-------
south (the Southeast Expressway). This is coupled with traffic sur-
veillance by a WBZ helicopter, and both have continued through the
present time.
WBZ continued its transportation public affairs involvement in the
summer of 1973 by introducing the "Commuter Computer Clubcar" car-
pool matching service, again in conjunction with the ALA Auto and
Travel Club, which is the oldest of several broadcaster-operated re-
gional carpool systems.
History of the carpool campaign - The WBZ-ALA carpool campaign -was
announced in August 1973. The system details were described, and
audience response requested, in a special 90-minute television pro-
gram in prime time. The matching system was intensely advertised
on the air, in other media, and through industrial groups and the
Chamber of Commerce. By March 1973, approximately 15,000 applica-
tions for matching had been received. The rate at which applica-
tions have arrived has varied, with a higher rate of applications
during periods of heavy promotion and during the periods of long
*
queues for gasoline. Applications are continuing to arrive at the
rate of a few hundred per week.
Organization - WBZ carries most of the burden for publicity and co-
ordination. The ALA Auto and Travel Club provides the computer match-
ing service, some promotional expenses, mailing costs, and all
clerical effort. WBZ and AL£ have also tnllsried the assistance and
endorsement of the Associated Industries of Massachusetts, the
Chamber of Commerce, several major empio5^rs, and the government
transportation agencies.
Gasoline shortages in the Boston area resulted in 1- to 2-hour
queues for gasoline in much of February 1974. In addition, the
collapse and subsequent repair of a major toll bridge in September
1973 provided another incentive for carpools.
30
-------
Publicity and data collection - Promotion of the Commuter Computer
system began in August with public service announcements and
editorials on WBZ radio and television. The Massachusetts
Secretary of Transportation endorsed the system in a press conference
on 23 August, and major newspaper coverage resulted. On Friday,
7 September, WBZ staged a "Great Commuter Computer Clubcar Race," in
which WBZ personalities and government officials used distinctive
automobiles (such as antique cars) and "raced" into Boston to
dramatize the frustrations of traffic and the potential for carpools.
That evening WBZ carried a 90-minute special on television, with
government officials (including the Governor) discussing transporta-
tion issues and Westinghouse personalities operating a telephone
service for commuter computer applications, much in the style of a
charity telethon. WBZ continued to heavily promote the carpool
system, through radio and television announcements (at least twice
an hour), press conferences, and advertisements in local newspapers
and on billboards. Besides offering questionnaires to the broad-
casting audience, they were distributed in bulk to large employers,
to central distribution points such as supermarkets, and given out
by toll collectors at the Boston tunnels, turnpike, and toll bridge.
The system was publicized to employers by industry and commerce
organizations, and by employers in their own house organs.
Over 500,000 questionnaires have been distributed. In addition the
questionnaire was printed in all the major Boston and suburban
newspapers in September (but few responses used those forms).
Articles about the campaign have been carried by several local and
national newspapers and magazines, as well as in national broadcast
news. In the past month or two the amount of publicity has been
reduced to a few spot announcements per day on the air, but applica-
tions continue because of the gasoline shortages. The public is
asked to mail questionnaires to WBZ, along with ten cents for postage.
Incentives - Few explicit incentives have been provided. The thrust
of the WBZ campaign has been to overcome public antipathy to
31
-------
carpooling, make it socially acceptable, and to make it fun.
Originally it was planned that members of the "WBZ Commuter Computer
Clubcar" would obtain promotional benefits such as free coffee from
roadside restaurants, but to date these plans have not come to
fruition. One inducement has been a contest in which an automobile
was given away to a carpool applicant. This resulted in an increase
in applicants but apparently also resulted in many duplicate
applications. A coincidental incentive was the opening of an
express lane for carpools and buses on Interstate 93 leading into
Boston from the north.
Matching process - ALA processes the applications on its own computer.
Matching is based on (1) home zip code, (2) place of work, (3) hours
of work, and (4) other preferences. Applicants are matched only with
other people with the same home address zip code. Destinations are
coded according to a list of approximately 60 landmarks in the Boston
Metropolitan area. (Only 36 landmarks are listed on the questionnaire;
the others have evolved from processing the applications. A blank
category for additional locations is provided on the form.) Work
destinations include major employment centers ("Government Center/
City Hall"), specific sections of downtown Boston ("Park Square"),
suburban employment areas ("Dedham/128 Industrial Parks"), and park-
and-ride locations for mass transit service ("Riverside Station"),,
Work hours can be separately identified by starting time, quitting
time, and whether occasional rides are needed for departures one hour
later than usual. One can indicate preferences for drive, ride,
alternate driving; and either male or female companions. An additional
feature is an option to indicate a desire for rides to major sporting
events.
The matching program is run on an IBM 360/20 computer with an 8K
core and a card system for input. For approximately 10,000 names in
the data base and three or four weeks of applications (several hun-
dred), it requires about 9 hours to do an entire processing cycle,
32
-------
including preparation, sorting, elimination of duplicates, and pre-
paring letters. About two hours of actual computer time is involved
for a typical run. With a tape or disc system, ALA believes the
time could be reduced substantially.
The computer matching results in a letter print-out for each new
applicant, indicating his potential carpool matches or stating that
no match can be made. The computer printout itself is suitable for
mailing. When matches are made, the list sent out is accompanied by
a package with literature on carpools, special "Commuter Computer
Clubcar" membership cards and window decals, highway and transit
system maps, and an accident check list. The matching process is
re-run every three or four weeks. Only new applicants receive print-
outs, but all applicants are retained in the data bank.
Employer participation - Several major and many smaller employers,
including government agencies, distributed questionnaires for the
WBZ-ALA system. Associated Industries of Massachusetts offered a
service in which they provided WBZ-ALA questionnaires with a firm's
own message and company insignia printed on the form. Few employers
collected the forms themselves, so it is not known how many of the
completed forms resulted from this process.
Results to date - By mid-March 1974, the total number of applications
received was 15,000. After separating duplicates and applications
which cannot be processed, the data base in the computer was 11,141.
Of these applicants, 25 percent (approximately 2,800) were matched
to at least one other applicant. The rate at which matches are made
has been steadily rising as the number of applicants increases.
(Also see Section IX for further analysis of these results.) Both
WBZ and ALA are disappointed that more applications have not been
received and that the matching process does not produce more matches,
but WBZ and ALA will continue the'program at least through August
1974. (Matching rates should be interpreted cautiously; see discussion
on page 68.)
33
-------
As was pointed out in Section I , the WBZ-ALA carpool system has been
made an official part of the Transportation Control Plan for the
Boston Air Quality Control Region, and the promulgated regulations
specify that the Commonwealth of Massachusetts must operate a replace-
ment system if WBZ and ALA cease their operation. Plans have not
been completed, but the Massachusetts Executive Office of Transporta-
tion and Construction is working with FHWA, WBZ, ALA, and major
employers to devise an expanded and improved system to replace the
ALA-WBZ system. According to discussions with the officials involved,
the state may work with large employers to provide destination-based
matching for them, and would also provide a regional matching system.
WBZ and ALA may continue to be involved. No decision to proceed with
planning the details of such a replacement system has been announced,
nor has WBZ or ALA altered their commitment to complete their one-
13,14,15
year campaign.
WIND, Chicago - WIND is an A.M. radio station in Chicago, operated by
Westinghouse Broadcasting Company. Station WIND has been conducting
a regional carpool matching campaign similar in several respects to
the WBZ-ALA campaign described earlier. WIND (and the other Westinghouse
(Group w)stations) was assisted by WBZ in organizing their campaign.
There are, however, some important differences between the WBZ-ALA
campaign in Boston and the WIND campaign in Chicago. For one, WIND is
actually conducting both a region-wide campaign, in which the general
public is solicited to submit applications, and destination-based
matching for specific employers. Because of the small number of
employers involved so far, the WIND campaign will be treated as a
regional carpool system for the purposes of this discussion.
Another difference between the WBZ-ALA system and the WIND system is
that the WBZ-ALA campaign is the only broadcaster-operated campaign
(and the only regional campaign) in Boston, whereas a second Chicago
34
-------
radio station (WBBM) is operating a carpool matching system. The
WBBM system is described in another portion of this section (page 17)
A third difference is that the WIND computer matching is done by a
private firm, Automatic Data Processing, using a different program
from the WBZ-ALA program. The matching program will be further dis-
cussed below and in Section IV. *
History - The WIND system has been operated in a manner similar to
the WBZ-ALA system, but with a slightly later start. The station
carried editorials throughout 1973 dealing with several aspects of
transportation. In an editorial on September 6 and 7, 1973, WIND
discussed the need for a "commuter computer", not specifically in
connection with carpools but rather on a more general basis in con-
nection with analyzing Chicago's transportation needs. Then on
October 11 WIND announced the official beginning of the "Commuter
Computer" carpool matching service. Publicity of the system was
coupled with publicity for other WIND transportation-related programs.
On November 5, a "Great WIND Chicago Commuter Race" was held to pub-
licize the city's traffic problems and dramatize the potential for
carpools. (WBZ staged a similar race in Boston.) Publicity and data
collection has continued to the present time, and a one-year campaign
is planned.
Publicity and data collection - In addition to the "Great Race" and
the editorials previously mentioned, the carpool system has been
promoted over the air, through press releases to other media, and in
newspaper advertisements. Applications are available to people who
either write or call the station; people who write for an application
are asked to send a stamped, self-addressed envelope. Applications
have also been delivered in bulk (approximately 150,000 so far) to
major employers and other mass distribution points. Employers have
distributed a large portion of the applications so far, and have
either collected them or asked employees to send in their own. The
35
-------
application form is quite simple, with only four questions - name,
address, phone number, and location of work; arrival time; departure
time; and preference of drive or ride. Destinations are identified
by numbered grid squares on a map on the questionnaire.
Matching process - The matching program was prepared by the data
processor, Automatic Data Processing, Inc. (ADP), which is serving
other Westinghouse Broadcasting Company stations with the same pro-
gram. The program matches applicants on (1) home address zip code,
(2) destination grid number, and (3) work times and drive/ride preference.
The program does not search adjacent home zones for additional
matches but does search adjacent work zones. The program is written
in COBOL and versions have been written for both Honewell and IBM
machines. As with the WBZ-ALA program, the WIND-ADP program keeps all
applicants to date in the file, and sends letters with output only to
new applicants. (Also see Section IV.) Some firms are performing
their own keypunching; otherwise ADP is performing all the data pro-
cessing for both the regional and destination-based matching.
Employer participation - Employers are provided the option to either
pool their data with the region-wide applicants or to keep them
separate. Employers distribute, publicize, and collect questionnaires
within their firms.
Results to date - By mid-March 1974 approximately 3500 applications
had been received from the general public and new applications were
arriving at the rate of about 100 per week. Between 10 and 20 percent
of applications are being matched. Employer responses are just
beginning and consequently no data are available. '
Other Westinghouse Broadcasting Company Stations - Several other
Westinghouse Broadcasting Company (Group W) radio or television
stations are conducting, or participating in, carpool matching sys-
tems. These campaigns are in most cases based upon the WBZ-ALA
36
-------
campaign, and several use the ADP computer program. Several of these
will be summarized here.
KDKA, Pittsburgh - The KDKA "Commuter Computer" campaign began in
October, and was publicized in a manner similar to the WBZ-ALA and
WIND campaigns. Over 200,000 application forms were distributed,
about 5000 were received, and approximately 500 carpool matches
resulted. Computer processing was done by ADP. The campaign is now
going to be combined with that of the Southwestern Pennsylvania
Regional Planning Commission (SPRPC), which is initiating a regional
carpool program to be funded by FHWA. The 5000 applicants to KDKA's
campaign will be sent new applications for the new SPRPC system.
An entirely new computer program, reportedly similar to one developed
by the U.S. Bureau of the Census, is now being tested. KDKA will
18
continue to be involved in the promotional aspects of the new campaign.
KYW, Philadelphia - The KYW campaign was structured after the WBZ-ALA
campaign, but began only in February, 1974. Approximately 1000
applications have been received and processed, but no matches resulted.
The matching process uses zip codes for home address matching, with
large zip code areas further divided into townships or boroughs in
some cases. Destination matching is by landmark or general area, as
in the WBZ-ALA system, not by grid or zone. Data processing and
clerical aid is provided by the Keystone Auto Club. Processing is
done by a custom program, written in COBOL for an IBM 360/40. The
program requires 15 minutes for 1000 applicants and uses 75K bytes of
19
core.
KPIX-TV, San Francisco - Several business and several government
agencies in the San Francisco Bay area have joined together to form
a carpool organization which is conducting a region-wide matching
campaign, called "RIDES". Overall coordination is the responsibility
of a steering committee which represents the business sponsors and
the governments; an official from the FHWA regional office acts as
37
-------
chairman of the committee. The role of KPIX-TV has been promotion;
another station, KSAN, is also involved. Other promotional and
printing expenses are borne by the private sector. The Association
of Bay Area Governments (ABAC) is performing the geo-coding, using a
computerized technique based upon the U.S. Census Bureau DIME files
(discussed elsewhere in this report). Destinations are coded land-
marks or areas, as in Boston and Philadelphia. Actual matching is
being done by the California Department of Transportation on a modi-
fied version of the U.S. Census carpool matching program. This pro-
gram searches for additional matches in zones adjacent to the home
zone. So far, approximately 25,000 applications have been received.
(The large number of applications results from the large population
in the area being served, and a lengthy gasoline shortage.) Of these,
about 15,000 were processed in late March and an 84 percent matching
20
rate resulted. (See also Section IV.)
WJZ-TV, Baltimore - The Baltimore carpool system is being operated by
WJZ-TV, WFBR Radio, and the Automobile Club of Maryland. Matching
is performed with an Automobile Club computer program, based upon a
grid system for origins and destinations. The system has been pro-
moted on the air by both the broadcasting stations. Questionnaires
were also distributed through all McDonald's restaurants in the area
and through major employers. So far (after 4 months of operation),
approximately 2000 questionnaires have been returned. No data pro-
21
cessing for matches had been performed as of the end of March, 1974.
Summary of Experience with Regional Matching Systems
Several systems designed to provide carpool matching services through-
out a metropolitan region have bee'h placed in operation within the
past year. In all of these systems, tadio or television broadcasters
have been involved and in several cases have been the chief supporter
of the system. Most such systems are being operated by commercial
enterprises as a public service, but in at least one system
38
-------
(San Francisco), a government agency (the California Transportation
Department) is also supplying technical effort.
The system in operation the longest, the WBZ-ALA carpool system in
Boston, has attracted and processed only 15,000 applications from a
metropolitan area population of more than 2.5 million people. The
system with the most applications to date is the San Francisco sys-
tem, which has been in full operation less than two months but has
attracted approximately 25,000 to 30,000 applications from a metro-
politan area population of more than 3.0 million people. In Boston
the matching rate is 25 percent, in San Francisco it is 84 percent
among the first 15,000 applicants (but note the matching pro'grams are
quite different).
Regional systems in operation at the present time have been accompanied
by extensive promotion but few explicitly-provided incentives. In no
city has a follow-up study been done to determine how many carpools
are formed either directly from the matching lists or indirectly as a
result of the promotion of the carpool concept.
One outgrowth of regional systems for carpool matching has been the
development of systems which combine the features of regional and
destination-based carpool matching systems. Such combination systems
will be discussed next.
EXPERIENCE WITH COMBINATION SYSTEMS
General
Combinations of destination-based and regional carpool systems have
evolved from the recent efforts to provide matching services to an
entire metropolitan area. Although the history of these systems was
not exhaustively researched for this report, discussions with the
principals in several carpool systems indicate that the combination,v
39
-------
or dual, approach probably grew out of the twin desire to serve a
wide area and also to capitalize on the high potential for destina-
tion-based systems, and thus avoid the disappointing response to the
WBZ-ALA system and other early regional carpool systems.
The typical dbmbination system involves the solicitation of applica-
tions for carpool matching from both the general public and from
employees indirectly through their employers. Some systems actually
maintain a separate data base for the general public from the
employer data bases, so such systems are actually multiple carpool
matching systems being operated simultaneously. In other cases the
employer data are combined with the general public data. So combina-
tion systems can be based upon either destination-based matching
systems or regional matching systems.
It appears that most publicly-operated systems now being begun in the
U.S. are combination systems of some sort. Few regional systems are
being operated without overt attempts to work through employers as
well as to appeal to the general public.
Case Histories of Combination Systems
General - As with the other case histories for regional and destination-
based systems, the emphasis in this section will be on highlights,
recent developments, and comparisons with other systems. Most of the
systems to be discussed here have also been reviewed in the recent
reports published by FHWA and referred to in the Bibliography.
Denver, Colorado - The carpool matching system being operated in the
Denver, Colorado, metropolitan area appears to be unique in several
respects. The system has evolved from the work of a computer class
at George Washington High School, rather than from an overt attempt
by a government agency or broadcaster to create such a system.
Furthermore, the computer program itself was written largely by high
40
-------
school students and is somewhat different in its matching logic from
any other program reviewed for this report. At the present time the
system is evolving from a destination-based system into a combination
system, and responsibility for it is being transferred from the high
school to local civic organizations. Eventually it is planned that
a government agency will coordinate the system.
History - Unlike most carpool matching computer programs, which were
created to satisfy a demand, the Denver matching computer program
began as a class exercise in the computer math class at George
Washington High School. Some employers had been operating a carpool
9
system before this, but with small and declining interest. When
the computer program was operational it received publicity from an
article in a Denver newspaper and from news items on Denver area
radio and television stations. As a result of this publicity, several
local businesses and government agencies made inquiries about the
program. The students and their teacher, Dr. Irwin Hoffman, began
to make presentations to interested groups, and several employers
have adopted the program.
Interest in the computer program has grown to the point that the
school cannot handle all the requests for assistance nor can it do
as much data processing as would be required to satisfy the demand.
Consequently an organization consisting of local civic groups and
the regional council of governments has been established in order to
take over the operation and expand it further. At the present time
the organizations are trying to obtain Federal and State financial
support in order to carry out the regional aspect of the system.
Most of the work done to date and in process has involved matching
on an employer-by-employer basis.
Organization - Initially, George Washington High School staff and
students handled most of the effort. The organization at present is
being coordinated by Downtown Denver, Inc. (DDI), a membership group
41
-------
of Denver business. The Rocky Mountain AAA Auto Club (AAA), a local
affiliate of the American Automobile Association, is assisting with
publicity, employer contact, and clerical and keypunching work for
some employers. Generally, DDI is responsible for the contacts with
downtown firms and AAA is responsible for employers outside the down-
town area. A local computer service company, Min-Comp Computer Corp.,
has agreed to do the data processing for firms who do not wish to do
their own, and for the regional matching when it is begun, at a fixed
price per person. The entire software package - the computer pro-
grams, personal instructions in its use, and instructions on geo-coding
and other functions - is being offered to any organization at no
charge. The George Washington High School staff and students are
still involved in altering the basic programs to suit the particular
needs of participating organizations. Several government agencies have
endorsed the program and participate in it, and the Region VIII office
of EPA has presented an award to the high school for their efforts,
but otherwise no government agencies are yet strongly involved.
Attempts are now underway to obtain funds from the
FHWA for the Denver regional Council of Governments to carry out the
bulk of the work to widely promulgate the system and establish the
regional data base. State funds have been obtained for DDI to operate
the system from April until the FHWA funds are obtained.
Publicity and data collection - At the present time the system is
largely a family of destination-based systems and as a result most of
the promotion has been directed at employers. Within participating
employers, the companies themselves have handled publicity and data
gathering. Promotion to the employers has thus far been handled by
DDI and AAA, as noted earlier.
With the destination-based system, a questionnaire is filled out for
each employee, either by himself or by a staff member who has access
to the required information. Next, the employee's home address is
converted to a digital code (that is, it is geo-coded) by a clerk.
42
-------
The data are then keypunched onto computer cards, either by the
employer or by AAA. The collected cards for one company are then run
through the matching process, either on the employer's computer, at
Min-Comp, or at the high school. The output is then either returned
to the employee through company channels or mailed to his home; the
computer output can be folded and used directly as a mailer.
Matching process - The software for matching is a family of three
computer programs, plus a process of geo-coding. One program creates
initial carpool listings for a data base. A second program updates
the listings to provide new printouts for added people, people who
have moved or changed jobs, or corrections. The third program pro-
duces a density matrix to show the distribution of potential car-
poolers (or transit riders) by geographical location.
The matching process itself appears to be unique. Rather than operat-
ing with grid cells or origin-destination designations, the program
uses actual x/y coordinates of homes and destinations. This enables
the use of a unique vectoring method for locating carpool partners
convenient to the candidate's route from home to work. This matching
program will be described in greater detail in Section IV.
Results to date - Several major employers have reported substantial
changes in driving habits as a result of the carpool system. Great
Western Sugar Company reported that 25 to 35 percent of their 250
employees are now carpooling, whereas few did previously. The Air
Force Accounting and Finance Center surveyed their 3,400 employees in
December and found about 25 percent were already in carpools. Of the
86.9 percent of the employees who returned the survey forms, 1,424
expressed an interest in carpool matching and were processed. By
February 1974, the number of people using carpools had risen to 1,547
(over 45 percent) (and 288 people were using public transit).
Altogether approximately 10,000 employees have been processed by one
43
-------
of the firms with the program. Typically, 99 percent of employees
22 23
have received at least one potential match. '
Dallas, Texas - The cities of Fort Worth and Dallas have coordinated
in a single carpool matching system in which employers and the general
public are being requested to participate. Again, however, the
emphasis at first is on obtaining cooperation from major employers.
History - In October 1973 a pilot project was undertaken to perform
computer matching for 650 Dallas city employees. The next step was
to extend the data collection to other city employees. In December
1973 and January 1974 a series of meetings was held in which employers
(private and government) were briefed on the carpool system and invited
to participate. In selecting employers to attend these meetings the
city used a Chamber of Commerce tabulation of the number of employees
in each firm; the city attempted to first reach major employers. So
far representatives of employers with a total of more than 100,000
employees have been given a presentation. By early March, 1974,
approximately 35 percent of Dallas employers had completed the data
collection phase, and data collection was continuing. Dallas has
submitted an application to FHWA for a grant to conduct this carpool
program over the next two years. If the grant is approved it is
intended to attempt to process all the employees in Dallas County.
Organization - Overall coordination has been the responsibility of the
Traffic Control Departments of the two cities. The Chamber of Commerce
assisted in promotion and contacting employers. Two broadcasting
stations are conducting their own compaigns but efforts are being made
to coordinate them with the city campaign. Employers asked to par-
ticipate are requested to appoint a coordinator to be responsible for
the effort within the firm. Employers are asked to obtain maps and
survey forms, distribute these to their employees, and then collect
them when completed. Employers were given the option of either pro-
cessing their own data or having the city perform this. Keypunching
44
-------
can be done either by the city or the employer, as can data processing.
The intention is to eventually process the entire region, so employers
are encouraged to pool their data together.
Matching process - The FHWA computer matching program is being used in
the Dallas-Fort Worth program. With that program, the matching is
based upon a grid system for both origins and destinations (i.e.,
home and work addresses). The grid system being used is a single
system of rectangular grids for the two-city metropolitan area.
Results to date - The City of Dallas reported in early March that not
enough data had been processed to permit evaluation of the matching
process. As noted above, approximately 35 percent of the employers
24 25
have submitted their questionnaires. '
Connecticut Cities - The Connecticut Department of Transportation has
been involved with programs to encourage or support carpool use for
several years. Recently these efforts have included provision of
carpool matching services for private employers and for regional sys-
tems.
History - In 1969 the Connecticut Department of Transportation
(ConnDOT) determined from a survey that a substantial number of
people were parking their cars at highway interchanges - often in an
7 f\
illegal or dangerous location - in order to join carpools. This
led ConnDOT to begin a program of constructing commuter carpool park-
ing lots at highway interchanges throughout the state. By late 1973
the state had built 11 such lots and 13 are scheduled for completion
in 1974. In addition, 79 emergency (non-permanent) lots are planned
for interim use during the present fuel shortage and until permanent
lots can be designed and constructed.
In parallel with the development of the carpool parking lots, since
June 1973 ConnDOT has provided computer carpool matching services to
45
-------
several parts of the state. The matching service has been used in
both destination-based, employer-operated systems and regional systems.
Organization - Several organizational structures are in use. Govern-
ment agencies, including ConnDOT, and other employers have been pro-
vided the carpool matching data processing service, with actual
administration performed by the employer involved. These carpool
operations have been destination-based matching. In addition, regional
carpool matching services have been developed in several cities. In
Hartford, the sponsors are ALA Auto and Travel Club, a major bank, the
Insurance Association of Connecticut, and an AM-FM radio station. In
Newington, the sponsors are an AM radio station, an insurance firm,
and a bank. The southeastern Connecticut (Norwich area) system is
sponsored by a consortium of local industries and government agencies.
The New Haven system is sponsored by AAA and a local radio station.
In each of these cities, the data are collected by the local sponsors,
on standardized application forms with local sponsors identified, and
processed by ConnDOT on a regional basis.
Matching process - ConnDOT has written its own matching program.
Geographic matching is based on Connecticut's 1,725 traffic zones,
which vary in size according to population density and land use.
Consequently, each application must be hand-coded to identify the
traffic zone corresponding to the applicant's address. Besides
address and work schedule, the program matches applicants on their
preference to drive or ride. The program is written in FORTRAN and
run on a UNIVAC 1106 computer, but is-not documented.
Results to date - By January 1974, matching services had been provided
to 50,000 employees of 20 firms. In Hartford, the program provided
matches for two-thirds of the first 2,500 applicants. Data collection
and promotion are continuing, with area-wide systems operating in*all
27
the major urban centers in Connecticut.
46
-------
Washington, D.C. - The Metropolitan Washington Council of Governments
(COG) has coordinated the efforts of several government agencies to do
destination-based matching throughout the metropolitan region.
History - The COG program was announced in September 1973, after plan-
ning for several months. The Northern Virginia Transportation Commission
(NVTC), one of the cooperating agencies, started a pilot matching cam-
paign at an industrial park in Virginia in September. The COG computer
program was actually developed for the NVTC operation. Promotion and
use of the system is still being expanded.
Organization - The carpool system is at present being operated through
employers. COG has worked with the Metropolitan Washington Board of
Trade, the Federal City Council, and other business organizations to
promote and encourage use of carpool matching and carpool incentives.
A special transportation committee, representing business and government
sectors for several years, has endorsed and is working to promote the
COG system. WTOP, a Washington radio and television station, also
participates in the promotion of the campaign, and supplies question-
naires .
Matching process - Actual computer matching is done by COG using a pro-
gram they prepared. Home address matching uses the Census Bureau DIME
files for geo-coding. Destinations are treated one at a time but the
program is reportedly capable of use in multiple destination systems.
(It is planned to solicit responses from the general public at a later
date.) The program was written in COBOL for an IBM 370/158 and requires
a large amount of core (100K). Documentation was not complete by mid-
March, 1974.
Results to date - By mid-March, approximately 35,000 questionnaires had
been completed, for several destinations. Matches are provided for 80
28 29
to 90 percent of these applicants. '
47
-------
Knoxville, Tennessee - The carpool matching system in Knoxville, Tennessee,
is another example (like Denver) of a system in which destination-based
matching forms the nucleus for the system, with a planned expansion to
regional matching in the future. In the Knoxville system the city
government has been heavily involved, which is unusual, and there is an
emphasis ort using the survey data for transit planning, which is also
unusual in practice if not in concept.
History - Planning for the "Car-Bus Pool" began in early 1973, largely
as a technique to improve transit planning and operation. Publicity and
detailed planning continued through the winter of 1973-1974, and survey
forms were distributed in February 1974. Survey forms are currently
being processed. Some changes to the transit routes have already been
made, and park-and-ride facilities are being initiated. A Knoxville
radio station which originally attempted to organize a separate, WBZ-type
system has now turned over their data to the city and combined efforts
with the city.
Organization - The organization of the Knoxville system is a unique
combination of public and private forces. This organization is dis-
9
cussed in detail in the literature, so it will only be summarized here.
The chairman of the carpool/buspool program is a. member of the Knoxville
Transit Authority, in keeping with the philosophy that carpools are a
portion of the mass transit system. A planning committee is composed of
representatives from both private and public sectors. An appointed
advisory committee represents the community, and technical staff is
provided by the University of Tennessee and the regional planning agency.
In addition, the mayor assigned a special assistant as a transportation
coordinator.
Major employers have been heavily involved. Knoxville has several large,
relatively separated employers who account for much of the workforce, so
destination-based matching was deemed to be appropriate for most employees.
Bus routes have been tailored for specific major employers, and emplojrers
48
-------
have distributed and collected the carpool/buspool survey forms. All the
media - newspaper, radio, television, and billboards - have been used
for promotion. The Mayor has personally given public support and the
survey forms begin with a letter from the major. The Chamber of Com-
merce is assisting by distributing survey forms to employers.
Matching process - The University of Tennessee has extensively revised
the FHWA matching program for use in the Knoxville system. Matching
uses the origin grid cell, destination, and work schedules, and in
addition uses an array of personal preference questions designed to pro-
duce matches which are personally compatible as well as convenient.
The program is being run on an IBM 360/65 computer. Program modifica-
tions are continuing to be developed, and it is planned to incorporate
a matching search on the destination end.
Results to ttate - Approximately 65,000 questionnaires have been dis-
tributed to employees of 360 firms in February and March 1974, and
approximately 20,000 have been completed and returned. Of those re-
turned, approximately 65 percent have expressed an interest in carpool/
buspool matching services. '
California Cities - The California Department of Transportation (Gal-
Trans) is organizing a carpool system for use in cities throughout the
states (except in San Francisco, where the system described elsewhere
had already been organized). The CalTrans participation includes the
matching service, promotion, research, planning incentives, and advice
on non-technical aspects such as law.
History - CalTrans began in early 1973 to prepare a carpool matching
program. In October 1973 the Governor asked for the establishment of
voluntary carpooling as an energy-saving technique. In December 1973
the Governor assigned CalTrans the responsibility of assisting in the
development of carpool systems for all state agencies. Also, the Dis-
trict Directors of Transportation were assigned to coordinate and assist
49
-------
local carpool efforts. Formal organization plans for the state head-
quarters effort were complete by February 1974, and detailed planning
and implementation of systems are now underway.
Organization - Within the state government, a Supervising Highway Engineer
has been assigned as Car Pool Coordinator by CalTrans. Assisting him
is a team from the CalTrans staff, representing services including
public information, computer services, finance, legal, supply, and en-
gineering. Other headquarters staff operate the computer matching
systems, carry out research, prepare publications, and provide coordina-
tion with the regional offices of CalTrans. Regional Coordinators have
been assigned, with responsibilities for promotion, liaison, consulting,
and coordination with employers. CalTrans is making available their
matching service and consulting services to other State agencies, arid
plans to offer these services to the private sector as well. State
involvement will include publicity and providing incentives, also.
Matching process - The FHWA matching program is being used (except in
San Francisco, where the modified Census Bureau program is being used).
Eventually CalTrans hopes to operate two parallel matching services,
one for destination-based matching and one for regional matching. In
the beginning, however, the emphasis is on a campaign to perform des-
tination-based matching for one organization at a time. This stems from
CalTrans1 conviction that people strongly prefer to carpool with others
from the same employer. The regional system would involve a telephone
advisory service which could assist individuals not matched through em-
ployer groups. The staff for this service would have the master output
from the destination-based system and would provide a broker-like
service to match the isolated individuals with other carpools. In addi-
tion a computer program is planned to provide custom matching service if
the organization-based output is not sufficient.
Results to date - The statewide campaign is fairly new so few results
have been reported. In an initial campaign at CalTrans headquarters in
50
-------
Sacramento, 1700 out of 2000 employees returned completed survey forms.
Preferential parking for carpools was instituted and the number of
carpools increased from 90 to 150.
Incentives - CalTrans emphasizes the importance of incentives. Within
CalTrans, incentives for carpooling include preferential parking and
flexible work hours in order to facilitate use of carpools (or of transit),
31 32 33
Other incentives are being studied. ' '
Summary of Experience with Combination Systems
Combination carpool matching systems incorporate some of the features of
destination-based matching systems and some of the features of regional
systems. Combination systems attempt to serve a metropolitan area but
emphasize strong employer participation in order to achieve a large
data base. Some combination systems are operated by the private sector
but the trend appears to be for government to organize such systems.
Combination systems are relatively new, so it is difficult to conclu-
sively show their effectiveness. Nonetheless, the systems reviewed for
this report show great promise as mechanisms for both encouraging car-
pooling and providing the matching service needed to facilitate the form-
ation of carpools.
The most successful combination systems (Denver and Knoxville, e.g.) are
characterized by enthusiastic support from a broad-based coalition of
government officials, business leaders, and major employers. Such sys-
tems also tend to have the most highly developed computer matching pro-
grams, but this may be a result of their recent beginnings rather than
a key to the success of the system. Operators of the combination sys-
tems believe that the key to their service is the dual approach of (1)
providing destination-based matching to large concentrations of employ-
ment and (2) later expanding to a regional data base to assist the minor-
ity of employees not reached through major employers.
51
-------
Of the systems in use and reviewed for this report, the Connecticut and
Denver systems have the greatest amount of operational experience. All
the systems are rapidly expanding their data bases, however, and in a few
months many should be fully operational and results of their carpool
campaigns should be available.
SPECIAL PURPOSE CARPOOL SYSTEMS
Throughout this report most of the discussion is of systems intended to
reduce the amount of vehicle travel associated with the journey to work.
Most carpool systems are designed for work-related travel rather than
other travel purposes because more travel is associated with employment
than with any other single purpose, and because travel to and from work
is regular and concentrated in time and place and is thus more readily
subject to carpooling.
There is some potential, however, for carpooling for non-work travel, so
this section will briefly review special purpose carpool systems. Car-
pool systems can be conceived for purposes such as recreation, shopping,
or school. A few such systems are operating at the present time.
One type of system is the supermarket shopping carpool matching service.
At least one supermarket chain in the Boston area has advertised such a
matching service and has made its computer program available to other
supermarkets through the National Association of Food Chains. The pro-
gram matches on home address zip code, store location, and shopping day
and time. From the brief review made for this study it does not appear
that there has been very much response to this service (fewer than 1000
applications from the customers of 60 stores) but. the service has been in
34 35
operation for only a few weeks. '
Another matching service deals with recreational trips. WBZ and ALA. in
Boston included on their questionnaire a question relating to carpocl
travel to major sports events. Few applicants have requested this
matching service, and matches are no longer processed on this question.
52
-------
From this review it appears there is little experience with special
purpose carpool matching systems. On the other hand, informal carpools
for non-work trips are certainly in wide use; it is not uncommon for
several people to travel together to shop, attend a football game, or
go on a weekend ski trip. In fact, vehicle occupancy is usually much
higher for non-work travel than for travel to work; auto occupancy ox-
rates of 1.6 to 2.4 persons per vehicle are typical for non-work trips.
The difficulty with matching services for non-work trips is that such
trips are irregular in time and place, often do not involve movements
of statistically large groups of people at one time, and are usually
undertaken at times when roadway congestion is slight and consequently
the incentives for carpooling are personal rather than societal. Mast
such trips - particularly the growing amount of social and recreational
travel - are undertaken outside of any institutional framework which
would readily provide the opportunity or incentives for formal matching
services. Certainly it would be undesirable and impractical to rigidly
schedule shopping and recreation. A possible exception to this would
be trips to and from school, which are fairly regular and may overlap
o o
peak commuting time as well. Some experience has been reported , but
not enough to be conclusive.
Social and economic forces may well encourage substantial increases -in
special purpose, carpooling, however. Whatever trends tend to provide
added inducements for work-related travel, such as frequent fuel short-
ages or preferential treatment of carpools on highway facilities, will
also tend to promote increased, but informal and temporary, carpools.
SUMMARY AND OBSERVATIONS ON MATCHING METHOD AND DATA BASE
Summary of Experience Reviewed for This Report
Carpool matching systems reviewed in this study can be categorized as
destination-based, regional, or combination systems. Experience with
each type was summarized in this section.
53
-------
Destination-based systems have been in use for the longest time and have
served the largest total number of people, but as isolated systems have
generally not affected the travel patterns of large numbers of commuters
in any one metropolitan area. Special investigation of several large
employers in the Boston area showed a lack of enthusiasm among these em-
ployers for the WBZ-ALA region-wide matching system. Employers cited
several reasons for preferring their own destination-based systems, in-
cluding the need for privacy of data and the low matching rate achieved
so far by the WBZ-ALA system. This study showed that manual or semi-
automated matching systems appear to be adequate for up to several thou-
sand employees at one site.
Regional matching systems have been operated in recent months in several
cities, chiefly by radio and television broadcasting organizations in
cooperation with other elements of the private sector. These systems
are characterized by intensive publicity and promotion and relatively
simple matching processes. Of these systems, the WBZ-ALA system in
Boston is the oldest and has attracted 15,000 applicants. The matching
rate is now 25 percent. A system in San Francisco, in which a broad-
caster participates but which is being operated by a coalition of orga-
nizations including both state and Federal government agencies, has
attracted approximately 25,000 applicants although it is younger than
the Boston system. Other systems have attracted even fewer applicants
for matching service. While the data bases in these systems are con-
tinuing to grow, it is generally conceded that these systems have not
attracted as many applications as had been anticipated.
Combination systems are the newest type of matching systems and appear
to be the type most often organized in the recent past. Systems reviewed
for this report are characterized by broad-based enthusiastic support
from the community. Generally such systems use a dual approach of first
attempting to encourage carpooling among employees of large employers
and then combining data bases to provide a matching service for the gen-
eral public. Of the systems reviewed, Connecticut's statewide system and
54
-------
Denver's system have provided matching services to the most people.
Several such systems can be expected to be in full operation and have
substantial data bases within a few months.
Observations
Matching Rates - Matching rates (i.e., the percentage of applicants for
matching who receive lists of potential carpool partners) vary widely
among the system reviewed. The matching rates depend to a large extent
upon the type of computer matching program being used. (These programs
will be discussed in more detail in the next section.) In general,
however, regional systems have matched 25 percent or less of their
applicants whereas destination-based systems and combination systems
based upon destination-based matching have exhibited matching rates of
50 percent or more. Regional systems appear not to have attracted
enough applicants to provide matches for the tremendous number of pos-
sible combinations of origins (home addresses), destinations (work
locations), work schedules, and personal preferences used for matching.
(The higher matching rates for destination-based and combination
systems result in part, however, because these systems tend to use
more sophisticated computer programs than do regional systems.)
Potential - Destination-based matching may be advantageous to an indi-
vidual employer and his employees, but unless such practices are wide-
spread in an urban area there would be little impact on travel or air
quality. Regional systems have the potential to provide matching ser-
vice throughout a metropolitan area and in theory could reduce travel
and air pollution. But such systems have not yet demonstrated their
efficiency; the Boston system, the oldest one, has produced matching
lists for only 2800 people. Regional systems have potential for air
quality improvements only if much greater segments of the population
apply for matching than have so far, or if such systems encourage
people to form carpools on their own. Combination systems are also
not yet proven, but because of their employer focus have a great poten-
tial for encouraging carpooling among large blocks of commuters.
55
-------
In the systems studied, there is some indication of the impact of
incentives. Generally, incentives have resulted in measurable increases
in the number of actual carpools, not just in applications for matching
service. This result occurred in cases with non-automated matching
services as well as in cases with sophisticated computer matching
services.
Advantages and Disadvantages - Destination-based matching systems have
the important advantage of a strong identification between the system
and the employer operating it. This can lead to a high degree of coop-
eration and participation both within the employer management and among
the employees. From a public standpoint, such systems have the major
disadvantage of being uncoordinated between employers and probably
being used only in isolated cases (i.e., only a few employers will use
them without the pressure of.outside forces) and consequently cannot be
relied upon for reducing vehicle travel and air pollution. (This dis-
advantage could be overcome by regulation or by other means to restrain
travel.)
Regional systems have the potential advantage of providing matching
service throughout a metropolitan area for all commuters regardless of
whether they work for an employer who participates in a matching sys-
tem. Such region-wide service can conceivably provide a means for car-
pool service for people who have unusual travel habits and need a large
data base in order to locate carpool partners. Unfortunately such sys-
tems have not yet demonstrated theiy ability to attract enough applicants
for large numbers of people to be matched. Most such systems lack the
personal identity, privacy of data, and personal service destination-
based systems have provided. On the other hand the promotion of
regional systems may tend to improve the public's concept ("image") of
carpooling and thus encourage greater use of carpooling outside of the
formal system.
56
-------
Combination systems have the advantage of destination-based systems
in being identified with the employer and having a demonstrably high
matching rate which can inspire confidence and wide use of the system.
For region-wide use such systems can potentially provide a very large
data base for general population matching, provided privacy issues can
be resolved. Such systems have the greatest potential for reductions
in travel and air pollution. One disadvantage of such systems may be
the amount of effort, time, and cost involved with preparation and
operation of a system satisfactory to large numbers of employer organ-
izations .
Best Approach - First, it must be said that it is probably premature to
conclusively decide what is a best approach; there simply is too little
experience. Second, local conditions vary so widely that there pro-
bably is no "best" approach. Within these limits, however, the experi-
ence reviewed for this report implies that a dual approach should be
used, similar to the Denver or Knoxville combination systems. A coor-
dinated effort should be made to encourage carpooling among as many
commuters as possible. To this end, the effort should be directed
first at the employers who employ the majority of the metropolitan
area; wherever possible, employer data should be pooled together to
form a substantial data base for matching people who do not work for
major employers or who have unusual travel requirements. Overall
system operation should be managed by a coalition of government, busi-
ness, and the public, with strong support elicited from major employers.
57
-------
SECTION IV
MATCHING SOFTWARE
GENERAL
This section of the report will discuss the computer programs which
perform the actual matching process to identify potential carpool
partners. Experience with several programs now in use will be sum-
marized, and programs being developed or being revised will be
reviewed.
Some destination-based carpool systems do not use computers for
matching) or use computers only for generating lists of potential
riders by home address rather than for actual matching. Some of
these non-automated matching systems were discussed in Section III
and consequently will not be further reviewed in this Section. For
additional information on manual matching techniques, the reader may
wish to refer to the FHWA or KUFSAM reports listed in the Bibliography.
REQUIREMENTS AND CONCEPTS
Any computer program, or family of programs, used for carpool matching
must perform certain functions. Of foremost importance is matching on
the basis of home address and work schedule. Matching on the basis of
work location may or may not be performed, depending upon whether the
system is used for regional matching or for destination-based matching.
Some computer programs match people on other criteria, including one
58
-------
or more of drive/ride preference, frequency of carpooling, smoking
preference, male/female companion preference, and carpooling for non-
work purposes. The requirements for matching of these various char-
acteristics will be discussed in the following sub-sections.
Origin Coding And Matching
There are two basic steps associated with matching origin (home address)
information for a particular carpool candidate:
1. Defining and identifying the area in which
the home address is, and
2. Determining other potential carpoolers with
home addresses near the candidate's.
Designing the matching system thus requires designing the method for
identifying home addresses in a way the computer can work with and
designing the method for searching among the data for addresses of
carpool partners near the candidate's home.
Origin Coding - Home addresses can be handled in the computer in
several ways. Theoretically, one way would be to work directly with
the street addresses and retain in the computer's memory some sort
of "map" which mathematically related the addresses to each other.
This map would enable the computer to calculate the distance between
one home address and another in order to determine the best potential
carpool partners.
Such a process could be done in two steps. First, the home address
would be converted to a numerical code representing the spatial loca-
tion of the home. For example, the address could be coded in rectan-
gular (x and y) coordinates from a fixed reference point on the map.
Next, the computer would work with the coordinates of all the candi-
dates to algebraically calculate home-to-home (or home-to-work)
distances.
59
-------
Another way to work with addresses is to divide the metropolitan area
into zones and identify each zone and all addresses within it by a
single number. Such zones can be either administrative designations
or arbitrary. Administrative zones could include the postal zip codes,
municipalities (assigning a unique number to each), census tracts, or
state highway department traffic zones. Arbitrary zones could include
such techniques as the use of a rectangular grid pattern which would
be superimposed on a map of the area. Irregular patterns or irregu-
larly numbered patterns, such as zip codes, would probably prevent the
use of some types of matching processes because it is difficult to
describe in mathematical terms how zones in these patterns are arranged.
Storing a "map" of the system would add to the length and cost of a
program. In contrast, regular systems such as rectangular grids are
more readily used for computations because there is a one-to-one rela-
tionship between the grid designation and its physical location.
These concepts are illustrated in Figure 1. On the other hand, ad-
ministrative zone systems may require less clerical work outside the
computer. The use of zip codes or municipalities requires little or
no extra work because they are directly identified by the address
supplied by the applicant, whereas the use of x and y coordinates re-
quires the determination of the coordinates for each home address,
which might be a time-consuming clerical process. With some arbitrary
systems such as grids it may be possible for applicants to code their
own addresses but such systems may lead to confusion and errors on
the part of many applicants.
Another choice involves the size of zones. Small zones may contain
too few candidates for good matching opportunities; large zones may
include more candidates but homes in one zone may actually be too far
separated for convenient matching. One way to overcome the difficulty
of zone size selection is to use a matching process which involves
matches among adjoining zones; with such a process the actual zone
size is less critical.
60
-------
4-1
to
4J
a,
0)
o
£
o
o
bO
a
H
73
o
o
R
CO
t-l
60
O
<0
O
00
H
61
-------
Origin Matching - Several types of matching logic can be used, depen-
ding upon the type of origin coding and the type of computer program.
Single-step matching selects all the home addresses in the same
designated area. All addresses in one zip code, one town, or one
grid square are identified as potential partners (subject to the
constraints of destination, work schedule, etc.).
Multiple-step matching selects all the home addresses in the same
designated area and then if necessary selects additional addresses
in adjacent areas. Typically, a minimum number of matches is estab-
lished, such as eight people with the same destination and work
schedule. If this minimum number of potential partners is not found
in the first zone searched, additional zones are searched. In a
rectangular grid, the first zone (or cell) would be the zone in which
the candidate lives. Next, all the eight adjacent zones in the grid
would be examined for potential partners, and so on. With a system
based upon x and y coordinates rather than zones, searching can proceed
inside circles of increasingly-wider radius until the desired quantity
of matches is determined.
Another type of searching involves searching along the route of travel
rather than searching in the origin's vicinity. Using some criterion
for nearness of the potential partners to the candidate's route, the
computer program would locate potential partners along the route from
the candidate's home to his job. The route could either be the actual
route along the street network, which may involve extensive mathematical
modelling of the highways, or the straight-line route between origin
and destination.
With any type of search mode, a related issue is that of the quantity
of matches. Two approaches can be used: exhaustive and minimum
number. In the exhaustive method, all candidates meeting certain
62
-------
criteria are listed together. Typically, a program with zip code
origin coding uses an exhaustive search. All persons living in the
same zip code are listed (provided they meet the destination and
schedule criteria), regardless of whether the quantity is one or one
hundred. With the minimum number approach, searching continues until
a specified minimum number of matches is obtained. With x-y coding,
for example, the program may search circular areas with increasing
radii until at least eight matches are obtained.
In summation, selection of the origin coding and matching process in-
volves some compromises among speed, complexity, flexibility, clerical
effort, and matching success.
Destination Coding And Matching
Destination coding and matching involves similar concepts to origin
coding and matching. Destinations are apt to be more concentrated
than origins, however, which alters the techniques somewhat. In addi-
tion, some computer programs are designed for destination-based systems
and destinations are thus not searched.
Destination (work location) matching involves the same two basic steps
as for origin matching, namely:
1. Defining and identifying the area in which
the destination is, and
2. Determining other potential carpoolers with
destinations near the candidate's.
Destination Coding - Destinations can be coded in the same ways as
origins (home addresses), including grid systems, direct x-y coding
from addresses, and use of zone designations. In practice, larger
administrative zones such as zip codes or municipalities would be
difficult to use because an excessive number of destinations or an
excessively large area could result. With too large an area,
63
-------
potentially matched locations might be widely separated in travel time
because of traffic congestion or might include awkwardly large numbers
of potential partners.
An additional type of coding useful for destinations is landmark or
area coding. General vicinities can be described verbally and assigned
an identifying number. Verbal descriptions would include terms famil-
iar to local people, such as "Government Center," "Post Office Square,"
or "Jonesville Industrial Park." Size and distribution of such areas
can be varied to suit local conditions, taking into account such con-
siderations as employment density, relative locations of designated
areas, highway network, and transit connections. Park-and-ride facil-
ities for railroad or mass transit service can be included in this type
of destination designation, thereby facilitating the coordination of
travel modes.
Destination Matching - Carpooling candidates can be expected to prefer
that potential carpool partners work relatively close to the candidate's
workplace. Traffic congestion is apt to be greater in employment cen-
ters than in residential areas, so it would involve more time and stress
to travel a certain distance among work locations to pick up passengers
than to travel the same distance among homes to pick up passengers.
People are more likely to know something about firms physically near
their own and thus may be more willing to carpool with employees of
nearby firms. In the matching process itself, the maximum distance
between the workplace of the applicant being matched and other work-
places should be chosen so that the matching process does not identify
either too many potential partners, making selection a cumbersome pro-
cess, or partners so distant they will be rejected by the applicant.
As with origin matching, either single-step or multiple-step matching
can be used for destinations. With single-step matching, all work-
places in one grid square or zone number would be searched for matches.
64
-------
With multiple-step matching, additional grid squares or destination
zones are searched if insufficient matching results from the initial
zone. Also, as discussed in connection with origin matching, destina-
tion matching can be based on an exhaustive listing of all matches
meeting certain criteria, or based on a criterion of a minimum number
of matches regardless of distance.
There is no need to use exactly the same criteria for destination
matching as for origin matching. In matching people who work in the
central business district (CBD) of a large city, for example, exhaus-
tive one-step matching may be suitable for the GBD destinations whereas
minimum-number multiple-step matching may be more suitable for origin
matching in the same program. Of course, use of two different match-
ing techniques may cause the computer program to be more complex, but
at the same time such techniques can increase the overall speed and
utility of the matching system.
Coding And Matching Of Work Schedules
The problem of matching work schedules is more difficult with regional
or combination systems than with destination-based systems because of
the greater variety of possible schedules. Yet this aspect of matching
is also very important, because carpooling will not succeed if the
carpool either jeopardizes the punctuality of a member or requires
significantly longer work days for him in order to accommodate the
other members of the pool. On the other hand, lack of flexibility
could result in fewer potential matches.
Work Schedule Coding - Coding of work schedules can either be continu-
ous or discrete. That is, applicants for matching may be asked to
either identify their own exact working hours (e.g. 8:10 a.m. to
4:45 p.m.) or to round off their work schedules to fit a specified
list (e.g. start at either 8:00 a.m. or 8:30 a.m. and finish at either
65
-------
4:30 p.m. or 5:00 p.m.)* If the applicant specifies his work hours
exactly, rounding may be done by the matching system operators or may
not be done at all. If the applicant does his own rounding off, he
may be able to more accurately Judge his flexibility than could the
operator of the matching system. For example, a professional person
whose office nominally works from 8:30 to 5:00 may be able to easily
accept arrival at 8:45 and departure at 5:15. So if rounding is to
be done, the applicant may be better suited to do it.
If rounding is done by the system operators, with the applicant's data,
some rules must be established. Without foreknowledge of the appli-
cant's flexibility, the system operator should presume that the appli-
cant specifies the latest time he may arrive; thus, if the applicant
specifies 8:10 a.m. as a start time and times are to be rounded to the
quarter-hour for processing, the next earlier quarter-hour, 8:00,
should be specified to the computer. (The questionnaire should ex-
plicitly identify to the applicant that he identify the latest time
he can arrive, earliest he can leave, etc.) Departure times would be
rounded to the next later time period, as from 5:10 to 5:15.
Work Schedule Matching - Computer programs can either match people only
with others whose schedules are identical, or match those whose work
schedules fall within a range. Matching to exact specified schedules
would only be done with previous rounding of work schedules so that
there would be a reasonable chance of matching. Matching within a
range would involve either pre-rounded or un-rounded times.
Matching within a range of schedules would be more appropriate and
more essential in cases where either few applicants are involved or
many work schedules are involved. The same sort of trade-off prevails
with schedule matching as with location matching, namely a compromise
between narrow criteria with few matches and broad criteria and many
matches.
66
-------
Coding And Matching Of Other Criteria
Besides location and schedule criteria, potential carpool partners can
be matched on numerous additional criteria. The more criteria there
are, however, the fewer the chances of matching each individual. With
large data bases and small variations in schedules, such as with des-
tination-based matching systems, matching on additional criteria may
be more feasible than with regional systems where good basic matching
service for an entire metropolitan area is desired.
Matching criteria in the optional category include the following:
1. Whether applicant wants to drive, ride,
or both.
2. Whether he has a preference regarding
smoking or non-smoking.
3. Whether he prefers only male or only
female carpool partners, or only married
or unmarried partners.
4. Whether he wishes to use a carpool
occasionally or regularly.
5. Whether he sometimes works overtime and
wants to be matched for additional work
schedules.
6. Whether the applicant wishes to be
matched on other preferences, such as
with only people from same employer,
only people of same age group, only
people of similar education, etc.
7. Whether the person wishes to obtain
matches for non-work purposes such as
sporting events.
These possible matching criteria all can be considered personal prefer-
ences. The use of such criteria for matching would be based upon the
concept that carpools are more apt to succeed with compatible people.
The validity of this concept is yet unclear but may be shown shortly
by experience with the Knoxville system, which uses an elaborate
questionnaire based upon this concept. The difficulty is that speci-
fying too many such preferences could lead to low matching rates. An
67
-------
alternative to actually matching on these preferences is to merely
identify the applicant's preferences in the data sent to each applicant;
for example, the list of potential partners could state the smoking/no
smoking preference for each name and allow the person using the list to
make his own decision of whether to contact: another person whose pref-
erences differed from his own.
Overall Matching Rates
Given the variety of matching criteria, it is apparent that the match-
ing system design will affect the rate at which applicants are matched
with potential carpool partners. Design of the matching system must
strike a balance between the quality of matches and the quantity of
matches.
One widely used indicator of the success of a matching program is the
rate at which applicants are matched, or the matching rate. In most
instances this is expressed as a percentage; for instance, one would
say that 30 percent of the applicants were matched with at least one
other person. The use of such rates are useful indicators of how well
the service is operating in the sense that higher matching rates imply
that more people will be provided data enabling them to form carpools.
In evaluating the design of a matching program and trying to decide
which of the matching criteria discussed above to use and how strictly
to use them, matching rates can be used as an indicator of the program's
utility. Programs with strict, narrow, multiple criteria will achieve
fewer matches than programs with few, broad criteria.
On the other hand, the use of matching rates can be deceptive. As
stated above, there is the issue of the quality of the matches; a long
list of carpool candidates is of small value to the applicant if most
of those on the list would require substantial compromises of travel
time, work schedule, travel distance, or personal preferences. Great
compromises will lead to unstable, short-lived carpools.
68
-------
Carpool matching programs which use a minimum-number-of-matches
criterion for matching can always have apparent high matching rates,
but may be no more effective at promoting carpools than other programs
which use better matching criteria but may exhibit lower matching rates.
Furthermore, any program will deliver higher matching rates with more
data. A low matching rate may not imply the failure of the program
but rather the failure of the total system to achieve enough applicants.
Matching rates should increase in time as the data base grows in size.
Consequently, matching rates should be compared carefully. While a
high matching rate is generally desirable, and while the matching rate
is one of the few quantitative indicators of the utility of a matching
system, the value of a system should be judged by other criteria as
well.
What is needed, but what is not yet available, is an indicator of the
rate at which a carpool matching system results in the formation of
carpools. As will be discussed elsewhere, there are as yet few data
on the rate at which carpools are formed through any system. More
important still, there are even fewer data on the long-term steady-
state use of carpools; most data are for short-term results from
programs instituted within the past few months.
Additional material on the carpool matching program matching rate is
in Section IX.
Output Form
Most matching programs are designed to produce output in a form which
can be directly used by the applicant. The most widely-used format
is that of a letter, addressed to the applicant, and identifying his
potential carpool matches. The output format is not critical except
69
-------
that it should include enough information for the applicant to decide
whom to contact and how to contact them. For destination-based systems
the output can be fairly simple. For region-wide systems more infor-
mation must be supplied.
For regional systems the minimum output information for each potential
partner appears to be:
1. Name
2. Telephone number where he can be reached
3. Some indicator of where he lives and works,
either his address or some portion of it or
the distance to his home, and his work loca-
tion or the distance to it
4. Work schedule.
Optional information which either could be on the output form or which
the applicant might ask for in a telephone call could include:
1. Information on preferences such as smoking,
ride/drive, etc.
2. Personal characteristics such as age, sex,
education
3. Employer name
4. More detail on employment location, home
address (but see discussion of privacy
issue below).
With respect to such optional information, there is an obvious need to
provide enough information for the applicant to decide who to contact.
On the other hand, the publishing of lists with preferences and other
data which may be considered "personal" may tend to discourage use of
the system. In the beginning, when such systems are first in use and
the data bases are small, there appears to be little danger in printing
only essential information. As data bases grow and lists of potential
partners become inordinately long, then perhaps more optional infor-
mation should be printed on the output.
70
-------
Privacy And Security
One major difference between destination-based, employer-operated
carpool matching systems and regional matching systems is the greater
need to provide privacy and security features in the regional system.
The need for privacy and security of personal data is perhaps not
apparent. In addition it is not clear that the security/privacy issue
would in fact arise to the degree postulated here. It may indeed vary
among the parts of the U.S., but throughout the research for this re-
port the privacy/security issue was continually mentioned in discussions
of the merits of various types of matching systems.
In the Boston area, for example, all the firms who were contacted and
who had decided to carry out their own carpool matching system rather
than only provide questionnaires for the WBZ-ALA. system mentioned the
privacy/security issue as at least one reason, and often a major reason,
to undertake a separate employer-based system.
There are two personal dimensions to this issue. One aspect can be
referred to as the privacy issue and amounts to a fear of misuse of
the data by those who are provided it by the matching service. Objec-
tions to providing personal information include the fear of being dis-
turbed by telephone calls or mail from numerous strangers, the fear of
persons who would use the information for criminal purposes (such as
to commit burglary during the work hours), and the fear of being asked
to form carpools by persons who may be intending assault, robbery, or
other illicit behavior, or may simply be people of a different social,
economic, or cultural class with whom the applicant would not wish to
ride.
The other aspect regards the security of the information and amounts
to concern over misuse of the data by those operating the system. One
71
-------
concern might be that lists of carpool applicants would be sold to
advertisers, much as license plate registration lists are now sold.
There is an additional institutional element to this issue when dis-
cussing combination systems wherein employers submit blocks of data
to a central system. Employers express reluctance to provide data
which they feel could be used to their disadvantage. Some seem to
fear that other firms (their competitors) will determine how many em-
ployees they have, or that someone will in some other way abuse
the collective data. One administrative officer objected to pooling
data with a nearby competitor because he feared the formation of car-
pools might lead to exchange of information on employee salaries.
Some of these fears may be ill-founded; others may be fears expressed
theoretically without their actually existing among many of the popu-
lace; others may be seen as trivial. Nonetheless, the existence of
these concerns among those who influence the operation of such systems,
such as company administrative officers, must be reckoned with. Re-
gardless of the merit of the fears they must either be assuaged or
used in the system design.
The best solution to the privacy issue appears to be three-fold:
1. Design the system to provide potential applicants
every reason to trust the system to protect their
privacy. The officials who administer it should
be as well-known and trustworthy as possible. As
an end to this, internal company distribution and
collection of carpool questionnaires has great
potential because it enables employees to deal
personally with people with whom they work every
day, and it minimizes the amount of personal in-
formation in the output. The literature should
clearly state how all personal information is
used and protected.
2. The output should provide as little personal in-
formation as is reasonable for good matching.
An office telephone can be printed in lieu of a
72
-------
home telephone. Street addresses should not be
printed.
3. Provide safeguards so that employer data cannot
be provided to others than the employer or the
system operator. That is, prevent the compila-
tion of lists by employer which could be misused
and violate an employer's trust in the system.
The concern over this issue may be premature; regardless, sensitivity
to it may result in more successful carpool systems.
Maintenance
One purpose of carpool systems is to initially cause an increase in
the use of carpools by assisting in their formation. A second purpose
is to maintain a desired amount of carpooling by providing a continuing
service to people. New service would be needed when a carpool member
changes jobs or home address or work schedule, when a new employee
joins the firm, or when a new resident moves into a region.
One way to provide the matching service for such changes would be to
re-run the entire program at stated intervals. For example, the entire
carpool survey could be re-run every six or twelve months. This would
be feasible for some destination-based systems but may be burdensome
for large systems. In addition, this delays the provision of carpool
matching data to people whose needs have changed.
Another way to provide this maintenance function is through central
files. With a suitable filing system all the existing data could be
organized to permit one-at-a-time manual matching. A new applicant
would write or telephone the agency which administers the carpool
system, provide his address, work location, and other matching data,
and be given the names and telephone numbers of several people with
compatible characteristics. This process could become expensive for
large data bases, however.
73
-------
A third way is to utilize a special computer program designed to only
edit, correct, or update the prior output. Such a program would search
the files for only the applicants whose characteristics are being re-
vised, match, and print out matching lists only for these revised
applicants. Such a program allows automated revisions in large data
bases, at modest cost.
Transportation Planning Data
Questionnaires for carpool matching service may be a useful way to
obtain data on transportation needs and characteristics. Several
operational computer programs include analysis of questions relating
to past or present transportation habits, preferences for transporta-
tion systems, travel times or routes, and other information of value
to transportation planning. Programs generally would separate these
data from any data used for matching candidates and instead prepare
special output for use by planners. One use of such data is to identify
the potential for vanpools or buspools, new bus routes, or increased
transit service among those interested in carpools.
One form of such output is a density matrix, which shows the number
of origins from within one area (one origin zone, say) to a particular
destination (the CBD, say), thus revealing where there are concentra-
tions of travelers with similar origins, destinations, and routes.
Inclusion of transportation planning functions in a matching program
has the potential for great increases in the utility of the program
for planning purposes with but small increases in overall costs.
74
-------
PRESENT CARPOOL MATCHING SOFTWARE
General
The first part of this section of the report has outlined some general
concepts of how matching programs operate. With these concepts in
view, this portion of the report will review several computer matching
programs now in use.
It should be borne in mind that computer carpool matching as a concept
is only a few years old and that few carpool matching systems have been
in operation for more than a few weeks or months. Thus it is probably
premature to judge how well particular programs will work as part of
an overall system, even though some indication may be gotten of how
the programs themselves operate. Thus, as was discussed earlier, what
is still needed is a measure of how much carpooling actually results
from the operation of the various matching systems.
Carpool matching is in a very dynamic state as this report is being
written. Programs are being revised as they are used, new programs
are being written, and other programs are being discarded. Many pro-
grams are written in several versions for different circumstances or
applications, and there is more experience with some versions than
with others. Furthermore, this report does not review all computer
programs and may omit some which later prove to be of great importance.
Examples Of Existing Programs
Widely Used Programs - Several computer programs for carpool matching
are well documented, have been widely distributed, and are in use in
multiple systems. Three of these will be described here. For addi-
tional detail .and for how to obtain these programs refer to the Refer-
ences and Bibliography. The descriptions below are based in part upon
References 29 and 37.
75
-------
FHWA Carpool Matching Program - The program prepared originally in
1972 by the Federal Highway Administration (FHWA) has subsequently
been distributed to approximately 190 agencies and companies. The
program is very well documented (see Bibliography), is available free
of charge to anyone planning to operate a matching system as a public
service, and is written in the American National Standards Institute
(ANSI) COBOL language.
Origin and destination coding are done by a grid system. Grid cells
are square or rectangular, and the array includes a maximum of 49 cells
in each direction, or a maximum total of 2401 cells in the area to be
coded. To provide more detail in downtown areas, the program permits
the use of two grid cell sizes simultaneously, with the inner (smaller)
grid cells being half the length and width and one-fourth the area of
the outer grid cells.
Origin matching searches for matches in the eight cells around -the
origin cell if at least eight matches are not found in the origin cell.
Work schedules can be matched exactly, or a range of work times can be
specified. There is no search at the work end. Each work grid cell
is matched separately. The program evidently was written chiefly for
destination-based matching, and is inefficient for region-wide match-
ing. FHWA plans to revise this program to include a search of adjacent
grid cells at the destination end.
The grid system can be limiting to the use of this program for regional
matching because of the coarseness of the grid. FHWA recommends a grid
cell of one mile square. In many metropolitan areas this would be too
large an area for destination matching in major employment centers.
If smaller grid cells are used, they may be too small for effective
matching in suburban or fringe areas where transit is poorest arid car-
pool potential the greatest.
76
-------
Special conditions can further complicate this. In Dallas and Fort
Worth the FHWA computer program is being used for a regional matching
system, but with the two cities it was not possible to use a detailed
grid in both cities. However, the grid used has a variable cell size,
varying from about one mile wide in the downtown areas up to two miles
wide in some outlying areas. Nonetheless, the one-square-mile grid
cells encompass dozens of city blocks in the two downtown areas.
(Dallas is considering modifications to the program to overcome this
potential problem.)
In their report to FHWA, a consultant surveying the available software
criticized the FHWA program for its excessive output of "extraneous
data." Also, some users have reportedly had difficulty adapting the
29
program to some types of computers.
Census Bureau Carpool Matching Program - The Census Bureau's "Census
Use Study" has developed a carpool matching computer program designed
to take advantage of certain computerized census data. There are
about 200 metropolitan areas where Dual Independent Map Encoded
Geographic Base Files (DIME/GBF, or just DIME files) have been compiled.
These are files of the topological features in a metropolitan area in
a form useful to a computer. In essence the DIME files identify which
census tract contains a given street address. The Census Bureau car-
pool program thus uses as an input the identity of the census tract in
which the applicant's origin and destination are located. In cities
where the DIME files are current, the use of this program reduces the
work involved in geo-coding; unfortunately, however, the DIME files
are not up to date in many cities and are most out of date in the
developing fringe of metropolitan areas where carpooling has the great-
est potential. Nonetheless the DIME files can assist in geo-coding;
the Census Bureau program is being used in San Francisco with the DIME
files, and manual geo-coding is used to supplement the DIME files.
77
-------
In addition, the Census Bureau states that their program can use any
alternative form of geo-coding such as zip codes, and the documentation
describes a manual process for using census tract maps with their
program.
The Census Bureau program documentation was recently published but the
program has subsequently been revised. Potential users should contact
the Data User Services Offices for the latest information. The follow-
ing description is based upon Reference 38.
In operation, the centroid of the census tract (or other zone) is es-
tablished and identified by x and y coordinates. Matching on the ori-
gin end involves first the origin census tract and then all surrounding
tracts whose centroids are within a predetermined distance from the
centroid of the origin tract. This process continues, with expanding
radii to the centroids of surrounding tracts, until either a minimum
number of matches is reached or until the distance to the centroid of
the surrounding tracts being searched is equal to the distance from
origin to destination. Searching surrounding tracts on the destination
end is not included in present documentation but is planned as a later
modification.
Matching on work schedules uses criteria established by the user.
Other criteria include common work days and drive/ride preferences.
The program was written in FORTRAN and is fairly complex. Little user
experience has been documented.
The revised version of this program being used for the San Francisco
carpool matching system incorporates several additional features.
Multiple destinations are now provided. These destinations are des-
cribed as landmarks, major employment centers, or other areas. Des-
tinations not in the standard list are coded by their zip code.
78
-------
Destinations are assigned x/y coordinates for use within the program.
Besides the search for matches among adjacent zones on the home end,
the modified program searches for matches among destinations whose
x/y coordinates are within a specified radius of the applicant's des-
tination. The potential matches are ranked for their suitability
according to a method which assigns weighted values to the nearness
of potential partners' homes, nearness of destinations, degree of
agreement in work schedules, etc. The matches with the best "score"
from this process are then listed in the output, rather than just
those matches meeting set criteria. This program has been used for
the first 15,400 applications, and the results are being analyzed.
1 Further refinements in the program itself and in the weighting factors
are expected to result from this analysis. (Of the first 15,400
applications processed, 13,000 received matching lists with an average
length of twelve names, but some matches appear to be of little use.)
Burroughs "Operation Energy" Matching Program - Of the firms in the
private sector, Burroughs Corp. has probably done the most to promote
the use of computer matching. Its "Operation Energy" package, con-
sisting of the computer program description as well as management
suggestions, is provided free of charge to any organization desiring
it. The description below is based upon that package, Reference 39.
Origin coding is based upon a grid system, with grid cell size deter-
mined by the user. Only a single cell size can be used, but the grid
can be as large as 99 by 99 (9801 cells). Within the program, the
x and y coordinates of each cell are used rather than just the cell
designations. For destinations, a combination of grid and numeric
designations is used. Up to 99 destination designations are defined.
These can be described to the applicant as areas, landmarks, buildings,
or other points. Within the program, the x and y coordinates of the
destinations are used in matching.
79
-------
In matching, the user specifies the minimum number of matches. The
program matches only applicants with the same destination code; there
is no work-end searching. Work schedules are matched within a range
of 15 minutes. Within the constraints of destination and work schedule,
the computer program locates suitable partners in the same origin grid
cell. If the predetermined minimum number is not found within the same
origin cell, adjacent cells are searched. The first adjacent cell
searched is the one most nearly in the direction from the origin of the
applicant to his destination; this is the reason the computer needs the
x/y coordinates of origins and destinations. The other adjacent cells
are searched until the minimum number of matches is achieved, working
from the cell most in the direction of travel to the cell in the oppo-
site direction. If the search of adjacent cells does not achieve the
number of matches desired, the next ring of surrounding cells is
searched, again beginning with the cell in the travel direction. This
process continues until either the desired minimum number of matches
is achieved or until a predetermined travel distance is reached; each
applicant may separately state how far (i.e., how many grid cells) away
from his home he'd be willing to travel. The directed, or vector, search
pattern is an unusual feature of this program.
The program is written in ANSI COBOL, is reported to be suitable for
small computers, is well documented, and has been used by several in-
dustrial sites for up to several hundred matches. Output can be
tailored to either one-company, multi-company, or regional use. One
option is that the operator can specify matches only among employees
of the same firm while still using the program for several firms
simultaneously.
Other Current Programs - The three programs described above are well-
documented and widely disseminated. The next programs to be described
are less well documented and less widely promulgated but merit interest.
80
-------
Denver, Colorado, Carpool Program - Section III of this report des-
cribed the Denver, Colorado, carpool matching system. As pointed out
there, the matching program was written by high school students, and
the program has not yet been well documented. Because of this, and
because of the unique matching logic, the program will be described
here in some detail. This description is based upon telephone conver-
sations with those involved. Note that the program has been changed
since Reference 29 was written.
The geo-coding (digitizing of addresses) process involves assignment
of numerical (x and y) coordinates to the home address and work address.
At the present time this is a manual process and is relatively time-
consuming. In the initial efforts this work has not been a major
obstacle because students or other volunteers can do the coding. But
for a full-scale effort geo-coding is a potentially significant cost.
The digitizing of addresses is essential, however, because of how the
matching program works, as will be described. It is estimated that a
person can digitize 200 to 300 addresses per day by hand.
A computer program has been developed which would permit machine digi-
tizing of addresses. It is estimated that this would cost $10-15,000
for the Denver area, and has not been done because of a lack of funds.
A similar program could be adapted to other cities, and reportedly
could be adapted to curving streets as well as to the more or less
regular street network of Denver.
The matching process itself operates in several steps. Before entering
the data, starting times for work are rounded to the next earlier half
hour and quitting times are rounded to the next later half hour. The
computer then groups all people with the same (rounded) work schedules
in a file and operates on one such file at a time. Next, the machine
selects a list of potential carpool partners for each person in the
file. To do this, it first examines the file for persons whose work
81
-------
addresses are within a one-mile radius of the applicant's work address.
(This radius can be altered and 0.2 mile has been used for some large
firms.) The computer calculates the distances between the homes of
the applicant and his potential partners. The machine keeps track of
the ten names whose home addresses are closest to the home of the
applicant and who work within the 1-mile radius. This is the "cluster
method." (See Figure 2.) If the third-closest person is more than
two miles from the applicant's home, however, the matching process
goes into another mode, the "vector search."
In the vector mode, the computer uses a more exhaustive way to examine
the list of people who work within the 1-mile radius of the applicant.
Potential partners are rejected if their homes are outside a circle
with its center at the applicant's place of work and a radius equal to
the distance from the applicant's home to his place of work (see dia-
gram in Figure 3 ). For each potential partner inside this circle, a
computation is made of the extra distance the applicant would have to
travel to pick up the potential partner. This extra distance, referred
to as "displacement," is a measure of the inconvenience or disruption
involved in carpooling. The measure is imperfect because airline
distances are used instead of travel distances, but provision of eight
names should reduce the difficulties with that compromise. It is be-
lieved that this "vector search" technique is unique to the Denver
program. Using this displacement as a measure of the suitability of
a potential carpool partner is considered by the program authors to
be superior to, and more flexible than, methods based on grid designa-
tions, zip codes, or other techniques of coding home addresses.
The matching program was written in FORTRAN IV, originally for a
UNIVAC 1106/1108 computer. It is now being adapted to an IBM 360/50.
On the UNIVAC, processing 500 people takes about 5 minutes. About
64K core storage words are needed for 2000 applicants. The amount
of memory needed is proportional to the number of applicants in the
82
-------
OJ
6
o
I
}-l
00
o
00
Ci
O
4-1
Cfl
o
o
CL
O
T3
cd
M
o
r-l
O
t-l
0)
c
Q)
o
a
o
CJ
Crt
I-l
0)
CM
U>
3
i-l
83
-------
e
(8
M
bO
O
00
C!
43
O
o
o
a.
1-1
«
CJ
o
u
a)
o
u
(-1
01
C
0)
o
d
r-i
O
43
4J
43
O
a)
(!)
Crt
U
(U
(U
M
3
00
r-l
84
-------
file with identical work schedules. Documentation has not been pre-
pared, but operation is apparently straightforward as about 20 other
organizations use the programs.
Conversion to regional matching - In present practice, matching is
being done on a destination basis. As can be seen from the preceding
description, the program can be used without change for the regional
matching when it is underway. The radii of the destination circle
would be adjusted to suit the data base; longer radii are needed for
fewer candidates. While some employers now use 0.2 mile for the des-
tination search, it is planned to use 1.0 mile for the region.
When the regional matching service is underway, the existing employer
data will be combined with new applications from the general public.
The version of the program designed for Min-Comp to use for the re-
gional matching has a built-in security feature which is intended to
prevent anyone from gaining access to another employer's files.
Printouts will include the home-to-home and work-to-work distances
and the work telephone number of potential carpool partners, but will
not include home addresses. Reportedly, with this feature no firms
have so far objected to pooling their data with the regional data.
(Of course, there has as yet been no reason to object inasmuch as the
regional system is not yet operational; the true test will occur when
the regional system operation begins.)
The overall process actually uses three programs, including one for
maintenance and corrections, and another which produces density matrices
23
which can be used for such purposes as planning transit service.
Automatic Data Processing/Westinghouse program - Several Westinghouse
Broadcasting Co. radio or television stations are operating regional
carpool matching systems. Some of these stations (including WIND in
Chicago and KDKA in Pittsburgh, both discussed in Section III) are
85
-------
using the firm of Automatic Data Processing, Inc. (ADP) to perform
the matching service. ADP has written their own matching program for
this purpose. This program is available to any organization which
needs it, but is sparsely documented.
The matching process is designed for regional systems, and uses a grid
system. The origins and destinations are both identified according to
the alpha-numeric designations of the grid cells in which they are lo-
cated. This geo-coding is performed by the applicant. The program
can, however, use other types of origin codes such as traffic zones
or zip codes. There is no searching of adjacent zones if zip codes
are used (as in Chicago for origin codes), but with a grid system
adjacent cells are searched for additional matches (as with the des-
tination coding for Chicago).
The program prepares output only for applicants not previously pro-
cessed, but retains previous data for making matches with new appli-
cants. Processing time is a function of the number in the data base,
the number of new applicants, and computer model, but a typical run
for new additions to a data base with a few thousand applicants would
require 15 to 30 minutes.
The program is written in COBOL and has been successfully run on
Honeywell 1250 and 2200 and IBM 360 computers. With the Honeywell
machines, the program used 32K core capacity and four tapes. With
the IBM machines, the program used 64K core capacity and one disc
and three tapes.
Other computer matching programs - The programs described here are
those judged to be of greatest interest to organizations who want to
establish regional carpool systems. Many other programs exist, and
some of these were mentioned in Section III of this report. Some
additional data on the programs available are in Reference 29, pre-
pared for FHWA.
86
-------
OBSERVATIONS AND SUMMARY
General
Of the numerous computer programs available for matching prospective
carpool candidates, only a small number have been well documented,
widely distributed, and used for substantial amounts of data. Even
among these few there is insufficient experience to enable a conclusion
of which program is best (if any is). Furthermore, there are some
additional programs which are not well known but may offer promise
for use in regional carpool systems.
Based upon the review undertaken for this report, some tentative
conclusions and recommendations can be offered. Above all, however,
it is important to bear in mind that some of the major computer pro-
grams are being revised or refined, documentation is becoming avail-
able, and experience is accumulating rapidly, so these conclusions
could be shortly rendered invalid by changing circumstances.
Tentative Conclusions On Matching Software
1. The central purpose of the matching program is to identify poten-
tial carpool partners to an applicant for the matching service. The
program thus should identify a reasonable number of persons whose
transportation route and schedule are nearly the same as the appli-
cant's. The program need not supply more than enough names to pro-
vide the applicant a modest amount of choice among his carpool part-
ners. Theoretically, the program should attempt to provide a list of
the best (i.e., most convenient) partners. Inasmuch as this is
actually impossible, the program should provide enough names that
the best choices are probably among them.
87
-------
2. To achieve the goal stated in (1) above probably requires a multi-
step matching process; that is, it is probably necessary to use a pro-
gram which searches for matches over a variable area around the origin
and/or destination, or which searches for matches along the applicant's
route.
3. The size of the geographical area searched for matches should be
related to the density of possible matches. Consequently, coding
should allow greater detail in densely developed areas such as central
business districts than in residential areas. This in turn implies
that with grid systems the grid cells need to be smaller in employment
centers than in residential areas, or that non-grid systems should be
used in employment centers.
4. Because of the need for a multi-step matching process, the pro-
grams now in use which seem likely to be most effective are those
which use either a grid or an x/y coordinate system for coding of
origins.
5. Matching criteria other than origin, destination, and work schedule
are probably of only minor importance and may be difficult to implement.
If personal preference data is desired it seems more practical to
merely list these data on the output forms for the use of the appli-r
cant, rather than actually forming matches on it.
6. Steps should be taken to safeguard the privacy of data which is
not essential to disclose in the output. Furthermore, these safe-
guards should be clearly identified to alleviate any fears and to
preclude complaints arising from erroneous impressions. Specifically,
home addresses and telephone numbers should not be disclosed by re-
gional carpool systems if alternatives are available. Alternatively,
the applicant should be allowed to specify what contact address or
telephone number he wishes to release.
88
-------
7. Any program should allow for readily revising the carpool data,
either manually through master lists or automatically using the com-
puter. The maintenance service should be publicized. Lack of this
feature is a potential major drawback of most broadcaster-operated
systems.
8. For destination-based matching, or for the destination-based por-
tion of a combination system, several programs are adequate. The
FHWA, Census, and Burroughs programs are all operational, documented,
and widely available. Choice among these must at the present time be
made on the bases of computer capability, preference, and circum-
stances .
9. For region-wide matching, the suitability of present programs is
less clear, in particular because so little experience has accumula-
ted. At the present time the Denver program, with both cluster and
vector search methods, appears to be very promising, but is not yet
proven for regional matching and is not documented. The Burroughs
program appears to be second best, with its version of vector search,
but apparently also has not been proven for region-wide service.
The Census Bureau program is complex and has proven difficult to
implement but may ultimately be very workable, especially in cities
with current DIME files. The FHWA program is limited in usefulness
for regional matching because of its grid system and one-by-one
destination matching. All other regional programs surveyed for
this report, chiefly those used by broadcasters, are either too
primitive in their matching (most use one-step geographic matching)
or are not well enough developed to recommend.
10. What will be needed before final conclusions can be drawn about
any carpool matching program is information on the long-term effec-
tiveness of such programs for encouraging the formation and use of
carpools, not just information on data base size, matching rates,
computational cost, or other characteristics of the program itself.
89
-------
11. Any matching system should be suited to the circumstances where
it will be used. In spite of the above comments, there may be areas
where, for example, single-step matching based upon zip codes is
effective, if not perfectly satisfactory. The matching process itself
is not carpooling, and only serves it. Selection of a program should
reflect the local needs, abilities, and preferences of the matching
system operator and the public.
90
-------
SECTION V
ORGANIZATION OF CARPOOL SYSTEMS
GENERAL
Some very helpful literature has previously been published on the sub-
ject of how to organize an effective carpool program. The reader is
directed to references 9 , 37, and 39 in particular. This section will
summarize the results of the survey for this report, with an emphasis
on regional systems.
EXPERIENCE WITH ORGANIZING CARPOOL SYSTEMS
Carpool matching systems have been operated by several types of
organizations. Experience with three of these types will be summarized,
Broadcasting Station Systems
Two of the carpooling systems investigated for this report which have
the largest data base are operated by broadcasting stations. These
two, operated by WBBM in Chicago and WBZ in Boston, were described in
Section HI- This section will review the organization aspects of their
carpool campaigns.
WBBM, Chicago - The WBBM carpool system is entirely destination-based,
and consequently the employers are the key to its operation. The
organization is simple in the sense that it consists only of an
arrangement between WBBM and each of the employers. The radio station
91
-------
provides the matching service, including keypunching, computer match-
ing, and the survey forms. The employer provides the distribution
and collection of forms and any incentives he may wish to use to en-
courage carpooling.
The incentive for the radio station is apparently the public relations
exposure, among major employers in particular. WBBM is an all-news
station and thus its audience (and its advertisers) are apt to include
people in the management of such firms, perhaps more so than stations
(such as WIND in Chicago) which appeal to a younger audience.
The incentives to the employers are the no-cost matching service and
the public relations impact of receiving mention on the air by WBBM
as a participating company.
WBBM does not prescribe how the employer should distribute, promote,
or collect the carpool survey, but the survey form is designed to be
inserted in pay envelopes.
This organizational structure is simple and apparently effective in
the sense that over 100 employers with over 100,000 employees have
signed up for the service. It is too early to determine, however,
how effective the organization will be at attracting applicants.
WBBM estimates that 10 percent of employees will participate.
WBZ, Boston - The WBZ carpool system is probably the oldest regional
system. It is nearly the opposite of the WBBM system. WBZ shares
responsibility for overall organization with the ALA Auto and Travel
Club. Solicitation of applicants is directed largely at the general
public, although some promotion has been directed at employers as well.
Formally, WBZ and AIA have responsibilities clearly defined by an
agreement between them. In essence, WBZ contributes promotion and
92
-------
materials, and ALA contributes data processing and clerical work.
This arrangement seems workable and is apparently amenable to the
participants, but the commitment expires in August 1974 (although
it could be extended).
The WBZ system has processed approximately 15,000 applications. This
is more than most regional systems, so in that sense the organization
seems effective. On the other hand, 15,000 applicants is a very
small fraction of the Boston area commuting population. Based upon
the experience in other cities with destination-based systems, and
based upon the comments from Boston-area employers, the WBZ-ALA
system would have drawn more applicants if more employers had strongly
supported the program, for example by providing incentives and by
collecting applications. Large employers have been wary of the lack
of privacy protection and for this and other reasons have preferred
to operat
page 18.)
to operate their own systems. (See the discussion which begins on
Summary - Broadcasting stations may be extremely useful in promoting
and publicizing carpooling and carpool matching systems. It is ques-
tionable whether a broadcaster can enlist enough public support to
operate an effective system on its own. Employer support appears to
be important to attracting a good data base. Furthermore, broad-
casters involved in present programs have planned only one-year com-
mitments; it appears unlikely that a broadcaster will wish to carry
the central role in a carpool system for more than a year or two, at
least without compensation.
Systems Operated By Other Elements Of The Private Sector
A few systems involve the private sector without broadcasters playing
a major role. The only system of this type investigated for the pres-
ent report is the Denver system, which is at present being operated
by a business organization and an auto club, with technical support
93
-------
from a high school. This organization has been effective so far but
the region-wide system is not yet in operation. In addition, the
present plan is to turn over the major role to the regional council
of governments if funds can be obtained; this may be indicative of
the trend for such organizations, because of factors such as the cost
involved. In Denver, for example, the equivalent of two or three
full-time people have been committed to this effort, at no charge to
23
the employers receiving the matching serviced
Systems Operated By Government Agencies
In Dallas, Knoxville, Washington, D.C., and in several cities in
Connecticut and California, government agencies are involved in or
are the chief operators of carpool matching systems. Many more
government agencies will become involved in the near future, particu-
larly as a result of the FHWA efforts to promote carpool systems by
providing Federal funds.
In Dallas and Fort Worth the organization is characterized by the
joint efforts of the two cities. Responsibility for the work is also
shared with the employer (see Section III). Dallas reports that
their system suffers because the Chamber of Commerce committed itself
25
to support of the systems operated by two radio stations.
In Knoxville the organization is led by a committee representing
government and commerce, which appears to have been effective inas-
much as at least half the employers are participating. The Knoxville
organization has the personal support of the Mayor and has included
30
the transit authority for coordination of bus and carpool service.
In Connecticut, California, and Washington, D.C., the government
agencies are supplying mostly technical advice and data processing.
Overall organizations include the business sector and broadcasters.
94
-------
In Connecticut, for example, several different organizational struc-
tures are used in the several cities involved. The chief character-
istic of these organizations seems to be an ad hoc coordination of
the private and public sectors with the state carrying the burden of
the long-term commitment to maintaining the data base. It may be
worthwhile noting that the latest of these three agencies to become
involved, California, is planning to charge for their services rather
than to donate them.
Summary Of Experience
Most carpool systems are being operated by ad hoc organizations con-
sisting of one or more private sector elements and one or more govern-
ment agencies. Several types of structures are in use and are
effective.
The most effective systems have had the strong support of employers,
especially major employers. This support includes active support and
promotion within the firm among the employees. Broadcasting stations
have been effective at promoting carpooling and carpool systems, but
it is questionable whether they can carry the chief responsibility
for operating a regional system, particularly because that role re-
quires a long-term commitment which broadcasters may wish to avoid.
Government agencies are becoming involved in carpool systems at a
rapid rate as a result of the provision of Federal funds. Because
government agencies are suited to long-term responsibility of the
type involved with maintenance of carpool systems, it seems clear
that regional systems should and probably will rely on such agencies
for leadership and data processing. .
Regional matching systems are most likely to be effective where major
employers strongly support the system and encourage the use of carpool
95
-------
matching and carpooling itt>elf among their employees. For this rea-
son, it may be that a dual approach will be needed. Some employers
may be reluctant for any of several reasons to pool their data with
those of other employers. In such a case it may be well to provide
good-quality destination-based matching service to these employers,
while at the same time providing region-wide matching service for
cooperating large firms, smaller firms who couldn't do their own
matching, and the general public. This dual approach may not maxi-
mize the number of applicants in the regional data base, but it will
maximize the number of applicants in matching systems. The probable
necessity for this dual approach is one reason there is a need for a
government agency to be the chief administrator of matching systems,
in order to coordinate the two approaches and provide technical sup-
port of not just one but two systems.
FUNCTIONS AND COSTS
*
Responsibilities In Regional Systems
Most initial carpooling system efforts have been done as a public
service by the private sector. As Federal funds become available
and as the numbers of systems and applicants grow, the burden of
costs will probably shift to government agencies.
The private sector should still be enlisted to support the carpool
system. Some ways this can be done are proposed in Reference 9 .
For regional systems, a brief outline of responsibilities is suggested
below.
Leadership - Especially at. the beginning of a system and while trying
to develop the data base, an ad hoc coordinating committee should
provide overall direction. Membership should include business lead-
ers, government agencies, and key employers.
Carpool systems serving a metropolitan region, whether defined as
regional or combination systems.
96
-------
Promotion - Broadcasters should be enlisted to promote the system,
especially in the beginning phases. Most will probably undertake
this at no cost to the public. During long-term system operation,
periodic public service announcements will no doubt be carried as
part of the station's regular programming. Promotion to employers
should be through the Chamber of Commerce, downtown business coali-
tions, or similar membership groups. (See also Section VI.)
Data Processing - With the dual approach recommended here, it is best
that a government agency retain and process the data. Private sector
elements may be a viable second choice, however, particularly where
such systems have been begun by private groups. In any event, no
firm should be prevented from doing its own. If employers wish to
operate their own systems,technical support should be provided by the
regional system management to the employers so that the employers'
internal programs will be effective.
Costs
Costs have not been extensively surveyed for this report, but from
the sources reviewed several useful indicators of the costs of ad-
ministering a system can be presented here.
Promotion Costs - Major promotion on radio and television could
probably be donated by broadcasters, much as they have donated time
to present efforts. Newspaper advertisements may even be donated,
although there is less precedent for this. News coverage in both
the press and other media will of course cost nothing. Paper promo-
tion tools such as bumper stickers, brochures, and flyers are avail-
able from the Highway Users Federation for a few cents apiece. Simi-
lar publicity materials custom-prepared for each system could be
produced in-house at employers at minimal expense.
97
-------
Data Processing Costs - The Denver system has arranged for data
processing at a price of 15 cents per person initially and 10 cents
per person for maintenance changes. When combined with geo-coding
(by hand), and other clerical and administration costs, Downtown
Denver estimates the cost to be $59.50 for 100 employees, or 60 cents
each. In California, the state now estimates the charge for the total
package of initial processing (instructions, maps, keypunching, data
processing) to be 50 cents per person, with a charge of 25 cents per
person for revisions (maintenance).
Distribution Costs - Mailing questionnaires and computer output would
cost 20 cents per person for postage alone unless collection and dis-
tribution were handled through the employers. The latter method is
less expensive, less complicated, and more effective.
Administration - Judging from systems now in operation, only a small
number of people are required for a full-scale system. This may be
judged from the fact that the systems now being run by the private
sector have all been done without hiring additional staff for promo-
tion or data processing.
Denver has submitted a proposal for funding an interim carpool office
for five months until the FHWA-sponsored COG office is in operation.
The total cost of this office for five months is estimated to be
$32,500, which includes three full-time and one half-time staff mem-
bers and clerical expenses. (Denver's metropolitan area had a L970
population of approximately 1.2 million.)
The city of Dallas estimated that approximately $170,000 would be
the total cost of a two-year carpool-buspool campaign. This cost
includes data processing, printing, publicity, and so on. Mainten-
ance of the system was estimated to cost $20,000 per year. (Dallas
had a 1970 metropolitan area population of 1.5 million.)
98
-------
The California DOT carpool program, which involves statewide matching
service plus administration of promotion and incentives, is estimated
by the state to require approximately 26 man-years of effort for the
CDOT effort at the headquarters level and in the district offices.
(California had a statewide population of 20.0 million in 1970.)
Costs Of Incentives - The costs of explicitly provided incentives
will not be estimated here, owing to their wide variability.
Total Costs - Based upon the estimates above, it can be estimated
that a regional carpool matching system, organized as described here,
will cost on the order of 10 to 50 cents per person to initiate, and
only a few cents per person per year to maintain, exclusive of major
promotion and incentives.
99
-------
SECTION VI
MARKETING
GENERAL
This section of the report will examine techniques for marketing both
the concept of carpooling and specific carpool matching programs. As
in past sections the emphasis here will be on the results of the
investigations undertaken for this report. Planners of carpool sys-
tems should also refer to previous literature on marketing carpooling;
the literature from the Highway User's Federation (HUFSAM) is particu-
larly helpful (Reference 37).
As used in this Section, "marketing" includes four concepts: promo-
tion, price, place, and product.
Promotion is the visible aspect of marketing, including advertising
and public relations. This Section will emphasize promotion because
the public needs not only to be informed about carpooling but also
must be persuaded to form carpools, much as consumers are persuaded to
desire other products or services. Promotion includes selling the car-
pool concept to employers and community leaders, as well.
Price involves two elements with respect to carpooling. First, there
is the price to the applicant for the matching service. In this sense
price also involves the effort or inconvenience associated with the
application. Second, there is the price of joining a carpool. This
price may be the inconvenience of sharing rides. In return, the car-
pooler receives certain benefits.
100
-------
Place is the idea of how, where, and from whom to obtain the service
being marketed. In this case, one function of marketing is to inform
people of who runs the carpool system and how to participate.
Product is both carpooling and the carpool matching service (actually,
both are services rather than products). The marketing, especially
promotion, must demonstrate the desirable features of the product so
that people want it. Defining what the product (or service) actually
is directly affects its appeal and thus is part of the marketing process.
Carpooling must be marketed, not merely "publicized," because people
must be encouraged to change their habits in order to increase car-
pooling. Successful carpool campaigns should include these four as-
pects of marketing in the initial planning in order to both develop
a good product and convince the public of its value.
The need to actually persuade people to form carpools, rather than
merely inform them of the carpool system, is supported by the findings
of a behavioral research study performed for the California Department
of Transportation (CalTrans) and reported in Reference 33. In that
study, it was found that 64 percent of surveyed drivers in Los Angeles
could be termed "hard-core non-poolers," or people who would be difficult
to motivate to form carpools and would require negative incentives, such
as government regulations to restrict parking, to achieve increases in
carpooling. This study found that the remaining minority of the popula-
tion would form carpools if adequately motivated. Motivation would
include explicit incentives, discussed in Section VII, but also would
include emotional desire, social considerations (peer approval, e.g.),
and an understanding of the personal benefits of carpooling. Marketing
can be seen as a process by which one can influence the perceived
desirability of a product, by informing, cajoling, and persuading.
(Actually part of the marketing may be done for the local system by
outside sources. The U.S. Department of Transportation and Exxon, at
least, have been widely advertising the merits of carpooling, which may
help to improve the public image of carpools.)
101
-------
MARKETING OF PRESENT CARPOOL SYSTEMS
Regional Systems
Existing carpool matching systems have had varying amounts of marketing.
The broadcaster-operated regional carpool systems have been characterized
by substantial promotion or publicity. The promotional campaigns for
several recent systems were described in Section. III. Promotion included
newspaper advertisements, billboards, special major television pro-
grams > automobile races during rush hour, and frequent announcements
(commercials) throughout the daily radio or television broadcasts.
Large quantities of questionnaires were delivered to bulk distribution
centers such as shopping centers, highway toll booths, and restaurants.
In the Boston area, for example, it would seem likely that nearly every
commuter had heard of the WBZ-AIA Commuter Computer by March, 1974.
The Boston plan attracted attention as news, also, in local newspapers,
34
national magazines, and even in the scientific press.
Besides these promotional efforts directed at the general public, other
promotional techniques have been used to attract: employers to participate.
Industrial groups and Chambers of Commerce have supported the existing
carpool systems by official endorsements, editorial support in organiza-
tion journals, special publicity letters and brochures mailed to member
firms, and by provision of questionnaires and other direct support*
Publicity of these regional systems has had good results in the sense
that the general public has become aware of the system and many thou-
sands of application forms have been distributed; on the other hand,
applications have been returned at a low rate. Another effect of these
regional marketing efforts no doubt has been a more widespread under-
standing of the benefits of carpooling. It is possible that carpools
have been formed as a result of these marketing efforts even without the
use of the matching systems; certainly people are talking and thinking
about carpooling to a greater degree than ever before (or, at least,
since World War II).
102
-------
Destination-Based Systems
Destination-based systems operated within just one firm have certainly
received less external publicity than the regional systems. But such
systems are often heavily marketed within the organization. Typically,
the carpool effort would begin with a letter from a chief executive officer
to all employees, then be followed up by supervisors distributing and
collecting questionnaires. Other devices have included displays in the
company cafeteria, listings of carpools and feature stories on carpooling
in company newspapers, bulletin board posters, and public address announce-
ments. In the firms whose systems were reviewed for this project and in
other firms whose programs are described in the literature, such tech-
niques have resulted in response rates which have varied widely but have
been as high as 50 percent or more of employees requesting the matching
service.
Combination Systems
Present combination systems have been promoted both externally (i.e.,
publicly) and internally (i.e., within the affected firms.) In effect
such systems have been promoted as if they were separate destination-
based systems but with public support.
Recommended Approach
For the dual approach tentatively recommended in Section IH> namely com-
bination systems with both destination-based and regional matching ser-
vices, both public and internal promotion appear to be needed. Aspects
of these promotion methods will be discussed further in the ensuing
portions of this Section.
103
-------
MARKETING METHODS FOR CARPOOL SYSTEMS
Appealing To Employers
For the success of any carpool system it is imperative that employers
support the system, so it is important to consider them explicitly in
the marketing plan. Perhaps the most important way to do this is to
include the employers in the planning of the overall system. In par-
ticular, support should be solicited from industry and commerce member-
ship organizations and perhaps labor leaders as well. These organiza-
tions in several cities have undertaken substantial public relations
campaigns among their member firms in support of carpool systems.
One reason industry groups should be included is that they know who
the key people are in the major employers. They may also have ready
access to otherwise confidential data on numbers of employees. Ob-
viously it would be easier to plan a carpool campaign with the assis-
tance of such groups than with their indifference or opposition.
In many cities, such industry or commerce membership organizations
probably have been among those who took an active part in hearings on
Transportation Control Plans (TCP's) where necessary to comply with
the Clean Air Act. Such groups may have opposed measures which would
reduce mobility or reduced parking supply. Because widespread carpool-
ing can reduce the amount of other travel restrictions necessary for a
TCP, business leaders should be shown how their support of carpooling
systems can prevent the necessity of other more stringent travel con-
trols. This perspective is one reason industry groups have widely
promoted carpooling in Boston, e.g.
Employers should be shown several benefits to them of carpooling among
their employees.
104
-------
One approach used by the Associated Industries of Massachusetts (A.I.M.)
in their newsletter to members was to emphasize the point that strong
support of carpooling could eliminate or at least reduce their role in
TCP reductions. A.I.M. suggested that more severe government travel
restrictions would be imposed if carpooling did not occur voluntarily,
and that public relations would be adversely affected. Other advan-
tages of carpools (e.g., cost savings to commuters) were also publicized.
To help encourage companies to promote carpooling among their employees,
A.I.M. acted as a distributor for the WBZ-ALA carpool survey forms.
A.I.M. printed them, charging only costs, and even arranged for special
company messages on the questionnaires. This was a useful technique
of personalizing the WBZ-ALA campaign for the specific firm. A.I.M.
also publicized the HUFSAM seminars on carpooling, and organized meet-
ings among the system operators, employers, civic leaders, and govern-
ment officials.
In its literature on carpooling, Burroughs Corporation emphasizes the
value of carpool systems for employee relations and public relations
39
(as well as the technical aspects of reduction in costs). Many firms
are very conscious of their community relations, particularly if their
employees are largely from the inner city or from lower-income groups.
The public relations aspect of company carpool operations has been a
basic incentive for the employers participating in the WBBM "Ride To-
gether/Drive Together" system in Chicago, for example. Firms receive
mention on the air and in press releases, and consequently obtain free
favorable publicity. In this connection, system organizers may want
to allow a few firms to use the system on a "trial basis," and arrange
for good press coverage with favorable publicity for the firms. This
will help entice additional participants.
The HUFSAM literature identifies several reasons which can motivate
employers to participate in carpool systems. Besides the general
105
-------
public relations, employee relations, and energy-savings aspects men-
tioned elsewhere, other incentives include:;
1. Plant facility expenses. Fewer cars in use
for commuting reduce the cost of providing
and maintaining parking facilities. (Park-
ing spaces can cost up to $5000 each.)
2. Neighborhood relations. Carpooling can re-
duce employee parking and traffic congestion ?
in areas adjacent to the employer's facility.
HUFSAM also publishes a number of clear, useful brochures and advertis-
ing devices which can be helpful for communicating with employers (or
employees). The HUFSAM literature emphasizes the economic benefits of
carpooling (and buses) and shows examples of successful programs.
An important aspect of marketing carpool systems to employers would in
most cases be the advantage of receiving free or low-cost matching
service. Employers may find the master lists most matching programs
produce useful for giving advice on transit or for advice to help a
temporary employee, new employee, or stranded employee surmount special
travel problems. If a central agency handles data processing, the em-
ployer should be shown the advantage of not having to design and operate
his own matching system with the attendant costs and problems. This
argument is probably most useful for the medium-size employer who would
need more than manual matching but would find it difficult to undertake
his own computerized system. (On the other hand, employers who can
afford the effort may support it more actively if they contribute enough
work to feel it is "their" program.)
Appealing To Employees
Individual employees may be a more difficult audience for marketing of
carpooling because of the diverse personalities, situations, and prefer-
ences involved. The problem is twofold: (1) the actual benefits vary
among people, and (2) perhaps more important, the perceived benefits
106
-------
vary considerably. Planning for carpool services should thus consider
the idea that no one incentive or marketing approach will suffice for
all potential carpoolers.
Another point to bear in mind is that people will be prepared to raise
numerous objections to carpooling.. There are disadvantages, and these
should be recognized. But the marketing campaign should anticipate the
drawbacks by emphasizing the compensating benefits. (This is part of
"defining the product.")
The employer can make his carpooling system participation very person-
alized. By doing so he makes the entire process more appealing. For
medium-size employers, or those which have only modest numbers of people
applying for carpool matching at any given time, it may be possible to
provide personal contact between the employer's carpool staff and each
applicant employee. This has been the approach in many firms perform-
ing their own matching, such as the five Boston-area employers discus-
sed in Section III.
One of the most universally useful arguments in favor of carpooling is
that of economics-riding together saves money for the commuters. Other
less tangible benefits such as benefits to the environment would prob-
ably persuade fewer people than would the simple savings in travel costs.
Unfortunately perhaps, most people do not perceive the true total cost
33
of operating an automobile. Consequently, the economic benefit of
carpooling is most easily demonstrated for the longer trips. The HUFSAM
literature provides some good examples of cost savings.
In this time of recent (and possibly future) gasoline shortages, another
motivating element may be convenience. Riding in a carpool reduces the
amount of time waiting in queues for gasoline, reduces the anxiety about
running out of fuel and, if gas-stops in rush hour can be eliminated, it
may save on travel time. (A possible arrangement to capitalize on this
107
-------
feature of carpooling is to have spouses fill the gas tanks during off-
peak hours when queues are shorter, thereby totally eliminating the
frantic search for fuel while travelling to and from work.)
Explicitly provided incentives such as preferential parking for carpools
(discussed in Section VII) should be well-publicized. Special contests
and games with one-time incentive prizes have proven to be popular parts
of campaigns to sign up carpoolers (but be wary of duplicate applications),
Several employers contacted during this study emphasized the idea of
showing that the company (or agency) is really behind the carpool sys-
tem, by the use of techniques such as letters from the president re-
questing participation, personal appearances by upper management at
kickoff publicity meetings, publicity techniques such as showing photo-
graphs of the managers arriving in their carpools (be sure they really
do take it seriously and sign up), use of former executive parking for
vanpools or carpools, providing incentives, and similar techniques.
Use of normal communication channels is helpful-carry items in the
company newsletter, on bulletin boards, over public address systems.
One company contacted has a pre-recorded message by a senior officer
which can be heard by any employee by dialing an advertised number on
company telephones. The message explained the status of carpooling and
extolled its virtues.
Employer campaigns should be credible, substantial in extent and dura-
tion, and well planned. An excellent overall approach mentioned several
times by those contacted in this study was to conduct a carpool campaign
in a manner similar to the way United Fund or other major charity drives
are carried out, with team captains with specific responsibilities for
obtaining results, inter-departmental competition for response, adver-
tised results, and generally a highly-organized, fully-supported effort.
108
-------
Another technique suggested by the radio station campaigns is to make
carpooling "fun" through the use of light-hearted advertising, limerick
contests, prizes for signing up your friends, perhaps pretty "carpool-
girls" touring the cafeteria at lunchtime to remind people to return
their questionnaires.
Appealing To The General Public
Obtaining support and participation from the general public (i.e., from
the people who are not reached by the employers' campaigns in a regional
or combination system) is most difficult of all. The isolated commuter
is difficult to appeal to in part because he may not readily identify
with the carpool system promoter. This is one reason the organization
operating the system should be broad-based and well-organized.
Many of the techniques and concepts discussed above under "Appealing To
Employees" will also apply to the general public. (For that matter,
some of the discussion in this sub-section may well apply to employer
campaigns for their employees.)
The exact nature of the promotion is, first of all, not capable of rigid
definition. No system is "best." Second, the system must suit the cir-
cumstances. In one city the regional promotion may need to appeal to
over half the region's employees, and these may be employed throughout
the area (Boston, e.g.). In another city the employees not covered by
employer programs may be only a fraction of total employment, and may
be concentrated in a few employment centers such as a small central
business district (Knoxville, e.g.). Marketing must reflect the local
fuel supply situation, the local air quality needs, the local geography.
Only general guides can be given here.
Commitment is important. Repeatedly throughout this study, carpool
system organizers have commented on the need for commitment on the part
109
-------
of the system operators. Some said it has been more work than was
expected. Those organizing and promoting the system must be committed
to a long, substantial effort, not just a momentary flurry of activity.
This is one reason for insuring that a government agency with a long-
term responsibility is at least part of the leadership organization.
Leadership should be both professional and popular. By this it is meant
that the staff running the operation should be capable and effective,
but there should be an externally visible popular leader or leadership
group. The mayor of a city or governor of a state could appoint a prom-
inent civic leader as general chairman, for example, much in the way he
would appoint a task force to tackle a community problem or would appoint
a leader for the United Fund campaign. Again, the United Fund type of
organization is a good model for how a carpool service can achieve popu-
lar support, especially at the beginning. Support from all quarters is
helpful: labor leaders, politicians, environmental groups, community
associations.
Media involvement is essential. In Dallas, for example, the lack of
broadcaster support because of competing carpool systems was reported
25
to have hindered development of the campaign. There is a valid issue
here of how to best arrange for good media coverage, however. Some
media people advocate allowing all media (e.g., all radio stations) to
simultaneously promote one carpool system. Others, however, contend
this is counterprcdurti ^. It is argued that if only one broadcaster
(who may actually S'<_ ,\i. ^M, one FM, and one television outlet) has the
rights to the publicity for the campaign he will carry it out more vig-
orously and will be more likely to live up to his commitment. Under
this argument, simultaneous support by all stations would probably
quickly degenerate to a few perfunctory public service announcements
each week, mixed into the programming with all the other public service
functions stations are asked to perform. Certainly the use of a single
broadcaster could add to the flavor of a promotional campaign and the
110
-------
degree of specific identification with it. This issue cannot be re-
solved here but is worthy of further consideration in planning specific
promotions.
Professional advice should always be sought. As was mentioned earlier,
a broad-based committee or board should be the leadership organization.
Before even forming the committee, the rough system plans should be
discussed with community leaders so that even the very beginnings of
the system and its structure are assembled with the advice of the com-
munity leaders who ultimately must support it. People from broadcasting
or newspaper organizations should be asked for their advice on how to
market the system, and perhaps one should be chosen as an honorary vice-
chairman to help design the details of promotion.
Objectives of the promotion include:
1. Inform the public of what is going on,
how to participate, and other technical
aspects of the operation. These should
include information on security and pri-
vacy, maintenance, and responsibility.
2. Demonstrate the benefits of participating,
both for individuals and the community.
3. Persuade people to join up.
The content of the promotion should thus reflect all three of these
elements. It is particularly important to accurately and completely
inform people of the procedures, advise them of their responsibilities,
explain the idea that carpooling is voluntary, and explain the idea
that people can change their mind and either sign up later or not par-
ticipate when he receives his list (i.e., the system is flexible).
Throughout the marketing process it may be well to bear in mind the
inherent limitations of the carpool system's appeal. The Los Angeles
study of attitudes toward carpooling showed that the majority of people
who now drive to work are "hard-core" lone drivers. Strong incentives
or disincentives might attract some of them to carpools, but unwillingly.
Ill
-------
Reportedly such people will not perceive the advantages to them of
carpools. The report suggests that it is most effective to market
carpooling to those who are considered "potential carpoolers." That
study also concluded that, of several motivation techniques available
to encourage carpooling, a carpool matching service would be the most
*
effective for the 24 percent of the sample deemed to be potential car-
poolers. Several psychological factors involved with carpooling are
33
identified in the report. On this thesis, the marketing approach
should emphasize the factors shown to be of interest specifically to
people who are potential carpoolers, such as convenience, economic
savings, and air pollution and fuel conservation benefits, rather than
to try to persuade the "hard core" non-poolers to join. This theory
suggests it is not cost-effective to try to convert the "hard-core
non-pooler."
Administration
Charges - One question is: Who pays the bill? It seems imperative
that individual commuters not be asked to pay the cost of the matching
service, for they would probably be deterred from applying for the
matching service. The FHWA is now providing funds for operating car-
pool systems, and several cities have applied for this aid. Unfortun-
ately, however, this is not additional money but instead is from a
fixed maximum amount of urban traffic improvements funds. Some states
may choose to use all these funds for projects other than carpooling,
thus necessitating other (non-Federal) sources for carpool system
operation.
The private sector may be able to provide financial support. Employers
will probably be willing to contribute effort or funds where large num-
bers of employers are involved so that costs are widely distributed.
The exact results of that study, performed in Los Angeles, may not
apply to areas with different travel habits and attitudes.
112
-------
As noted elsewhere, however, it is likely that government agencies will
play a key role in most regional or combination systems. Consequently,
it is also likely that the government will generally be required to
provide much of the resources needed for operating the system.
Distribution and Collection - An effective system will minimize the
effort required of applicants. Questionnaires should be distributed
through employers wherever possible. Supplementary distribution should
include the places people visit frequently-grocery stores, shopping
centers, and similar locations. Ideally, questionnaires should be
available with just a telephone call, and postage should be prepaid by
the system operator. Minimizing effort will maximize participation.
Replies from the matching service should be prompt. Frequent computer
runs will be rewarded by better public response. (Some systems now
involve a discouraging lag of several weeks from the time questionnaires
are requested until a match list is received.) Distribution through
employers should be encouraged, and firms should consider giving an
employee time on the job to contact fellow employees on his match list.
Questionnaires - Format of the matching questionnaires or application
forms will be determined by the data required. Simplicity is to be
encouraged. Small, readily understood forms would appear to be best,
and would thus be suitable for easily stuffing into pay envelopes or
postcard mailing. Smaller forms will be less expensive to print in
newspapers. One-piece self-addressed forms would add to convenience.
If larger forms are used, an advantage could be the use of one side
for endorsements, explanations, or letters from company management.
Existing systems use forms which run the gamut from simple to complex.
Among the smallest and simplest is that for WBBM in Chicago; among
the most complex is the four-page form for Knoxville. All forms should
be pre-tested in trial surveys (as should the entire marketing approach).
113
-------
An associated concern is a map for those systems which involve the use
of a grid. If applicants themselves identify their own grid cells,
errors will result unless the maps are very clear and logical. This
problem may be more severe in some areas because of the particular
geography involved. Where maps are to be used, extensive pre-testing
is recommended in order to insure the adequacy of the map.
SUMMARY
Carpooling and carpool matching services must both be actively marketed
rather than merely publicized, in order to overcome widespread disinter-
est, objections, and apprehensions about carpools among the commuting
public. The marketing effort should both inform and persuade, and in-
clude thoughtful determinations of what the product is and what the best
promotional techniques are in the circumstances.
Present carpool systems have been marketed in a variety of ways. Re-
gional, broadcaster-operated systems have been characterized by sub-
stantial, highly visible, colorful promotion. Many destination-based
systems have been strongly promoted within the organization served,
even if there has been little externally visible promotion. Combination
systems have involved both internal and external promotion.
It is recommended that promotion be directed towards both the general
public and through employers at employees. It is particularly important
to obtain well-known top leadership and broadly-based support from com-
munity and business groups. The management organization must have a
firm, long-term commitment. Industry and business organizations should
be enlisted to aid in marketing carpooling to employers. Special cam-
paigns for employers should emphasize the advantages to them of promoting
carpooling among employees. Overall operation could effectively be pat-
terned after major United Fund-type charity drives.
114
-------
Marketing to the employees and the general public should emphasize the
personal advantages to the individual of carpooling. As part of this,
it should be clear to the public that employer management and community
leaders support the system.
Community involvement in planning the system and in planning the pro-
motion of the system will be worthwhile, and should involve professional
advice from experienced marketing forces such as broadcasters.
The promotion should recognize that not everyone is a potential car-
pooler and should concentrate on the segment of the commuting public
likely to become carpoolers with adequate motivation. To encourage
participation, effort and cost to the commuter should be minimized.
Promotion and data collection techniques should be designed with local
conditions in mind and pre-tested before widely used.
115
-------
SECTION VII
INCENTIVES AND DISINCENTIVES
GENERAL
This section will discuss incentives for carpooling. In this report,
incentives are actions takent to explicitly encourage carpooling.
Personal benefits of carpooling, such as reduced costs to the commu-
ter, are not explicitly provided and were discussed in Section II.
(These benefits must, however, be emphasized in promotional efforts.)
Other detailed discussions of incentives are listed in the bibliography.
Planned incentives alter one or more of the following characteristics
of travel:
1. Travel cost
2. Travel time
3. Convenience
4. Intangible and nontravel-related factors.
These are the elements which a person either explicitly or implicitly
considers in making his travel decision. In a multi-modal situation,
the communter uses these considerations to decide whether to drive,
ride the train, take a bus, or whatever. Many commuters do not have
a true mode choice, however. Such people more or less must travel by
car because of where their home or job is located or because of their
work schedule. One difficulty with trying to encourage carpooling is
that many auto commuters are in a well-established habit of driving
alone, and may never have seriously considered forming or joining a
carpool. The marketing program described in Section VI is intended
116
-------
to bring the carpooling choice to the commuter's attention and in-
fluence his decision; incentives can provide added weight to the argu-
ment in favor of carpooling by providing benefits to the carpooler
above and beyond those inherently involved in carpooling.
The role of incentives in encouraging the use of carpools is at this
point an arguable one. Some experts contend they are essential while
others consider a good matching system is the key to increased car-
pooling. During the course of this study it became clear that car-
pool systems were attracting greatly increased interest because of
the fuel shortages in early 1974, particularly in cities (San Francisco,
e.g.) where shortages were especially acute. This qualitatively shows
the effect of disincentives. Unfortunately, there are only a few
studies which show the quantitative effect of incentives on carpooling.
One such study, Reference 33, has been mentioned earlier in this re-
port. As part of the work for that report, an evaluation was made of
how receptive people were to a number of proposed incentives. Of
those people who were at least somewhat interested in carpools, the
following percentages were stated to be likely to carpool if the pro-
posed incentives were provided.
Percentage of People
Who Would be Likely
Proposed Incentive to Carpool
Carpool parking lots near freeways 70%
Use of company cars for carpools 63%
Gas tax rebates 61%
Free or lower parking fees 60%
Reserved freeway lanes 57%
Preferential (more convenient) parking 57%
Note that this survey was taken in the Los Angeles area, and that
most respondents use low-cost or free parking facilities near their
employment sites at the present time. Also note that these are postu-
lated conditions; people sometimes behave differently from the way
117
-------
they expect to. Also, the survey was taken in the Fall of 1972, be-
fore gasoline became less available and more expensive.
Most actual experience with incentives to carpooling has been with
destination-based carpool systems or with isolated public incentives.
From a review of the literature and from the review of carpool systems
conducted for this report, the most commonly used carpool incentive is
preferential parking. Most such cases involve a restricted supply of
parking, however, so that vehicle occupancy would be expected to be
higher than average even without explicit carpool preference. Several
examples are cited in the literature of average vehicle occupancies up to
two or three times the U.S. average where carpools are given preference.
Other types of incentives have also provided increases in carpooling.
Priority lanes for carpools (and buses) have initiated measurable in-
creases in the number of carpools using Oakland Bay Bridge and Golden
Gate Bridge in San Francisco and Shirley Highway in Arlington, Virginia.
While these examples show that carpooling can increase in the presence
of incentives, it is not clear how much additional carpooling would
occur if major incentives were provided. There may be diminishing
rates of return as larger numbers of people begin to form carpools, and
effectiveness may vary from area to area. In spite of these open ques-
tions, incentives offer the potential for producing higher rates of
carpooling, and should be given serious consideration in planning car-
pool systems. This section will discuss a number of possible incen-
tives.
INCENTIVES FROM THE EMPLOYER
Introduction
Most commuters in many metropolitan areas park at facilities provided
by their employer. In addition, this report has emphasized the role
of employers in promoting carpooling among employees. Consequently,
118
-------
employers have a major potential as a source of incentives for car-
pooling. On the other hand, it may be difficult for an outside agency
to persuade an employer to establish incentives, and more particularly
it may be difficult to achieve any uniformity of incentives among em-
ployers in one metropolitan area. With these limitations in mind,
several employer incentives will be reviewed.
Potential Incentives From Employers
Preferential Parking - Parking specifically reserved for carpools and
conveniently located can encourage carpools if parking is in short
supply or if the carpool parking would substantially reduce the time
or distance between parking lot and building. A possible disadvantage
with this technique is opposition from those who may now have preferen-
tial treatment on the basis of rank or seniority and compromises may
be necessary.
Carpool Loading Areas - Special loading/unloading areas provided for
carpools can be an added convenience, especially where large numbers
of personnel work in one building or where parking lots are large.
Such loading areas could provide nearly as much convenience as prefer-
ential parking locations and may not involve so many objectiions from
people with existing parking privileges.
Financial Bonus - At least a few firms are experimenting with cash
payments to employees who don't drive their cars to work. One firm
pays non-drivers one dollar a day. (This type of incentive should be
coordinated with incentives for mass transit, so that people don't
switch from transit to carpools and thus increase vehicle travel.)
Work Schedules - Carpools may be difficult to form if people find they
are delayed in leaving work by the need to wait for someone who leaves
later or must walk substantially further to reach the parking lot.
Such delays may be especially discouraging if they cause the person
waiting to experience greater traffic congestion in the parking lot or
119
-------
on the street system. So one incentive could be to allow carpoolers
to leave a few minutes early (or arrive a few minutes late) in order
to optimize their schedule. Another aspect of this is to publicize
the importance of letting employees leave on schedule; meetings or
other tasks which run "just a few minutes longer" can rapidly defeat
carpools.
Vanpools/Buspools - This report refers to "carpools" but just as easily
in nearly every case one could substitute the term "carpools/buspools/
vanpools," because all such high-occupancy vehicles could be involved.
Vanpools and buspools are generally more highly organized than carpools
and generally imply more direct financial support from employers, in-
cluding provision of vehicles for employee use. Providing vanpools or
buspools would obviously be a strong incentive to employees because
they need not use their own car to commute. Detailed evaluation of
buspools and vanpools is beyond the scope of this report, however;
the reader is referred to the existing literature (see Bibliography).
Marketing Process - An effective marketing process will itself be an
incentive, because it will be convenient, personal, helpful, and other-
wise conducive to participation in carpooling. (See Section VI.) In
this connection, an effective marketing campaign should reward the em-
ployee with recognition, besides more tangible benefits.
Related Services - An employee will not leave his car at home and ride
in a carpobl if he believes he may need his car at work. To offset
this objection to carpools, employers can provide several incentives:
Company cars - Employers can provide vehicles for between-piant: travel,
unexpected daytime business travel, or for contact with clients. (This
may be less expensive than compensating employees for use of their own
cars.) Note that this incentive was rated very high in the attitude
33
survey in Los Angeles.
120
-------
Food service - Larger employers who provide lunch rooms may be able
to improve the rate of carpooling by improving food service so that
employees are less likely to want to travel elsewhere for lunch.
Mass transit connections - If an employee leaves his car at home but
unexpectedly needs a car during the day (for either business or per-
sonal errands), he may be able to use mass transit instead. If mass
transit service is available but not within walking distance, a
shuttle service could be established between the employer's facility
and the nearest public transportation.
Summary - Several incentives have been summarized here. Some are more
attractive than others, but all should be considered as potential ways
to encourage carpooling at employers. Organizations involved in man-
aging carpool systems (government agencies, in particular) should
themselves provide as many of these incentives as possible to their
own employees to demonstrate them and to help promote them. The next
sub-section will discuss incentives available to the general public.
INCENTIVES FOR THE GENERAL PUBLIC
General
Incentives (and disincentives) provided to the general public either
*
by government agencies or the private sector can equitably influence
all automobile commuters regardless of where they are employed. Further-
more, a broader range of incentives can be provided than employers could
provide and there would be no need to rely upon the generosity or com-
mitment of individual employers.
Preferential Treatment On Highways
General - The use of exclusive lanes, reduced tolls, and other prefer-
ential treatment of high occupancy vehicles on highways have received
a great deal of attention in recent years. Some experience has been
121
-------
accumulated which shows that the number of carpools using a facility
will be increased by such techniques. These techniques are generally
site-specific, however; that is, an exclusive bus/carpool lane on a
freeway will only encourage carpools in the travel corridor served by
that freeway. Furthermore, carpools may be attracted to this corridor
from other facilities, resulting in an increase in vehicular travel
rather than a decrease. Thus, the use of this type of incentive re-
quires detailed transportation planning evaluation on a case-by-case
basis, particularly where increased congestion will be incurred by
non-carpoolers. Nonetheless, these now appear to be viable and use-
ful for encouraging carpooling.
Exclusive Lanes or Roadways - Exclusive lanes or roadways involve
giving priority treatment to high-occupancy vehicles, thereby reducing
travel time. Dramatic travel time savings have been demonstrated in
San Francisco and Washington with such facilities. The benefits of
this technique are greater where new facilities, or at least new lanes,
are involved. Where the exclusive lane must be an existing lane re-
moved from general traffic service, there are disadvantages which re-
quire detailed study. Removing one lane from use on an expressway may
increase congestion in the remaining lanes to the extent that there is
no net gain in travel efficiency or air quality. Safety and cost are
other concerns. Several reports dealing with this topic are listed in
the bibliography. This type of incentive has more potential for long-
run, permanent effect than it does for short-term change" in travel-
Preferential Bypass of Queues - Congestion points may provide a rela-
tively simple opportunity for preferential treatment of carpools. Time-
consuming queues may form at toll booths, certain freeway entrance or
exit ramps, wherever traffic flow metering is installed, at certain
intersections, or other high-traffic locations. In such situations
provision of a bypass for carpools would provide the incentive of re-
duced travel time. This technique has been used in the approaches to
toll facilities on the Oakland Bay Bridge and Golden Gate Bridge in
122
-------
San Francisco, and at a metered freeway entrance ramp on the San Diego,
California, Freeway. Physical arrangements may not permit the use of
this technique in many locations without some new construction (of an
extra ramp, for example). Also, unless such locations are widespread
or carry particularly large amounts of traffic, the net effect on
total vehicle occupancy may be small.
Reduced Tolls - Providing lower tolls for carpools would be a direct
economic advantage for carpools, and could be combined with exclusive
lanes at toll booths for greater effect. This type of incentive is
in use in Philadelphia and San Francisco. Such an incentive could be
low in initial cost but might require additional toll collectors or
additional enforcement officers. Loss of revenue may be a considera-
tion as well. On the other hand, this could be an effective technique
if several toll facilities in a metropolitan area carry a substantial
fraction of present traffic and all utilize this incentive.
Preferential Treatment In Parking
Municipalities that operate parking facilities with substantial amounts
of employee parking could institute reduced parking rates, preferential
locations, reserved space, or guaranteed space for carpools. If other
parking rates are increased to compensate for the lower rate for car-
pools, the net effect should be no loss in revenue and no loss in
transit ridership. Private facilities could perhaps be required to
institute similar policies through municipal licensing regulations.
Transportation Control Plan regulations often contain parking regula-
tions and should also be considered as possible mechanisms for imple-
menting such incentives throughout a metropolitan area. (The legal
aspect of this incentive was not investigated for this report.) This
incentive could effectively be coordinated with either encouraging or
requiring employers to provide preferential parking for carpools, so
that together nearly all employees in the region would be treated
123
-------
equitably. From experience to date, and analyses performed for trans-
portation control plans, parking supply regulations are among the most
effective incentives for carpooling and should be encouraged.
Other Public Incentives
Reduced Taxes - It is conceivable to reduce license, fuel, or vehicle
taxes for carpoolers. Such an incentive could be regressive, however,
and would probably be costly and difficult to regulate. This incen-
tive did appear to be attractive to the potential carpoolers in the
Los Angeles attitudinal survey, however.
Reduced Insurance Rates - Several insurance commissions have recently
taken action to allow reduced automobile insurance rates for drivers
who regularly use carpools. This should be encouraged, but the effec-
tiveness of such a technique is unclear at. this time.
Relationship to Transportation Control Plans
Operating carpool systems could perhaps be offered as an incentive to
employers by arranging for preferential treatment of such employees in
Transportation Control Plans (TCP's). For example, some TCP's involve
reductions in employee parking. Employers could be exempted from some
or all of this requirement in return for operation of a carpool system
plus provision for incentives such as transfer of existing parking
spaces to carpool use. Alternatively, transit fare discounts could be
given to firms who encourage carpooling and operate employer mass
transit incentives.
Virtually any other significant impediment to travel, such as parking
supply reductions, increased parking fees, and other disincentives,
will tend to increase carpooling if the disincentive causes an increase
in cost or loss of convenience which carpooling will ameliorate. From
the standpoint of TCP's, however, these effects are not additive;
124
-------
rather, carpooling is a major mechanism for accommodating the travel
prevented by the disincentive.
SUMMARY
This section has reviewed a number of incentives that can be explicitly
provided in order to encourage the use of carpools (and other high-
occupancy vehicles in most cases). Such incentives should be well
publicized if they are used, along with publicity of the direct per-
sonal benefits of carpooling discussed in Section II.
Incentives can reduce travel cost, reduce travel time, increase con-
venience, or otherwise provide additional reasons for a person to
choose carpools over driving alone. Incentives can help to induce a
person to change his longstanding habit of travelling alone.
Experience shows that incentives can lead to increases in carpooling,
but the duration and extent of such effects are not yet clear. Studies
of commuter attitudes imply that at least a minority of present drivers
can be motivated to join carpools, and that certain incentives would
help to induce carpooling among this minority. Gasoline shortages in
early 1974 showed the potential value of public disincentives for
encouraging carpooling.
Employers can provide significant incentives to carpooling, but it may
be difficult to motivate employers to provide such incentives or to
achieve uniformity of incentives among employers in a metropolitan area.
Several types of incentives from employers are in use and are of poten-
tial value.
Incentives provided for the general public by government agencies in
the private sector could be more equitably distributed than employer-
provided incentives, and a wider range is possible. Preferential treat-
ment of carpools on highways is of proven local value but is site-
specific, is perhaps not significant to a metropolitan area, and re-
125
-------
quires detailed evaluation on a case basis. Preferential treatment in
parking price, supply, or location could be most widely distributed
and would be readily and effectively coordinated with employer parking
incentives and thus is recommended as the best major public incentive
for carpooling. Other public incentives are available and worthy of
experimentation but are of unproven value.
Incentives have the potential for assisting the operation of a carpool
system, especially in initial operation. To be effective, their opera-
tion must be well planned and their existence well publicized as part
of the carpool system marketing effort. Incentives should be provided
to both the general public and to specific employee groups. Transporta-
tion Control Plans should provide incentives for employers to operate
carpool systems.
Table 3, reproduced from Reference 40, summarizes the major types of
incentives and several characteristics of each. Each of these types
of incentives was discussed in this Section.
126
-------
Table 3. SUMMARY OF INCENTIVES
Cost-Related Incentives
Parking Cost
Automobile Ownership Cost
Automobile Operating Cost
Travel Time Incentives
Vehicle Priority Traffic Control
Convenience Incentives
Preferential Parking Space Allocation
Facilities for Carpoolers
Adjustments to Working Hours
Vehicle Restriction
Parking Restriction
Legislation
Intangible. Non-Travel Related and Other
Organizational Incentives
on
r
Potential Sponsor
Government |
Agency
*
*
*
*
*
*
*
*
*
*
#
*
Employer
*
*
*
*
*
*
*
*
*
Citizen
Group
*
*
*
University/ '
College
*
#
*
*
*
*
*
*
*
Other
Organization
#
*
*
#
Cost to Sponsor
m
0
U
"rt
c
Low
High
High
Moderate/
High
Low
Low/
Moderate
Low/
Moderate
Low
Low
Low
Variable
Variable
Continuous
Cost
Low (or net
revenue
Varies, can
break even
Low (or net
revenue
Moderate/
High
Low
Low
Moderate
High
Moderate
Low
Low
Variable
Variable
Implementation
Operational
#
*
*
#
*
*
-
-
Planned
*
*
*
*
-
-
127
-------
SECTION VIII
REVIEW OF OTHER ISSUES
GENERAL
This Section treats several miscellaneous issues related to carpooling
and carpooling systems.
LEGAL ASPECTS
Several legal questions can arise in connection with carpools or car-
pool systems. These questions can be categorized as issues relating
to (1) legal authority, (2) regulations, (3) liability, (4) compensa-
tion, and (5) insurance. Each of these will be briefly reviewed.
Legal Authority
The question of legal authority to impose incentives or disincentives
must be investigated in each locality contemplating such action. The
U.S. Secretary of Transportation apparently has no authority under pres-
ent laws to establish preferential lanes for carpools or to initiate other
incentives. Some local authorities or states have such authority but
others do not. Special legislation may thus be required. Preferential
parking may also require new legislation in some states or municipalities.
A related question is the power to set tolls. Such authority varies
among jurisdictions but, except in special circumstances, tolls cannot
be established on the Federal Interstate highway system so the use of
new tolls as a disincentive is impossible on many major urban highways.
128
-------
Regulations
Carpooling itself appears to be the subject of little or no explicit
government regulation. Carrying passengers for a profit is, however,
subject to either local or Interstate Commerce Commission regulations,
or both. Carpools organized strictly for mutual convenience and cost
sharing, which this report has concentrated upon, should encounter no
41
difficulty with common carrier regulations. Vanpools or other
operations involving either profit to the operator or institutional
sponsorship may be subject to common carrier regulation, and may re-
37
quire special driver's licenses for the drivers (as in Minnesota).
Liability
Carpool system operators should investigate their state motor vehicle
laws to determine the liability of a carpool driver to his passengers.
In a study for FHWA, it was concluded that these guest provisions in
motor vehicle laws vary among the states, but that it is generally
considered that a reciprocal driving arrangement (as in carpools)
makes the guest statutes inapplicable.
The FHWA's study also investigated the question of the liability of
the sponsor of a carpool system. Generally, the type of carpool sys-
tem described here, in which the operator provides no coercion but
only provides matching information, would involve little or no ad-
ditional liability beyond that resulting from workmen's compensation
laws. Greater involvement, such as providing company vehicles for
use by carpoolers, involves added legal responsibility which should
41
be investigated by an employer's legal advisors.
In this connection, it would be prudent to clearly state the sponsor's
responsibility and liability in the literature sent to system appli-
cants. Many carpool questionnaires reviewed for this report incorpo-
rate a "disclaimer of liability" and ask that an applicant sign this
129
-------
statement and thus acknowledge the sponsor's lack of liability. While
such disclaimers may be useful as information tools, the FHWA considers
the disclaimers themselves do not alter the sponsor's liability and
thus it may be unnecessary to ask carpoolers to sign such statements.
Compensation and Taxation
FHWA and HUFSAM have examined the conditions under which a carpool
would involve compensation subject to Federal income tax. According
to FHWA, the Internal Revenue Service position is that carpools do not
alter the taxable income of a carpool participant so long as any com-
pensation (to the driver, e.g.) does not exceed costs. Incentives
provided by employers may, however, be treated as fringe benefits and
37 41
thus be taxable. ' More detailed review should thus be undertaken
where potentially taxable benefits to carpoolers are contemplated.
Insurance
Carpool system operators have found that one of the most frequently
asked questions concerning carpools is the effect on a carpooler's
insurance. Almost universally the answer which is provided is that
the potential carpooler should contact his insurance company to be
sure of his coverage and the effects on his rates. This is the recom-
mended position. Generally, the insurance industry has taken a posi-
tive attitude towards carpooling; as this is being written action
is underway to revise insurance plans and insurance laws to encourage
carpooling and clarify the insurance issue. In the interim, carpool
system operators should urge carpoolers to inform their insurers and
check their coverage, and operators should publicize any local statutes
which affect this issue.
130
-------
Summary
This Section has described the major legal issues which may arise
in connection with carpool systems. Most carpool systems would not
be expected to generate serious legal difficulties. Applicable laws
and regulations vary considerably among locations, however, so car-
pool system organizers should obtain local legal advice regarding
their system.
RELATIONSHIP BETWEEN CARPOOLING AND OTHER TRANSPORTATION ELEMENTS
Planners of carpool systems should anticipate the need to coordinate
their systems with other elements of transportation, notably mass
transit systems and taxi operators. Section V recommended that mass
transit system operators be included in planning and operating car-
pool systems and cited Knoxville as an example of where this is being
effectively done. There are two elements to this coordination:
technical and political.
From a technical viewpoint, carpools can be seen as an adjunct to, or
a part of, a coordinated transportation system. Data being collected
from carpool surveys can be used for revising or expanding bus service,
planning park-and-ride facilities, and other transportation planning
functions. Carpool destination lists should include key transit ter-
minals. In fact, carpool systems can be used to increase transit
ridership,, Many people probably drive all the way to work because
the transit system doesn't reach their home. Some of these people
might be diverted to transit if they could join a carpool destined
for a transit terminal (with preferential parking for transit, per-
haps). Buspools or express bus routes might readily be formed in
areas where the density of demand shown by carpool surveys would war-
rant such service. Carpool information kits mailed with the matching
lists could well follow the example of the WBZ-AIA system in Boston
and include maps of the transit system, bus schedules, and other in-
formation to encourage use of available public transportation.
131
-------
On the political side, carpool system planners should coordinate with
transit and taxi operators so that these elements will not feel the
carpool system is a competitor and contrary to their interests.
Transit and taxi operators should be shown the advantages to them of
increased carpooling, such as increased off-peak travel on transit
and in taxis by the carpool riders who need to make off-peak trips
but don't drive their own cars. Also, transit operators should be
offered the use of the carpool survey data, and be asked for advice
on planning what is, after all, a transportation system.
The problem of coordinating transit and taxi service with carpools
has been explored in greater detail in one of the reports prepared
for FHWA. Rather than duplicate that material here, the reader is
referred to that report, Reference 42.
MONITORING RESULTS
Throughout this study, it has been difficult to quantitatively assess
the impact of carpool systems on the amount of carpooling, vehicle
occupancy, fuel consumption, air pollution, total travel, or traffic
congestion. Yet it is essential in the long run to determine these
impacts in order to understand how effective carpooling is at achiev-
ing the objectives of clean air and efficient transportation, particu-
larly when compared with other techniques in Transportation Control
Plans.
To this end it is suggested that carpool systems include provisions
for measuring the degree of change in travel and in carpool use. The
details of doing this are the subject of another research project now
underway for the U. S. Environmental Protection Agency and thus it
would be impossible to rigidly define here how to undertake this moni-
toring. The following tentative guidelines are, however, offered:
132
-------
Carpool questionnaires should ask the applicant to identify
his current mode of travel to work. To keep the question-
naire simple, this should be just one, simple multiple-
choice question. The survey form should allow the person
to identify how many people are in his carpool if he rides
in one.
2. The matching programs should include a master list output
which can be used to form mailing lists for subsequent
surveys. Security and privacy guarantees should be ob-
served, however,, (Employers could be asked to distribute
such "after" surveys rather than mail them to employees
of participating employers.)
Periodically a mail survey should be performed of (1) ap-
plicants to the matching service and (2) randomly chosen
members of the general public to ascertain their mode of
travel. Analysis of such surveys will show the changes
in vehicle occupancy over time.
4. Transportation Control Plans should require the type of
surveys described here both before and after system
operation as part of any mandated carpool system.
More elaborate monitoring plans may be conceived. Relatively straight-
forward monitoring is suggested so that its cost will not be excessive
and so that the monitoring effort will not become a burden upon the
carpool system applicants. Experience with carpool systems will show
how such monitoring efforts should evolve.
133
-------
SECTION IX
ESTIMATING THE POTENTIAL FOR CARPOOL SYSTEMS
GENERAL
Most of this report has consisted of a practical review of present
and proposed carpool systems. Such systems have evolved on a rather
ad hoc basis, there being little detailed scientific basis for plan-
ning carpool systems. In the process of this study some concepts of
a more theoretical nature have been encountered which are relevant to
the question of increasing automobile occupancy; this Section will
review these concepts and also examine the role of carpools in urban
transportation planning.
CARPOOLS AND TRANSPORTATION PLANNING
One way a transportation planner would view carpool systems is as a
mechanism which alters some of the mathematical relationships he uses
to describe existing traffic and plan future transportation networks.
Three of these relationships which could be altered are automobile
occupancy, mode split, and traffic assignment. Each will be dis-
cussed in brief.
Automobile Occupancy
Automobile occupancy is the element which relates the number of per-
son trips to the number of automobile trips. In the urban transporta-
tion planning process, planners often use a mathematical model which
describes how automobile occupancy varies with trip purpose or by land
134
-------
use. Auto occupancy data used for these models are normally collected
in traffic surveys. Past practice has been that the measured auto oc-
cupancy data are also used for forecasts of future travel. The trend
Q £
has been for average auto occupancy to decrease with time. Carpool
systems are intended to increase the number of people who ride to-
gether and thus to increase average auto occupancy. If such systems
succeed in reversing the trend towards lower auto occupancy, there
are major implications for planning of future transportation systems
or elements. The effect of carpools is particularly important for
transportation facility planning because average auto occupancy is
generally lowest during the peak traffic flows of the day; the size
and location of most urban transportation facilities are determined
by these peak traffic demands.
Transportation planners do not generally consider the number of car-
pools directly; rather, the usual practice is to deal with average
auto occupancy. Furthermore there is no standard definition of whether
a "carpool" is 2, 3, or more persons. Consequently it is necessary
to examine the relationship between the number of carpools in traffic
and the average auto occupancy.
In the study of the Hollywood Freeway referred to earlier, a count was
made of automobile occupancy during the morning peak period. The re-
33
suits of that count were as follows:
Number of Persons Percentage Percentage
in Vehicle of Vehicles of Persons
1
2
3
4
Total
88.6%
10.1
0.9
0.4
100.0%
79.0%
18.0
1.6
1.4
100.0%
Average occupancy = 1.131 persons/car
135
-------
The above data shew that more than 11 percent of the vehicles carry
two or more occupants. But it would be erroneous to conclude that all
vehicles with two or more (or three or more) persons were carpools. A
carpool is generally considered to be an arrangement in which people
regularly ride together for a specific purpose such as the journey to
work. Some of the persons recorded in the above sample are probably only
occasional passengers; their purpose may merely be to accompany another
person in order to obtain the vehicle for use during the day, for example.
In fact, a mail survey of drivers using the Hollywood Freeway showed that
fewer than 7 percent of the respondents consider themselves as belonging
to a carpool. The report comments upon this apparent discrepancy as
follows:
"Among the 1896 respondents in this study, 6 percent or
108 belong to a carpool; 93 percent (1763) do not, and
1 percent, or 25, did not indicate whether or not they
carpool. Auto occupancy samples during morning commut-
ing hours on the freeway indicate that 10 to 12 percent
of the vehicles carry more than one occupant, and know-
ledge of this fact has led to the speculation that the
survey might be biased in that a disproportionately low
number of carpoolers returned the questionnaire. We do
not believe that the bias speculated upon actually ex-
ists. Rather, it is believed that many of those com-
muters who join with others to ride together only occa-
sionally, or under special circumstances, probably do
not consider themselves as 'members of carpools1 and
would, therefore, answer negatively to the carpool mem-
bership question
"Pursuing this rationale, a substantial number of two-
or-more-occupant vehicles would likely appear on the
freeway on any given day, occupied by commuters who do
not think of themselves as regular carpoolers."33
Another way of stating this is that measured auto occupancy provides an
6
overestimate of the amount of organized carpooling. The implication of
this observation is that the amount of increased carpooling must be more
than one would initially expect in order to achieve increases in average
auto occupancy. In the present example, the average auto occupancy was
1.13. If all vehicles with two or more occupants were considered to be
carpools, then doubling the number of carpools would double the number
136
-------
of persons in cars with more than one occupant. In the present example,
average auto occupancy would become 1.30, a substantial increase. For
the same volume of persons, the vehicle volume would be reduced by 13
percent. In contrast, if the number of carpools is considered to have
been indicated by the mail survey, then doubling the number of carpools
would increase the number of sampled drivers in carpools from 6 percent
of the respondents to the mail survey to 12 percent. Average reported
carpool size was 2.3 persons. Doubling the reported rate of carpools
would increase average vehicle occupancy to only 1.20, and would reduce
the number of vehicles by only 6.1 percent or less than half as much as
with the prior assumption.
This exercise illustrates the idea that the amount of carpooling must be
increased by more than would be expected from examination of only vehicle
occupancy data. It also demonstrates the sensitivity of traffic volume
estimates to assumed vehicle occupancy or amounts of carpooling.
Some analytical work has been done by the U.S. Department of Transporta-
tion to examine the potential for altering vehicle occupancy through
greater amounts of carpooling. In this work, national survey data were
used to establish the distribution of occupants among vehicles, and then
to examine the average vehicle occupancy separately for vehicles with
multiple occupants. It was found that cars with more than one person
carry an average total of 2.5 persons, and that such multiple-occupancy
vehicles carry roughly 30 percent of all auto commuters. (This study
separately reached the same conclusion reported earlier in this section
that the national level of carpooling is much less than would be indi-
cated by average auto occupancy.) Calculations were made to determine
the effects on average occupancy of various postulated changes in the
amount of commuters in multiple-occupancy vehicles. One calculation was
to determine how much change it the use of multiple occupancy vehicles
would be needed to increase average auto occupancy by 50 percent. The
report states:
137
-------
"There is frequent discussion of raising average total
occupancy from its present 1.2 to (for example) 1.8.
To achieve such a considerable shift would require an
increase in the proportion of total individuals travel-
ing in multiple occupancy autos from 27.6% to about 73%.
The corresponding rise in the proportion of multiple
occupancy autos is from 13.3% to approximately 53%.
The associated changes required in terms of other vari-
ables certainly suggest that the proposed increase in
average total occupancy would be quite difficult
" consider the hypothetical average occupancy in-
crease to 1.8, which was discussed earlier. Maintain-
ing the average multiple occupancy at the current 2.5,
this results in a 35% reduction to work trip auto
usage, which is significant. However, as mentioned in
the previous section, the occupancy increase (to 1.8)
required to achieve such a reduction is substantial
and possibly unattainable.""
As another case, the DOT report calculates the effect of an average auto
occupancy of 2.5:
"As discussed earlier, the average multiple occupancy
is currently 2.5. As a limit, an ideal case in which
all cars carry an average of 2.5 commuters is examined.
This means that the average total, auto occupancy level
is 2.5 and that the proportion of total travelers in
multiple occupancy autos is 100%, both of which are
quite improbable. Under these highly idealized circum-
stances, the curves indicate a maximum reduction in
auto usage (for commuting) by approximately 48%.
The DOT report demonstrates some of the difficulties associated with
marked increases in auto occupancy. This study also demonstrates that
the greatest number of commuters in multiple-occupancy vehicles are in
vehicles with one other person; it is apparently much more difficult to
form groups of three or four. The relatively large number of people
riding alone in cars and the small number with two, three, or more sug-
gests that "the greatest marginal benefit to society in terms of reduced
traffic congestion and lower numbers of mobile pollution sources comes
with the commuter's decision to join with one other individual to share
the ride to work." This argues against defining carpools as three
people or four people for the purpose of creating incentives.
138
-------
Mode Split
Another part of the transportation planning process is the estimate of
the manner in which person trips are divided ("split") among the various
travel modes (auto, rail, bus, etc.)- Increased carpooling may signi-
ficantly alter modal split. It is conceivable, for example, that people
43
will switch to carpools from transit, not only from other cars. Or
carpool service to transit terminals could result in ridership increase
on some transit lines. Furthermore, the use of carpools may alter the
distribution of mode split, lhat is, carpooling may disproportionately
attract people from outer suburbs versus inner suburbs, from certain in-
come groups, or from certain travel corridors. (This possibility will
be further discussed in the next sub-section.) Another element in mode
split is that people who have a choice between modes often make that
choice based upon some weighing of cost and travel time. If carpooling
reduces highway congestion, the highway routes may become more attractive
to some people who now use transit. This could result in lower transit
ridership or otherwise alter mode split relationships. The relationships
between carpooling and mode split are not well understood and require
further study.
Traffic Assignment (Route Selection)
Once a person has chosen which mode to use, he must next choose his
route. (Actually the route choice and mode choice processes are probably
iterative to some degree.) The planning process which analyzes this
route choice is referred to as traffic assignment; on paper the analyst
"assigns" the forecasted traffic to a route in order to analyze the be-
havior of the facility with the forecasted traffic. As noted earlier,
travelers consider travel time, travel cost, and other factors in choos-
ing among routes.
Increased carpooling could alter route choice and forecasting of traffic
assignments because carpooling could reduce congestion and alter rela-
139
-------
tive travel times for alternative routes; incentives such as altered toll
schedules could revise the cost relationships; and disincentives such as
ramp metering or parking restrictions could alter time, cost, or conven-
ience elements.
Furthermore, there is some evidence that carpooling alters travel patterns
in a more basic way. In an analysis of present carpool habits in Cali-
fornia, the Hollywood Freeway study report showed that people who carpool
differ in their travel patterns from those who do not. Table 4 sum-
marizes some of the differences found to be statistically significant.
Carpoolers travel further; have a greater tendency to work for govern-
ment agencies, large employers, and in high-density areas; and are more
apt to arrive at work during the peak period and have to pay for parking.
This implies that increased carpooling will not draw evenly from the non-
carpooling population but will draw more people from some locations,
travel patterns, and travel periods than others. In short, carpooling
could alter present travel patterns.
Table 4. DIFFERENCES IN TRAVEL BEHAVIOR BETWEEN
CARPOOLERS AND NON-POOLERS ON HOLLYWOOD
FREEWAY, 1972
Characteristic
Average distance to work
Average travel time to work
Employed in high- density areas
Employed in low-density areas
Employer has 25 employees or less
Employer has 100 employees or more
Employer is government agency
Use free parking
Arrive at work in peak period (7-8)
Data (Sample of 1896)
Carpoolers
19 miles
40 minutes
66%
34%
16%
66%
27%
42%
58%
Non-Poolers
16 miles
36 minutes
49%
51%
28%
48%
16%
62%
36%
Source: Reference 33.
140
-------
In another study, a review was made of the characteristics of applicants
to the WBZ-ALA carpool system in Boston. This study showed that the ap-
plicants for the WBZ-ALA. matching service are apparently not representa-
tive of Boston area commuters in general, but rather live further from
their jobs and tend to live further from Boston than do average commu-
44
ters. Another analysis of the WBZ-ALA system showed that applicants
are much more likely to work in the Boston central business district than
14
are Boston area employees on the average. Both of these results imply
that increases in carpooling will alter travel patterns. Because these
analyses apply to very specific conditions, however, it would be diffi-
cult to generalize upon the exact type of travel pattern changes that
would occur. Clearly the effect would depend upon which employers parti-
cipated, what incentives were provided, and other elements of system de-
sign and operation.
CARPOOLS AND AMOUNT OF TRAVEL
General
Carpooling is of interest to EPA chiefly because of its potential for
reducing the amount of vehicle travel, and consequently the amount of air
pollution emitted by motor vehicles, in metropolitan areas. Transporta-
tion Control Plans (TCP's) which include carpool matching systems usually
have claimed no explicit reduction in vehicle-miles of travel (VMT) as a
result of carpool matching systems; rather, the carpool systems have been
analyzed as a mechanism for accommodating the travel eliminated by park-
ing supply reductions or other strategies. Experience to date shows this
to be a rational approach; there simply has been no demonstration of
marked reductions in VMT as a result of carpool systems. On the other
hand, carpooling, technically at least, offers the apparent potential for
drastic reductions in travel over and above the effect of transit systems
or parking restraints. In concept, widespread use of carpooling (meaning,
widespread increases in vehicle occupancy to three, four, or five persons
per automobile) could reduce travel to one-half, one-third, or less of its
141
-------
present amount. But it should be recognized that substantial amounts of
carpooling are probably practical only for travel to and from work. The
work journey is the only travel occurring in sufficient quantities in
place or time to allow efficient, regular riding together. Travel to
work accounts for only about one-third of total travel by automobiles in
typical U.S. urban areas. Thus, there is a very real limitation on the
amount of VMT reduction which carpooling can achieve.
Number of Carpools and Data Base
Some work has been done to examine the amount of VMT reduction potential
of carpool systems. The DOT paper discussed above included the example
of increasing carpooling to the point that all commuters would ride in
vehicles with an average auto occupancy of 2.5, thereby reducing the
average commuting travel by 48 percent. As commuting travel is only-
one- third of total metropolitan area VMT, the maximum effect of such a
change in occupancy would be a reduction of approximately 16 percent.
To achieve such increased amounts of carpooling with a carpool matching
system would require a substantial data base. As an example, consider a
hypothetical case in which the goal is to reduce total VMT by 10 percent.
This implies reducing commuter VMT by 30 percent. If multiple-occupant
vehicles maintain an average of 2.5 persons per auto, and thus new car-
pools are formed only among present drivers of single-occupant vehicles,
then it can be shown that 50 percent of present drivers must form carpools
leaving 25 percent of present drivers still driving alone. If the chief
source of carpool formation is a matching system, then the matching sys-
tem must include at least 50 percent of all present drivers to reduce VMT
by 10 percent through carpooling. This 50 percent figure in the Boston
area would amount to more than three hundred thousand drivers, or several
times the number of applicants to any present carpool system.
142
-------
Matching Rate
A related issue is the rate of matching applicants. To form a given num-
ber of carpools will require many extra applicants to the matching system
because no system can achieve 100 percent matching. Factors which influ-
ence the rate of matching include the size and number of zones or cells
into which the metropolitan area is divided, the number of work schedules,
and the number of applicants.
Number of Applicants - Given a system of matching and a distribution of
origins, destinations, and schedules, it is apparent that the probability
of finding a match will increase as the size of the data base increases.
The exact nature of this relationship is not clear, however. One writer
has suggested that the probability function for making a match is S-
shaped, so that the matching rate is initially low but then begins to in-
crease rapidly as the number of applicants increases beyond a certain
45
critical value. If true, this would explain the relatively low initial
matching rates found with some systems (such as WBZ's 25 percent). In
another paper, a computer simulation process was used to predict the
change in matching rate as the number of applicants increased. The simu-
lation indicated that there is no critical point for the matching rate,
but that matching rate will increase at a diminishing rate as the number
of applicants increases. The simulation was for the WBZ-ALA system, and
predicted that 43,400 applicants would be required for 50-percent match-
ing. Other programs have shown high matching rates with fewer data.
Variety of Origins, Destinations, and Schedules - With a given number of
applicants to a carpool matching service, the matching rate will decrease
as the number of combinations of origins, destinations, schedules, and
preferences is increased. For example, the simulation study referred to
above predicts that approximately twice as many applicants would be needed
to achieve a 50 percent matching rate with seven work schedules as with
44
five work schedules. One of the DOT papers referred to earlier developed
143
-------
a probability method for calculating the size of origin or destination
cells based upon the desired matching rate. Fewer combinations produces
a higher matching rate, but of course some of the resulting matches may
be of little utility. Larger origin zones or destination zones would
increase matching rates but the zones would be so large that two people
in one zone may be too far apart for practical carpooling.
One implication of this has already been observed: matching rates are
inherently higher in destination-based systems, where the number of com-
binations is much smaller than with regional systems. For a given data
base, and probably for a given amount of effort, higher matching rates
can be expected with destination-based matching systems than with region-
al systems. Fairly high matching rates have been attained with data
bases of only a few thousand with some destination-based systems.
Number of Carpools From Matches
An unknown question in most cases reviewed for this study is how many
carpools are formed from the proposed matches provided to applicants for
carpool matching service. There is no known correlation between carpool
matches and carpool formation. The first attempt to discover this cor-
relation appears to be a survey planned for this spring in Boston. The
U.S. Department of Transportation intends to mail a detailed follow-up
questionnaire to each person matched by the WBZ-ALA Commuter Computer
campaign. This study will include questions on travel mode (including
carpools) both before and after the person applied for matching. The
DOT survey promises to be very important as it will be the first measure-
ment of how effective a regional carpool system is at actually aiding
people to form carpools.
CARPOOLS AND AIR QUALITY
Strategies to reduce emissions of air pollutants from passenger cars can
do so in two ways. Reducing the amount of vehicle travel (VMT) directly
144
-------
eliminates emissions. Improving traffic flow reduces the rate of emis-
sions per vehicle-mile of travel. Carpooling can produce both types of
effects.
VMT Reduction Effects
The formation of carpools directly eliminates vehicle trips and thus can
be expected to concomitantly eliminate the air pollutant emissions asso-
ciated with those trips. If enough carpools are formed that, say, 3 per-
cent fewer people drive to work, this would initially appear to eliminate
(roughly) 1 percent of total trips and thus 1 percent of all automobile
emissions in an urban area.
There are two elements which may alter this expected amount of emissions
reduction, however.
First, there is the potential for changing the mode split. Some of the
incentives for carpooling, such as parking supply restrictions, may in-
duce more transit travel and this could reduce emissions by more than
expected. On the other hand, some people may shift from transit to car-
pools, thereby resulting in less emissions reduction. The latter possi-
bility seems more likely than the former, in view of the well-established
propensity for Americans to prefer automobile travel to mass transit.
Second, carpooling could result in an increase of non-work travel owing
to the greater number of vehicles available at home for people to use.
The amount of this counter-productive travel will depend upon many ele-
ments and cannot be predicted. It would be hoped that, in the long run,
carpooling would result in a lower specific rate of automobile ownership
so that the amount of extra non-work travel would gradually be reduced.
Prediction of the air quality impact of carpooling should thus include
an awareness of these two effects.
145
-------
Congestion Effects
Substantial carpooling would reduce the number of vehicles using partic-
ular highway facilities in any unit of time and could therefore reduce
congestion, increase speed, and reduce emissions per vehicle-mile. It
is likely that these effects, if they occur, will not be distributed
evenly in a metropolitan area. Rather, one would expect the effect to
be concentrated near major employment centers where the concentration of
employment would both encourage and facilitate carpooling. This points
to an advantage of carpooling programs being used where congestion is
worst; under such conditions, carpooling could reduce emissions by more
than VMT. That is, a 1-percent reduction in VMT would result in more
than 1 percent reduction in emissions, because oE the combined effect of
less congestion and less travel.
This benefit to traffic flow may be offset, however, by traffic induced
or diverted from other routes by the reduced congestion. (In this re-
spect, carpooling is no different from other measures to improve traffic
flow.)
ENERGY ASPECTS
Carpooling, to the extent that it reduces vehicle travel, will aid in
current efforts to consume less fuel for transportation, without any
loss in mobility. It has been estimated that "Increasing the auto oc-
cupancy 30 percent at rush hours can save 5 percent of the total highway
fuel consumed and 3 percent of the nation's total transportation fuel
needs " In comparison, "Approximately 3.5 percent of highway fuel
would be saved by a shift of 10 percent of the auto mileage to buses.
This would require a 100 percent increase in bus patronage " In
fact, "Already more passengers are carried in carpools that (in) all
46
other forms of mass transportation combined."
146
-------
In another sense, carpools can be seen as a means for coping with future
increases in person travel volumes without any increases in vehicle tra-
vel or fuel consumption, if the empty seats in passenger cars are seen
as unused transportation syst-em capacity. As population and employment
grow, carpooling could conceivably absorb all or nearly all of the in-
creased work-related movements of people with no additional fuel consump-
tion (or construction of transportation facilities). Automobiles can
thus be considered as an under-utilized transportation resource which
carpooling would more efficiently use.
147
-------
SECTION X
SUMMARY AND RECOMMENDATIONS
SUMMARY
This report presents the findings of a study of systems designed to en-
courage the use of carpools for the journey to work. The initial intent
of this report was to evaluate several existing systems which are attempt-
ing to encourage carpooling in certain metropolitan regions in the U.S.
by providing a computer service for matching prospective carpool part-
ners. During the study it became apparent that experience with other
kinds of carpool systems, notably those intended for one destination or
one employer, would also be worthwhile to review because experience with
such systems could be helpful in planning and evaluating systems for
region-wide use. Several region-wide and several single-destination car-
poll systems were then critically studied.
In general it was found that carpooling is in an extremely dynamic state.
Systems are being planned or put into initial operation throughout the
U.S., and existing systems are rapidly evolving as experience accumu-
lates. Few data were found to be available by which effectiveness of
these systems could be judged. Consequently, most of the conclusions
of this report must be considered tentative. In some cases, even tenta-
tive conclusions were not yet possible because of the lack of sufficient
experience.
Section I of this report provided background information on carpool sys-
tems. The present interest of the U.S. Environmental Protection Agency
148
-------
(EPA) stems largely from plans to use carpooling as one of the strategies
for reducing air pollution in many U.S. metropolitan areas. In addition,
recent shortages of gasoline in some parts of the U.S. have led to in-
creased interest in carpooling as a means to conserve fuel. Congress
has recently asked EPA to work with the U.S. Department of Transporta-
tion to determine what measures might lead to significant increases in
carpooling.
Substantial activity is underway throughout the U.S. to organize and
operate systems to encourage the use of carpools. A number of these have
come about as a result of requirements in Transportation Control Plans
to provide a carpool matching service. Most of these plans do not in-
clude detailed requirements for carpool systems; thus, this report is
addressed to the need for development of detailed guidelines for organ-
izing carpool systems.
Section II summarized the benefits of the widespread use of carpools.
Both public and private (personal) benefits are involved.
Chief among the expected public benefits of widespread carpooling would
be improvements in air quality. Greater use of carpools would improve
air quality in two ways. One, the greater use of carpools would reduce
the number of vehicles being used, thus directly reducing the amount of
pollutant emissions- Second, with fewer vehicles on the road there
would be less congestion and higher average speeds. Automobiles emit
less pollution per vehicle-mile at higher average speeds, thus further
reducing total vehicular emissions. Carpooling is more appealing and
less costly than many other alternative transportation controls which
could reduce air pollution.
Other advantages of increased carpooling include reduced energy consump-
tion, monetary savings for the commuter, and the opportunity to take
advantage of incentives provided for carpoolers.
149
-------
Summary Evaluation of Garpool Systems
Sections III through VIII of this report have reviewed the experience
with several current carpool systems in the U.S. One section has been
devoted to each of several aspects of planning and operating such systems,
namely (III) Matching Method and Data Base, (IV) Matching Software, (V)
Organization, (VI) Marketing, (VII) Incentives and Disincentives, and
(VIII) Other Issues. Content and recommendations of each section will be
summarized here.
Each of these six sections included general discussion of the topic in ques-
tion and specific review of actual experience with present carpool sys-
tems. Attention was focussed chiefly on the applicability of the informa-
tion to carpool systems which could serve an entire metropolitan area.
Section III discussed basic carpool matching methods and data bases.
Three basic organizational plans for carpool systems were defined:
1. Destination-based systems serve employees who
work for one or a small number of employers at
one location.
2. Regional systems serve an entire metropolitan
area.
3. Combination systems combine the features of
destination-based and regional systems in order
to capitalize on the simplicity of destination-
based systems while serving regional needs.
Destination-based carpool systems have been in use for several years.
Many older systems use manual methods for matching prospective carpool
partners, whereas latest systems use computer programs for matching.
Most experience is with destination-based systems operated by employers
to facilitate the allocation of limited parking supplies. Although the
focus of this study has been regional carpool systems, some destination-
based carpool systems were reviewed in this report because the experience
with these systems is valuable for planning regional (or combination)
systems.
150
-------
These destination-based carpool systems have achieved very high vehicle
occupancy compared to the national average. At Federal Highway Adminis-
tration (FHWA) headquarters in Washington, D.C., an average of 2.45 per-
sons per auto was achieved with the aid of computer matching services;
higher vehicle occupancies have also been achieved in some locations.
At a Burroughs Corp. plant in Pasadena, California, computerized carpool
matching resulted in a 35 percent decrease in parking space demand.
Special reviews were made of five employer-operated Boston-area carpool
systems. These five systems were investigated to determine what alterna-
tives for carpool matching were utilized by employers who choose not to
rely upon the computerized regional carpool system being operated by
WBZ and the ALA Auto and Travel Club. All five of these employer sys-
tems involve manual or semi-automated matching systems even though up to
several thousand employees are involved. All five appear to offer flex-
ible, personal, adequate carpool matching service, but it is not possible
to conclude whether these systems would be adequate if there were a
greater demand for the matching service.
Experience with destination-based matching systems shows they can be ef-
fective if combined with suitable promotion and incentives. (Promotion
was discussed in detail in Section VI and incentives in Section VII,
both summarized below.)
Region-wide computerized matching of carpool candidates is a relatively
new concept in carpools. The potential advantage of regional matching
is that it can provide more people with potential carpool partners than
destination-based systems usually can, which is particularly useful for
employees of small firms and people who live unusually far from their
place of work. This report reviews the case histories of several cur-
rent regional carpool systems, all operated by or in conjunction with
broadcasting stations as a public service.
WBZ radio and television and the ALA Auto and Travel Club in Boston oper-
ate what is probably the oldest and most well-known regional carpool sys-
tem. This system has been heavily promoted, and all costs and effort
151
-------
has been donated by the private sector (largely by WBZ and ALA). By
late March 1974, the system had provided computer matches to over 2800
people (out of 11,141 applicants). This response has been less than was
anticipated but applications continue to arrive and the campaign will
continue through at least August 1974.
Several other regional carpool systems operated by broadcasting stations
were studied for this report, but all have had substantially less re-
sponse than WBZ in Boston. The only regional system reviewed that has
had a higher response has been a system operated by a coalition of
government and private groups in the San Francisco Bay area. In a short
time the San Francisco system has attracted about 25,000 applicants of
which 15,000 had been processed by April first, with a high matching rate.
The regional systems reviewed for this report have provided matching ser-
vice to only a small fraction of the metropolitan commuter population
they are intended to serve, in spite of extensive marketing. There are
no data on how many carpools have been formed as a result of these car-
pool systems.
Combination matching systems typically involve serving both employers
and the general public, with either single or multiple data bases. These
are the newest type of carpool systems, and most of those reviewed for
this report have been providing matching service only to employers so
far, with expansion to the general public planned for the future.
The Denver, Colorado matching system is at present being operated by the
private sector, but plans are underway for a government agency to operate
it. Approximately 10,000 persons from 20 firms have been serviced so
far.
In Connecticut, the State Department of Transportation is providing com-
puter matching services for carpool systems in several cities through-
out the state. Local sponsors vary among the cities. Approximately
152
-------
50,000 employees of 20 employers had received matching services by
January 1974.
The carpool matching system in Knoxville, Tennessee, is being operated
with the philosophy that carpools are part of the transit system and car-
pool survey data can be used to dynamically plan transit service. Ap-
proximately 20,000 carpool survey forms have been processed, and some
changes have already been made in bus service.
In several other cities and states, system operation is only just be-
ginning.
Combination systems reviewed for this report show great promise as mech-
anisms for both encouraging carpooling and providing the matching ser-
vice needed to facilitate the formation of carpools. The most success-
ful combination systems are characterized by enthusiastic support from
a broad-based coalition of government, business leaders, and major em-
ployers.
Special purpose carpool systems for non-work trips were briefly reviewed,
but appear to be attracting little interest and cannot be considered to
be significant at this time.
Of the carpool systems reviewed for this report, combination systems
appear to have the greatest potential for substantially increasing the
auto occupancy in metropolitan areas. Within the limits of local con-
ditions, it is recommended that a dual approach be taken to matching.
Effort should be directed first at the major employers, and then the
employer data should be combined (where possible) to form a data base
for region-wide matching. Overall system operation should be managed
by a team with broad public support.
Section IV discussed the computer programs that perform the actual match-
ing process to identify potential carpool partners. Discussion included
experience with several programs now in use.
153
-------
Several types of methods for coding and matching origins (home addresses)
and destinations (work locations) were reviewed. Some coding systems de-
fine zones, or grid cells, into which a metropolitan area is divided.
Others us x-y coordinates or specific landmark locations. Some systems
match only people with exactly the same origin and destination designa-
tion, others also search adjacent locations for additional matches.,
Other criteria for matching include work schedules and personal prefer-
ences. It was concluded that the number of matching criteria should be
minimized in order to prevent low matching rates, but no rigid rules can
be stated. The better matching programs were found to produce reason-
able matching rates with only a few thousand employees, in destination-
based systems.
Privacy and security of personal data are important considerations in the
matching process, so it is recommended that matching programs should pro-
vide the minimum necessary amount of output data to other applicants;
home addresses, in particular, should not be disclosed.
An essential consideration is that the carpool data be readily revised,
corrected, or updated to reflect such factors as changes in home or work
addresses. Several methods to do this were reviewed.
Three widely-used computer programs were examined. The Federal Highway
Administration (FHWA) program is probably the most widely distributed,
is well documented, and is fairly effective. It appears to be somewhat
limited in its usefulness for region-wide matching, however, because of
its grid system, large amounts of output, and lack of efficiency for
multiple-destination service.
Another widely-disseminated government-written program is that from the
U.S. Census Bureau. This program is designed for use with computerized
address location files, uses zone matching rather than a grid, and has
154
-------
been documented. Recent changes to allow its application to multiple
destinations have not been documented, however.
The third program, by Burroughs Corp., uses a combination of grid cells
and x-y coordinates for matching and can match for up to 99 destinations.
A useful feature is a vector search technique for locating potential
partners nearest the direction of travel from home to work.
Some new and less well-known programs were also studied. Of these, the
most promising appears to be one written for the Denver system and this
report described its operation in detail. The matching process incor-
porates a vector search more sophisticated than other search methods; in
this mode, the computer determines the eight potential carpool partners
who would involve the least extra travel distance for the applicant
being matched. The program has been adapted to 20 or more firms, al-
though it has not been documented, and is suitable for either destination-
based or regional matching.
From the review of computer programs, several recommendations were made
for choosing a program. For region-wide matching, the best programs
appear to be the Denver, Burroughs, and Census Bureau programs, but all
either have limitations or are unproven in regional use. In any case,
program choice will depend upon the particular circumstances and require-
ments. (Sources of programs are identified in Section XII.)
Section V analyzed carpool system organizations. Three types of organi-
zations have operated carpool systems: broadcasters, other elements of
the private sector, and government agencies.
Broadcasters appear to be extremely useful in marketing carpooling and
carpool matching systems, but it is questionable whether a broadcaster
can enlist enough public support on its own. Employer support is im-
portant to attracting a good data base.
155
-------
This report covered only one regional or combination system operated
exclusively by the private sector and involving no broadcaster, namely
the Denver system. While it has been effective so far, the region-wide
matching is not yet in operation and plans are underway to turn the
major role over to a government agency.
Government agencies operate or coordinate the systems in Knoxville,
Dallas-Fort Worth, Connecticut, California, and Washington, D.C. The
chief characteristic of these organizations seems to be ad hoc coordi-
nation of the private and public sectors, with the government carrying
the long-term commitment for system operation. Government involvement
in carpool systems is becoming more prevalent, especially because of
Federal funding.
Of the systems reviewed, the most effective ones have had the strong
support of employers and civic leaders. It was concluded that regional
(or combination) matching systems are likely to be most successful if
the system utilizes a dual approach, to both employers and the general
public, to maximize the number of applicants in matching systems. It
appears that a government agency should administer or oversee a region-
wide system, but with broad community support. Overal organization
could well be patterned after major charity drives, such as United Fund.
Total costs to initially provide carpool matching services for a
metropolitan area can be estimated to be 10 to 50 cents per applicant,
exclusive of maintenance, incentives and promotion.
Section VI analyzed techniques for marketing carpooling and carpool
matching services. Carpooling must be actively marketed because people
must be encouraged and persuaded to change their habits and form car-
pools.
156
-------
Existing regional carpool matching systems have had varying amounts of
marketing. Broadcaster-supported systems have been characterized by
substantial promotional efforts, which appear to have been effective at
making the public aware of the system. Low response rate from the public
may result from lack of support from employers or lack of incentives.
Destination-based systems, and the employer portions of combination sys-
tems, have often been heavily marketed within the employer organization.
Support of top management, coupled with widespread publicity, have re-
sulted in response from 50 percent or more of the employees.
It is recommended that both public and internal employer promotion be
used with the recommended dual approach carpool systems. To this end,
the report presented suggestions for marketing the system to employers,
employees, and the general public.
Major employers and other civic leaders from the private sector should
be involved with the carpool system beginning in the planning stage.
Support should be solicited from industry, commerce, and labor, and es-
pecially from membership organizations representing these elements. Em-
ployers should be shown the benefits to them of carpooling among their
employees, such as reduced facility costs and improved public relations.
Employees should be approached through a comprehensive, personalized
marketing effort fully supported by their company management. Benefits
to them of carpooling, and special incentives available, should be well
publicized. Overall, the carpool system could be marketed in a manner
similar to that used for major in-house charity drives or other employer
activities.
The marketing plan for the general public must be tailored to the local
circumstances. The leadership is particularly important, as it should
be professional, popular, and committed. Media involvement is essential,
and consideration should be given to allowing a single broadcaster or-
ganization to exclusively provide the broadcasting portion of promotion.
157
-------
Before marketing is begun, local professionals should be invited to help
develop the marketing plan (and the overall system).
Promotion should inform the public, demonstrate the benefits of partici-
pation, and persuade people to join. Providing questionnaires and col-
lecting data should be free of cost to the participants and questionnaires
should be simple to complete.
The entire marketing program, especially survey forms and associated
maps, should be tested before widespread use, to ensure their adequacy.
Section VII reviewed incentives and disincentives which can be explicit-
ly provided to encourage carpooling. With the desination-based carpool
systems in use for some time, the most commonly used incentive has
been preferential parking. Few regional or combination systems have
incorporated any explicit incentives. Limited experience to date shows
that incentives can increase carpooling and should be given serious con-
sideration in planning ccrpool systems, but few data are available to
indicate the effectiveness of specific incentives. The utility of
incentives was demonstrated, however, by the markedly increased interest
in carpooling during the gasoline shortages of January to March, 1974.
A wide variety of incentives can be provided, so it is essential to
have a cohesive plan for coordination among employers and with public
agencies.
Employers have an important potential as a source of incentives. Avail-
able major incentives include preferential parking, loading areas,
direct monetary subsidy, flexible work schedules, and provision of ve-
hicles.
Incentives for the general public encompass a broader range and in-
fluence many more drivers than can employer-provided incentives. Public
incentives generally require detailed case-by-case analysis, however,
because of their potential widespread effects.
158
-------
Preferential treatment of carpools on highway facilities has been shown
to increase the number of carpools using a facility. Techniques include
exclusive lanes, reduced tolls, and preferential bypass of queues at
congestion points. These reduce the cost or travel time for carpools
compared to single-occupant vehicles. Detailed technical analysis is
necessary to insure that such devices do not produce counter-productive
effects on non-carpool traffic.
Preferential treatment in parking supply, cost, or availability should
effectively be coordinated with employers' preferential parking for car-
poolers so that together nearly all employees in a region would be
treated equitably. Parking regulation appears to have the greatest
potential for widespread use as an incentive to carpooling of the
incentives studied.
Incentives have the potential for assisting the operation of a carpool
system, especially in the beginning. Incentives should be well planned
and well publicized; and provided to the general public, employees, and
employers.
Section VIII discussed several miscellaneous issues related to carpooling.
Most carpool systems would not be expected to encounter serious legal
difficulties, but carpool systems should be reviewed for several legal
issues to ensure compliance with local laws and regulations. Carpool
system operators should urge carpoolers to inform their auto insurers
and to verify the adequacy of their insurance protection for carpools.
Insurers are moving towards lower rates for carpoolers and generally
appear to support the carpooling concept.
Planners of carpool systems should anticipate the need to coordinate
their systems with operators of taxis and transit systems. Carpool sur-
veys can provide data useful for dynamically planning transit system
operations. Carpoolers should also be encouraged to use transit systems,
and transit and taxi operators should be shown how increased carpooling
159
-------
can lead to increased use of their services. That this can be done is
demonstrated by Knoxville, where the transit commission is helping to
organize the carpool system.
Planning of carpool systems should include provisions for measuring the
changes in carpool usage. At present, few data are available to cor-
relate carpool use with carpool system operation. Obtaining such data,
through techniques such as mail surveys, would enable design of more
cost-effective carpool systems.
Section IX briefly reviewed several more theoretical aspects of carpool
system planning and explored the relationship of carpool systems to the
transportation planning process.
A review of literature on auto occupancy showed that the amount of car-
pooling is less than one would expect based on only average auto occu-
pancy data. At present, about 28 percent of all commuters travel in
vehicles with two or more people. Only about half of the vehicles with
two or more occupants are actually carpools used for commuting on a
regular basis. Increasing commuting auto occupancy by 50 percent (from
1.2 to 1.8 persons per vehicle) would reduce commuter travel by about
one-third, but to achieve this increase would require increasing the
proportion of commuters who travel in multiple-occupancy vehicles from
28 percent to 73 percent, or to nearly three times the present rate.
Increased carpooling could alter the division of commuters among travel
modes (mode split). It is possible that widespread carpooling ma)^ shift
some commuters away from transit. On the other hand, ridership on some
specific lines could be increased by coordinating carpooling and transit
service. Also, off-peak transit use may increase as carpooling reduces
the number of people without their own cars in employment centers.
Carpooling could also alter travel times on highway routes and thus al-
ter the distribution of traffic throughout the transportation network.
160
-------
There is some evidence that carpooling would alter overall travel pat-
terns because carpoolers differ in their travel characteristics from
non-poolers; they tend to drive further and work for larger firms than
does the "average" driver, for example.
There is only a limited basis at the present time for predicting (1) the
number of applicants to a matching service who will receive useful
matches, (2) the number of persons who would form carpools after receiv-
ing lists of potential carpool partners from a matching service, (3) the
amount of actual travel reduction that would occur as a result of the
carpools formed, and (4) the concomitant improvement in air quality.
Some analytical work on these questions has been reviewed in this report,
but more research and more experience are both needed. Some further re-
search on these topics is underway or planned.
RECOMMENDATIONS
From the study undertaken for this report and summarized in this section,
several major recommendations are offered. (Detailed recommendations on
certain aspects of planning carpool systems are also contained in Sections
III through VIII, and were summarized in the earlier portion of this
Section.)
1. To be effective, carpool systems should be combina-
tion systems with the dual foci of employers and
the general public. Employers should be involved
because experience shows destination-based systems
are much more effective than regional systems. The
general public should be involved because the people
who most need the carpool matching service are those
who work for small employers or live unusually far
from their place of work and thus are least well
served by destination-based systems.
2. Carpool system organization should be tailored
to the community, but as a rule should include
a broad representation of the public and private
sectors. Overall organization could well be
based upon major charity drives such as the United
Fund.
161
-------
3. Carpooling and carpool systems must be actively
marketed in order to alter people's long-standing
habit of driving alone to work. Simple public-
ity will not suffice, and professional market-
ing guidance should be sought for each system.
4. Incentives should be provided to encourage car-
pooling. A wide variety of incentives is avail-
able, and those used should be suitable for local
circumstances and coordinated between the employers
and the public agencies. Parking regulation has
major potential.
5. Additional study and research is needed to under-
stand and plan effective carpool systems, because
little experience with these systems has yet been
documented. A major need is for an understanding
of how many carpools are actually formed with the
various carpool systems. Research should be under-
taken to investigate the carpool formation rate
from several regional or combination carpool sys-
tems.
162
-------
SECTION XI
REFERENCES
1. 42 United States Code 1857 et seq.
2. Public Law 93-239, January 2, 1974.
3. U.S. Department of Transportation. National Transportation Energy
Conservation Action Plan/Executive Summary. (Presented at DOT
Carpool Seminars. February 1974.) Photocopied.
4. Federal Register. 38:30966-30967. (40 C.F.R. 52.1138)
5. The Conservation Foundation. A Citizen's Guide to Clean Air.
The Conservation Foundation, Washington, D.C. January 1972.
6. Kendall, D. and J. Brevard. Auto Occupancy and the National Level
of Carpooling. U.S. Department of Transportation, Transportation
Systems Center, Cambridge, Mass. Report No. RP-SA-13. December 28,
1973.
7. Highway Users Federation for Safety and Mobility. News Release.
August 29, 1973. Photocopied.
8. Transit and Traffic Engineering Branch, Urban Planning Division,
Federal Highway Administration, U.S. Department of Transportation.
Carpooling Case Studies. U.S. Department of Transportation,
Federal Highway Administration, Washington, D.C. January 1974.
9. Alan M. Voorhees and Associates, Inc. Organization for Carpooling.
U.S. Department of Transportation, Washington, D.C. January 1974.
10. Pratsch, L. Carpool and Buspool Matching Guide. Third Edition.
U.S. Department of Transportation, Federal Highway Administration,
Washington, D.C. November 1973.
11. Personal communication. Personnel at Station WBBM, Chicago.
March 18, 1974.
12. Alan M. Voorhees and Associates, Inc. Manual Carpool Matching
Methods. U.S. Department of Transportation, Washington, D.C.
January 1974.
163
-------
13. Station WBZ. WBZ's Commuter Computer Clubcar. Press kit.
Boston, 1974.
14. Heaton, C. Preliminary Evaluation of the Boston Area Carpooling
Program (WBZ/AIA Commuter Computer Clubcar Campaign). U.S. De-
partment of Transportation, Transportation Systems Center, Cam-
bridge, Mass. Report No. RP-SA-10. December 11, 1973.
15. Personal communications. Personnel at WBZ, ALA, Associated
Industries of Massachusetts, Massachusetts Executive Office of
Transportation and Construction, and Federal Highway Administra-
tion Division Office. January, February, and March 1974.
16. Station WIND. The WIND Commuter Computer. Press kit. Chicago,
1974.
17. Personal communications. Personnel at WIND, and ADP of Chicago
and Boston. March 1974.
18. Personal communications. Personnel at KDKA and Southeastern
Pennsylvania Regional Planning Commission;, Pittsburgh. March 1974.
19. Personal communications. Personnel at Keystone Auto Club,
Philadelphia. March 1974.
20. Personal communications. Personnel at FHWA Regional Office and
Hoefer/Amidei Assoc., San Francisco; and California Department
of Transportation, Sacramento. March 1974.
21. Personal communications. Personnel at WJZ-TV, Baltimore.
March 1974.
22. Rocky Mountain AAA. The Care and Feeding of a Carpool. Instruc-
tion kit. Denver, February 1974.
23. Personal communications. Personnel at George Washington High
School; Downtown Denver, Inc.; EPA Region VIII; Air Force Account-
ing and Finance Center; all in Denver. February and March 1974.
24. City of Dallas. Car Pool. Information kit. Dallas, Texas, 1974.
25. Personal communications. Personnel at Dallas Traffic Control
Department. March 1974.
26. Connecticut Department of Transportation. Planning for the
Future. Part III: Expressway Interchange Parking Study in
Connecticut. Connecticut Department of Transportation, Hartford,
Connecticut. November 1970.
164
-------
27. Personal communications. Personnel at Connecticut Department of
Transportation. January, February, and March 1974.
28. Personal communications. Personnel at Metropolitan Washington
Council of Governments. February and March 1974.
29. Alan M. Voorhees and Associates, Inc. Review of Matching Software
and Procedures. U.S. Department of Transportation, Washington,
D.C. January 1974.
30. Personal communications. Personnel at Office of Mayor and University
of Tennessee, Knoxville. February, March, and April 1974.
31. Personal communications. Personnel at California Department of
Transportation, Sacramento. March 1974.
32. California Department of Transportation. Car Pool Program Im-
plementation Plan. First Revision. Department of Transportation,
Sacramento, California. February 15, 1974. Photocopied.
33. Alan M. Voorhees and Associates, Inc., and Behavior Science
Corporation. A Study of Techniques to Increase Commuter Vehicle
Occupancy on the Hollywood Freeway. For: California Department
of Transportation. November 5, 1973.
34. Ups and Downs of Car-Pooling. New Scientist, p. 760-761.
March 21, 1974.
35. Personal communication. Personnel at Star Markets, Boston,
Massachusetts. March 1974.
36. U.S. Department of Transportation, Federal Highway Administration,
Public Transportation Branch. Estimating Auto Occupancy: A Re-
view of Methodology. U.S. Department of Transportation, Federal
Highway Administration, Washington, D.C. 1972.
37. Highway Users Federation for Safety and Mobility. The 'Pool It1
Work Kit. Highway Users Federation for Safety and Mobility,
Washington, D.C. 1974.
38. U.S. Department of Commerce, Census Use Study, and Southern
California Regional Information Study, County of Los Angeles.
CARPOL/an approach to large-scale carpooling using DIME tech-
nology. U.S. Department of Commerce, Bureau of the Census,
Washington, D.C. January 1974.
39. Burroughs Corporation. Operation Energy. Burroughs Corporation,
Detroit. 1973.
165
-------
40. Alan M. Voorhees and Associates, Inc. Incentives to CarpoolLng.
U.S. Department of Transportation, Washington, B.C. January 1974.
41. Alan M. Voorhees and Associates, Inc. Legal and Institutional
Issues of Carpooling. U.S. Department of Transportation, Washing-
ton, D.C. January 1974.
42. Alan M. Voorhees and Associates, Inc. Transit/Taxi Coordination.
U.S. Department of Transportation, Washington, D.C. January 1974.
43. TRW, Inc. Socio-Economic Impacts of the Proposed State Transporta-
tion Control Plans: An Overview. U.S. Environmental Protection
Agency, Durham, N.C. November 1973.
44. Berry, W. L. Regional Carpool Matching Programs: A Simulation
of the Boston Experience. Harvard College, Cambridge, Mass. 1974.
Unpublished paper.
45. Kendall, D. Planning for Carpool Matching, Services. U.S. Depart-
ment of Transportation, Transportation Systems Center, Cambridge,
Mass. Report No. RP-SA-18. February 6, 1974. Photocopied.
46. Pratsch, L. Carpools: The Underutilized Resource. Civil Engin-
eering. 44:49-52. January 1974.
Note: Personal communications include correspondence, telephone con-
versations, and meetings.
166
-------
SECTION XII
ANNOTATED BIBLIOGRAPHY
This section lists and describes the contents of selected literature
of major interest to those responsible for planning and analysis of
carpooling and carpool systems. Sources for key documents are also
listed.
167
-------
Alan M. Voorhees and Associates, Inc. Buspools. In DOT Carpool
Seminar. U.S. Department of Transportation, Washington, D.C.
January 1974. (Source: A.)
Criteria for and organization of buspools. Buspools viewed as
larger version of carpools or vanpools. Good description of
case histories of several buspool programs.
Alan M. Voorhees and Associates, Inc. Incentives to Carpooling. In
DOT Carpool Seminar. U.S. Department of Transportation, Washing-
ton, D.C. January 1974. (Source: A.)
Useful summary of the advantages (and disadvantages) of several
types of incentives. Includes a tabulation of which incentives
are useful for various types of sponsors (government, employer,
etc.), and information on relative costs.
Alan M. Voorhees and Associates, Inc. Legal and Institutional Issues
of Carpooling. In DOT Carpool Seminar. U.S. Department of
Transportation, Washington, D.C. January 1974. (Source: A.)
Questions and conclusions regarding: legal aspects of incentives,
expense sharing, liability, competition; security; compensation;
insurance.
Alan M. Voorhees and Associates, Inc. Manual Carpool Matching Methods.
In DOT Carpool Seminar. U.S. Department of Transportation,
Washington, D.C. January 1974. (Source: A.)
Describes noncomputerized methods for organizing systems to match
people for carpools. Intended for organizations with 100 to 1,000
personnel. Describes the mechanics of several methods in moderate
detail and gives examples of existing manual employer systems.
Includes information on costs, incentives, and problems. Should
be reviewed by any organization undertaking an employer-based car-
pool program, because these manual methods may suffice even for
employers with 1,000 employees or more, depending upon employer
and location.
Alan M. Voorhees and Associates, Inc. Organization for Carpooling.
In DOT Carpool Seminar. U.S. Department of Transportation,
Washington, D.C. January 1974. (Source:: A.)
How to organize a carpool/buspool/vanpool program. Descriptions
of several existing organizational arrangements. Tabulation of
carpool organizations in 25 U.S. cities. Emphasis is on employer-
based programs or on single destination areas, not regional pro-
grams .
168
-------
Alan M. Voorhees and Associates, Inc. Review of Matching Software
and Procedures. In DOT Carpool Seminar. U.S. Department of
Transportation, Washington, D.C. January 1974. (Source: A.)
Description and analysis of nine computer programs for perform-
ing carpool matching, with brief outline of 13 additional pro-
grams. Identified characteristics of each program include com-
puter language, computer core size needed, source of program
and status of documentation, name of person to contact, etc.
Alan M. Voorhees and Associates, Inc. Transit/Taxi Coordination. In
DOT Carpool Seminar. U.S. Department of Transportation, Washing-
ton, D.C. January 1974. (Source: A.)
Discusses coordination of carpool campaigns with transit and taxi
operators to help ensure maximum overall efficiency of the trans-
portation system and to minimize possible problems for transit
and taxi operators from large-scale carpool efforts. Important
reference for government agencies desiring to sponsor carpool
programs because it points out problems and shows how to minimize
them.
Alan M. Voorhees and Associates, Inc. Vanpools. In DOT Carpool Semi-
nar. U.S. Department of Transportation, Washington, D.C.
January 1974. (Source: A.)
Presents discussion of methods for organizing carpool programs
based upon 7- to 15-seat vans instead of sedans. Outlines prob-
lems and solutions. Covers programs in which vans are owned by
employers, riders, individuals, or other agencies. Probably of
greatest value to employers considering vanpools as part of car-
pool program.
Alan M. Voorhees and Associates, Inc., and Behavior Science Corpora-
tion. A Study of Techniques to Increase Commuter Vehicle
Occupancy on the Hollywood Freeway. For California Department
of Transportation, Sacramento, California. November 5, 1973.
Report of a study which used the Hollywood Freeway for research
on commuter behavior, potential for carpooling, and methods for
increasing commuter vehicle occupancy. Included: a useful re-
view of carpool literature; behavioral survey of present Holly-
wood Freeway commuters, including exploration of psychological
differences between carpoolers and nonpoolers; evaluation of 16
techniques for motivating carpooling; recommended program of
action on carpooling by California.
169
-------
Berry, William L. Regional Carpool Matching Programs: A Simulation
of the Boston Experience. Harvard College, Cambridge, Mass.
1974 (unpublished paper).
Analysis of the travel characteristics of the first 6,000 re-
sponses to the WBZ-ALA carpool matching service in Boston. Com-
puter simulation of how the WBZ-ALA matching service would operate
with varying numbers of additional applicants. Discussion of the
effectiveness of such carpool systems. (Author is a research
assistant in the Graduate School of Business Administration;
further research is planned.)
Burroughs Corporation. Operation Energy. Book 1: Management Guide-
lines. Burroughs Corp., Detroit, Mich. 1973. (Source: B.)
Brief but comprehensive manual for organizing a carpool program.
Includes discussion of the computer program (described in detail
in Book 2), Burrough's experience, organization and implementa-
tion of carpool program, incentives, sample questionnaires.
Available free from Burroughs Corp., Burroughs Place, Detroit,
Mich. 48232.
Burroughs Corporation. Operation Energy. Book 2: Program Documenta-
tion. Burroughs Corp., Detroit, Mich. 1973. (Source: B.)
User and program documentation for a computer carpool matching
program, CARPOL, developed by Burroughs. Written in ANSI COBOL
(higher level), reported by USDOT (Reference 29) to require
20,000 bytes of core. Based upon a grid system of map coordinates
for residences, up to 99 coded destinations. Uses a vector
search if specified number of matches is not found in home grid
cells. User can specify number of potential matches, number of
grids to search.
Connecticut Department of Transportation. Bureau of Planning and Re-
search. Division of Transportation Planning. Planning for the
Future; Part III: Expressway Interchange Parking Study in
Connecticut. Connecticut Department of Transportation, Hartford,
Conn. November 1970.
Report of survey of extra-legal, ad hoc parking at highway inter-
changes in Connecticut. Survey showed that the majority (93.6
percent) of such parking was for work trips, most of which (88
percent) were completed by carpool. Discussion of need for and
recommendations on formal provision of interchange parking facil-
ities to encourage carpools.
170
-------
Heaton, Carla. Preliminary Evaluation of the Boston Area Carpooling
Program (WBZ/ALA Commuter Computer Clubcar Campaign). Cambridge,
Mass.: U.S. Department of Transportation, Transportation Sys-
tems Center, Cambridge, Mass. Report No. RP-SA-10. December 11,
1973.
A preliminary, internal research paper with analysis of the first
4 months of operating the WBZ/ALA campaign. Useful examination
of origins, destinations, and work times of first 4,600 respon-
dents to campaign. Also includes extensive bibliography on car-
pooling; many of the references are nontechnical items (newspaper
or general circulation magazine articles, for example).
Highway Users Federation for Safety and Mobility. The "Pool It" Work
Kit. Highway Users Federation, Washington, D.C. 1974. (Source:
c.)
Handbook for planning and operating carpool systems, directed
mainly at employers. Generally less detail than in FHWA report
series. Contents; introduction, overview, types of ride pools
(car, bus, and vanpools are discussed), planning a carpool pro-
gram, motivation (incentives), public information and promotion,
matching (duplicates much of FHWA literature on matching com-
puter programs), legal aspects; lists of Federal Energy Office
Regional Administrators, DOT Secretarial Representative, DOT
Division Engineers, HUFSAM Regional Representatives.
Kendall, Donald. Planning for Carpool Matching Services. U.S. De-
partment of Transportation, Transportation Systems Center,
Cambridge, Mass. Report No. RP-SA-18. February 6, 1974.
A preliminary, internal research paper that discusses the mathe-
matical basis for matching carpool candidates. Includes a model
for estimating the number of matches from a given set of circum-
stances .
Kendall, Donald and Joe Brevard. Auto Occupancy and the National
Level of Carpooling. U.S. Department of Transportation, Trans-
portation Center, Cambridge, Mass. Report No. RP-SA-13. Decem-
ber 28, 1973.
A preliminary internal research paper with analysis of the
national statistics on automobile occupancy and carpooling. Ex-
amines the distribution of commuters by number of vehicle occu-
pants and discusses effectiveness of various postulated changes
in auto occupancy. Concludes that relatively large increases in
amount of formal carpools would be needed to achieve modes in-
creases in average vehicle occupancy.
171
-------
Meyers, Phil and John Walker. The Effects of 'Share a Ride Day" on
Los Angeles Freeways. Traffic Engineering. 72 (11): 35-37,
August 1972.
Description of experiment to increase vehicle occupancy through
public relations campaign. No apparent increase in vehicle
occupancy occurred. Detailed tabulation of vehicle occupancy
data for three freeways.
Morin, Donald A. Commuter bus clubs are showing the 'professionals'
how to meet transit demand. Traffic Engineering. 44 (4): 24-27,
January 1974.
Description of experience with buspools organized by commuters,
and criticism of transit systems' failure to meet demand.
Pratsch, Lew. Carpool and Buspool Matching Guide. 3rd edition. In
DOT Carpool Seminar. U.S. Department of Transportation, Federal
Highway Administration, Washington, D.C. November 1973.
(Source: A or D.)
Overview of how to organize carpool or bus pool campaign. In-
cludes description of successful carpool, vanpool, bus programs.
Describes public information, incentives, matching and mainte-
nance. Includes a short bibliography of publications on car-
pools and buspools. General emphasis is on employer-based car-
pool schemes and the DOT/FHWA matching program in particular.
Pratsch, Lew. Carpools: the underutilized resource. Civil Engineer-
ing. 44: 49-52, January 1974.
Essay on the values of carpools to society and the individual.
Some estimates of the impact of massive carpooling on highway
congestion, air quality, fuel consumption. Useful as an over-
view of the potential advantages of widespread carpool programs.
U.S. Department of Commerce, Census Use Study and Southern California
Regional Information Study, County of Los Angeles. CARPOL/an
approach to large-scale carpooling using DIME technology. U.S.
Department of Commerce, Bureau of the Census, Washington, D.C.
January 1974. ($3.) (Source: D or E.)
Report from Census Use Study. Describes a project to use a com-
puter program to generate lists of carpool candidates for em-
ployees in Los Angeles Civic Center. Program (CARPOL) is written
in FORTRAN, is available to public, and at present is limited to
one destination. Flexible matching criteria (variable geographic
area for matching origins). Report includes detailed descrip-
tion of L.A. experience, including costs and times, plus user
documentation for CARPOL. Designed for use with Census DIME files
172
-------
(information relating street addresses to geographic grid sys-
tem) but can be used with other types of geographic coding of
origins. Report includes description of a method for manual con-
version of topological information (such as boundaries of census
tracts or zip codes) to mathematical coordinates for use with
computer program.
U.S. Department of Transportation. DOT Carpool Seminar. U.S. De-
partment of Transportation, Washington, D.C. 1974. (Source: A)
Ring-bound collection of materials provided to participants in
DOT carpool seminars throughout the U.S. during early 1974.
Includes several booklets separately listed in this bibliography,
plus other materials on preferential treatment of multiple-
occupancy vehicles, legislation, and public relations literature.
U.S. Department of Transportation. National Transportation Energy
Conservation Action Plan/Executive Summary. Paper presented at
DOT Carpool Seminars. February 1974 (photocopied).
Summary of the administrative actions by DOT to reflect the
Emergency Highway Energy Conservation Act. Specifies types of
funds available for energy-reduction transportation projects
(such as carpool matching systems) and outlines requirements
for regional emergency transportation programs. Includes text
of relevant sections of the Act.
U.S. Department of Transportation. Federal Highway Administration.
Computer Services Division. Mission Support Branch. Program
Documentation for the FHWA Carpool Matching Program. In DOT
Carpool Seminar. U.S. Department of Transportation, Washington,
D.C. January 1974. (Source: A.)
Detailed description of the internal operation of the FHWA
matching program. ("Available upon request from the Urban Plan-
ning Division on a limited basis.")
U.S. Department of Transportation. Federal Highway Administration.
Urban Planning Division. Public Transportation Branch. Esti-
mating Auto Occupancy: A Review of Methodology. U.S. Govern-
ment Printing Office, Washington, D.C. 1972 (stock number 5001-
0035). (Source: D.)
How auto occupancy data is used in transportation planning.
Determination of average auto occupancy. Modeling methods.
Good compilation of average auto occupancy used in part studies,
including variations by city, year, trip purpose, land use, dis-
tance, car ownership; but much of data are several years old.
Twenty-five references.
173
-------
U.S. Department of Transportation. Federal Highway Administration.
Urban Planning Division. Transit and Traffic Engineering
Branch. Carpooling Case Studies. In DOT Carpool Seminar. U.S.
Department of Transportation, Federal Highway Administration,
Washington, D.C. January 1974. (Source: A.)
Description and analysis of carpool programs organized at the
Federal Highway Administration Headquarters in Washington, D.C.,
in 1972, and at the Baltimore Federal Executive Board offices
in 1973. A total of 1,200 employees at the Baltimore office.
Report includes sample questionnaires, publicity materials, and
results. (Both cases involved the FHWA computer program.)
U.S. Department of Transportation. Federal Highway Administration.
Urban planning Division. Transit and Traffic Engineering Branch.
User Documentation for the FHWA Carpool Matching Program. In
DOT Carpool Seminar. U.S. Department of Transportation, Federal
Highway Administration, Washington, D.C. January 1974.
(Source: A.)
Guide for operating the FHWA matching program. Describes how to
obtain the program, how to establish grid system, logic of
matching process, input and output, and operating the program.
Probably most widely used program, and has been adapted to
several computers. Requires a maximum core capacity of 122K,
or less without certain optional features; written in ANSI COBOL.
Suitable for regional use in that multiple destinations can be
specified, but some users find not enough geographic detail can
be specified in the grid system.
Zevin, Israel. Carpooling in Connecticut. Connecticut Department of
Transportation, Division of Planning, Traffic Statistics
Section, Hartford, Conn. April 1972.
Results of several surveys to determine the present carpooling
habits of commuters. Analysis of Hartford traffic study to show
passenger car occupancy as a function of commuting distance,
type of employer. Analysis of special counts of vehicle occu-
pancy among several types of employers throughout Connecticut,
including data on effect of population and other factors on
occupancy. Analysis of survey of vehicle occupancy at fringe
parking lots.
174
-------
Sources for obtaining published materials:
Source A (DOT Carpool Seminar reports)
Source B
Source C
Source D
Source E
Available at no charge from either the FHWA
Division offices located in the capital city
of each state, or from:
Urban Planning Division HHP-26
Federal Highway Administration
Washington, D.C. 20590
T.S. Grier, Coordinator
"Operation Energy"
Burroughs Corporation
Burroughs Place
Detroit, Michigan 48232
Highway Users' Federation
1776 Massachusetts Avenue, N.W.
Washington, D.C. 20036
U.S. Government Printing Office
Washington, D.C.
Users ' Service Staff
Data Users Office
Bureau of the Census
Washington, D.C. 20233
175
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-450/3-74-041
2.
3. RECIPIENT'S ACCESSIOf*NO.
4. TITLE AND SUBTITLE
Study and Evaluation of Computer Carpool Programs
in Certain Metropolitan Areas
5. REPORT DATE
April 1974
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
David A. Bryant
8. PERFORMING ORGANIZATION REPORT NO.
GCA-TR-74-8-G
9. PERFORMING ORGANIZATION NAME AND ADDRESS
10. PROGRAM ELEMENT NO.
6CA Corporation
Burlington Road
Bedford, Massachusetts
11. CONTRACT/GRANT NO.
01730
68-02-1337, Task Order No. 3
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental ^Protection Agency
Office of Air Quality Planning and Standards
Strategies and Air Standards Division
Research Tn'anqle Park, North Carolinia 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final - April 1974
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
Study initiated to develop and present technique for initiating computer/carpool
programs.
16. ABSTRACT
The report methodology is adopted in certain metropolitan areas to arouse public
interest and support for carpooling by the use of computer matching to reduce the
total number of vehicles committed to work trips, reduce the demand for scarce
motor fuels, and reduce vehicle miles traveled. All these combined will contri-
bute to the improvement of air quality in congested work areas in metropolitan
centers.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Carpool Systems
Matching Method
Data Base
Incentives - Disincentives
13. DISTRIBUTION STATEMENT
Release Unlimited
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
186
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
176
tl Q P O 1974 747-796 / 368. REGION NO
-------
-------
*Q
c
w
r-1
HH
n
>
H
HH
o
Z
2
O
w
"a
>
i
4X
^ji
O
UJ
-sj
^
I
O
^
02O
mo
-»2O C_
m
.2 *n
.; moo
I -irn
i 2NT1
:f OO
J>
O
Tl
$ ^3
if N) J>
i-j i
J!1 -J r-
«>-
N O
"0 ^
R-
Q- 01
CD 3
2. =T
CD
W CD
Q. ~
Q. _
CD
Q.
CD
n
n CD
CD CD
o
CD QJ
- CD
n ^
<2_ cu
ID
D
I s
o i
T3 n
s ?
2 ?
Z w
-a
r-
O
H O B'
-1 (J( QJ
" 3 ix ~j
a- m
- w e n
23 » =!
N)
-J
-J
2
n
T>
m
O
q
o 5
o
§
1°
S u>
m -<
Z t>
m ~* O
O > m
"> _ -"
m m
O m
Z >
>6
o
n
M
------- |