AN ANALYSIS OF OPPORTUNITIES TO
INCREASE CARBON SEQUESTRATION BY
PLANTING TREES ON TIMBERLAND AND
AGRICULTURAL LAND IN THE
UNITED STATES, 1993-2035
-------
AN ANALYSIS OF OPPORTUNITIES TO
INCREASE CARBON SEQUESTRATION BY
PLANTING TREES ON TEVtBERLAND AND
AGRICULTURAL LAND IN THE
UNITED STATES, 1993-2035
Prepared for.
Steven Winnett
Kenneth Andrasko
William Hohenstein
Climate Change Division
U.S. Environmental Protection Agency
401 M Street, S.W. Mall 3220-H
Washington, DC 20460
(202) 260-6923
Prepared by:
Mac Callaway
Shannon Ragland
RCG/Hagler Bailly
P. O. Drawer O
Boulder, CO 80306-1906
(303)449-5515
-------
CONTENTS
EXECUTIVE SUMMARY ES-1
CHAPTER 1 INTRODUCTION 1-1
CHAPTER 2 SCENARIOS 2-1
2.1 Tree Planting Initiative (TPI) Analyses 2-1
2.1.1 TPI20CS 2-4
2.1.2 TPI20 2-5
2.1.3 TPI70CS 2-5
2.1.4 TPI70 2-5
2.2 Cropland Programs 2-5
2.2.1 CROPBASE 2-6
2.2.2 CROP1RC 2-6
2.2.3 CROP1REC 2-6
2.2.4 CROP5RC 2-7
2.2.5 CROP5EC 2-8
2.2.6 CROP10RC 2-8
2.2.7 CROP10EC 2-8
2.2.8 CROP10 2-9
2.3 Pastureland Planting 2-9
2.3.1 PAS7 2-10
2.3.2 PAS7EC 2-10
2.3.3 PAS15 2-10
2.3.4 PAS15EC 2-11
2.3.5 PAS25 2-11
2.3.6 PAS25EC 2-11
2.3.7 PAS25ECM 2-11
2.4 Bottomland Hardwood Planting 2-11
2.4.1 WETS 2-12
2.4.2 WET5U 2-12
2.4.3 WET10 2-13
2.4.4 WET10U 2-13
CHAPTER 3 DATA AND METHODS 3-1
3.1 Data 3-1
3.1.1 Treatment Cost Data 3-1
3.1.2 Timber and Carbon Yield Data 3-2
3.1.3 Eligible Acreage 3-2
RCG/Hagler Bailly
-------
CONTENTS » page ii
3.2 Methods 3-3
3.2.1 Determine Enrollment Distribution/Schedule During the
Planting Period 3-3
3.2.2 Calculate Carbon Sequestration in Each Year of the
Program 3-4
3.2.3 Cost Calculation 3-5
CHAPTER 4 RESULTS 4-1
4.1 TPI Results 4-1
4.2 Cropland Results 4-5
4.3 Pastureland Results 4-12
4.4 Bottomland Hardwood Results 4-16
CHAPTERS SUMMARY AND MAJOR CONCLUSIONS 5-1
5.1 Summary 5-1
5.2 Conclusions 5-1
5.2.1 Regional Shares 5-2
5.2.2 Annual Carbon Increments 5-3
5.2.3 Increasing Unit Costs 5-3
5.2.4 Enrollment Timing 5-4
5.2.5 Funding Mechanism 5-5
CHAPTER 6 REFERENCES «-l
RCG/Hagler Bailly
-------
EXECUTIVE SUMMARY
This study explores twenty-two scenarios for planting trees on private timberland and
agricultural land in the United States for the purpose of sequestering carbon. The study
estimates carbon and cost results for each scenario and draws several conclusions based on
these results. Traditionally, afforestation and reforestation programs in the U.S. have not
specifically targeted carbon sequestration as an objective. For example, the Conservation
Reserve Program (CRP) is geared toward reducing soil erosion, and while tree planting is a
component of this program, the same incentives and programmatic rules and guidelines that
lead to optimal soil stabilization may not be consistent with carbon sequestration. While these
objectives are clearly compatible with carbon sequestration, they may affect program
efficiency with respect to carbon sequestration.
The purpose of this study is to explore the carbon implications of alternative design
considerations for afforestation and reforestation programs in the United States. Program size,
which refers to either the number of acres targeted or the level of government funding
appropriated to the program, is one design consideration that is critical to the total carbon
sequestration potential of a program. Historically, the levels at which funds have been
appropriated to existing programs have often been less than originally proposed or requested
by the administering agency, thereby altering the carbon sequestration capability of the
proposed program. To understand how to better design programs specifically to sequester
carbon, this study analyzes how the amount of carbon sequestered by programs will be
affected by different assumptions regarding program variables such as:
» Regional Distribution. Some programs, such as the Stewardship Incentives
Program (SIP), have used regional allocations to "share" out scarce resources.
We compare cost-effectiveness results of such restricted enrollments with least-
cost approaches that enroll acreage based on cost.
*> Funding Mechanism. The SIP program employs cost-sharing arrangements,
whereby a certain proportion of the establishment cost is paid by the program
in a lump sum, while the rest is paid by the farmer. We also consider various
rental payment structures that are designed to encourage the landowner to
retain the acreage in trees throughout the analysis period, 1993-203S.
» Land Type Most afforestation and reforestation programs have focused on a
single land use, such as forestland, cropland or pastureland. A more
economically efficient approach is to select acres for carbon sequestration
based strictly on least-cost per unit of carbon, regardless of land type. Barring
RCG/Hagler Bailly
-------
EXECUTIVE SUMMARY * ES-2
this, it is important to be sensitive to the variations in carbon capture and cost
as a result of targeting a specific land type.
» Program Size. To date, afforestation and reforestation programs have been
relatively small and as such have not tapped the low-cost carbon sequestration
opportunities in the U.S. We investigate the effects of larger programs on cost
and carbon sequestration. The program size variable also shows the effects of
policy on programs because we use it to model both program funding levels
that were proposed to Congress, and actual funding levels granted by Congress.
+ Enrollment Constraints. Sometimes there are resource constraints, which are
readily not included in supply curves, that limit the ability of a region to
implement a tree planting program. For example, available nursery capacity
may limit the number of acres that can be planted in any region. We
approximate such constraints in our model and analyze their effect on program
cost-effectiveness.
The study includes four timberland scenarios that explore opportunities for increasing the
acreage in the Stewardship Incentive Program (or some related program that encourages
nonindustrial, private timberland1 owners to increase timber plantings or undertake timber
stand improvement measures) with additional annual funds of $20 and $70 million.
Seven cropland scenarios focus on expanding tree planting on cropland from the current level
of about 2 4 million acres up to 3.7, 8.7 and 18.7 million acres. The current 2.4 million acres
belong to the CRP program, which encourages farmers to remove marginal, highly erodible
cropland from production, and plant trees or grasses. The cropland scenarios consider program
sizes of 1 3, 5, and 10 million acres, and borrow the CRP funding structure, which covers
planting costs and land rents. These scenarios are structured in a way that permits aggregation
across the scenarios to achieve the respective total acreages of 3.7, 8.7, and 18.7 million.
There are seven pastureland scenarios that study the possibility of expanding tree planting
activities on agricultural land to include pastureland. These scenarios consider tree planting
programs of 75, 15, and 25 million acres, and also adopt a funding structure similar to the
CRP
Finally, there are four wetlands scenarios that take approximately 4.9 million acres of
bottomland out of agricultural production and plant trees on this land as a part of a wetlands
reserve tree planting program.
1 Land is considered timberland if forest trees cover at least 10% of the area or if it was formerly
forested and has not been converted to another use, e.g . agriculture. Consequently, programs that plant
trees on timberland can reforest a previously afforested area or augment the existing trees through
additional planting or improved stand management practices.
RCG/Hagler BailJy
-------
EXECUTIVE SUMMARY ES-3
Within a category of scenarios, the scenarios essentially differ from one another by program
size and enrollment pattern (i.e., the manner in which the acres are planted over time). The
research involved constructing models to accomplish the following for each scenario:
* Construct an enrollment pattern given specific regional and enrollment
constraints, and acreage or program cost goals
» Formulate carbon sequestration curves that describe how much carbon is stored
annually over a study period from 1993 to 2035 for each enrollment cohort
(i.e., trees planted in the same year)
» Estimate annual and total costs to the government and program participants
over the study period, based on treatment or planting costs and opportunity
costs (i.e., land rent).
The model used data for per acre treatment costs and rental costs from Moulton and Richards
(1990) and Richards (1992). Carbon content data was based on information from Richards
(1992) and unpublished growth and yield curves developed by Birdsey.
By comparing carbon sequestration and costs within the scenario categories and between the
categories, the following conclusions are drawn about the effects of various program features.
» Total unit cost, equal to total cost divided by total carbon sequestered through
2035, was lowest for the timberland scenarios. These unit costs tended to be
about $3 per ton of carbon over the life of the program, compared to a range of
approximately $8 to SI6 per ton for the cropland scenarios, a range of $3 to $6
per ton for the pastureland scenarios, and a unit cost of about $8 per ton for the
wetland scenarios.
» Program unit cost, which equals program cost divided by total carbon
sequestered through 2035 (i.e., excludes federal funding costs for out-year land
rental payments), is lowest for the wetland scenarios at about $0.60 per ton.
Program unit costs for the timberland scenarios were in the $1 to $2 range.
Pastureland program unit costs were in the $2 to $4 range, while pastureland
costs were in the $3 to $5 range.
» Program expansion did not substantially increase total unit costs for the
timberland and cropland scenarios, but did appear to increase total unit costs
for the pastureland scenarios.
» Scenarios that enrolled the eligible acres on a least-cost carbon basis tended to
be more cost-effective (i.e., have lower unit carbon costs) than related scenarios
that incorporated various regional enrollment constraints.
RCG/Hagler Bailly
-------
EXECUTIVE SUMMARY ES-4
» Scenarios that had enrollment patterns that staggered the planting of trees over
time tended to have lower total costs in net present value terms, and lower total
unit costs (where discemable). This occurred because the staggered enrollment
shifted costs into the future, which ceteris paribus reduced the present value of
costs.
* Staggering enrollment over time reduces the total amount of carbon sequestered
through 2035 by a relatively small amount. However, it reduces carbon
sequestration totals for the years 2000 and 2010 considerably, sometimes to
one-fifth of the carbon sequestered by a faster enrollment schedule.
» Annual carbon increments tended to peak around the year 2015, and to remain
relatively high in the timberland, pastureland and wetland scenarios throughout
the remainder of the analysis period, while declining more noticeably in the
cropland scenarios.
> Program sizes that are modeled after either proposed or funded programs
enable us to compare actual program results with those that would have
occurred under the proposed funding level. For example, the timberland
scenarios consider programs funded at $20 and $70 millionunder varying
assumptions about program cost-sharing levels and funding allocations to
program components such as tree planting and timber stand improvement. The
$20 million scenario represents actual Congressional appropriations for the SIP
in recent years, while the $70 million represents USDA funding requests.
Comparing the carbon sequestration results, for scenarios having comparable
assumptions, we note that while the larger program was slightly less cost-
effective, its carbon sequestration rates were more than three times greater than
the smaller program's rates in the years 2000 and 2010.
» Finally, we suggest that programs which cover the full cost of sequestering
carbon will be more effective in maintaining participation levels than those
which depend on the good will of landowners not to harvest their land. This
assumption is based on economic logic rather than study results, and the way
costs were structured in this analysis further reflect that logic.
RCG/Haglcr Bailly
-------
CHAPTER 1
INTRODUCTION
The purpose of this study is to explore the carbon and cost consequences of expanding tree
planting in the United States for the purpose of sequestering carbon on private timberland and
agricultural land. Existing federal programs that are targeted at increasing tree planting
include the Stewardship Incentive Program (SIP), which is a USDA Forest Service Program
that encourages private owners of nonindustrial timberland to better manage their lands for
both timber supply and environmental quality objectives. This program is fairly limited in size
($5 to $10 million, annually), but extremely popular. Land owners are required to submit a
management plan, in return for which the cost of the actions proposed in that plan are shared
with the Federal government. By and large, the objectives of these programs are focused on
enhancing timber supplies, either through increasing plantings or else through timber stand
improvement. On average about 65% of the acreage enrolled in the SIP program is devoted to
timber supply, while the remainder is devoted to a variety of environmental quality objectives,
such as creating windbreaks to lower erosion and fostering habitat to benefit wildlife.
Another existing federal program is the Conservation Reserve Program (CRP). This is a
USDA program targeted at removing marginal, highly erodible cropland from production.
Fanners are required to submit management plans showing how they would manage the
"reserved lands." Farmers are paid the opportunity cost of their land to engage in these
programs, over a ten year period. Currently about 36 million acres are in the CRP program.
Of this amount, about 2.4 million acres is planted in trees, with the remainder planted in
grasses.
Traditionally, afforestation and reforestation programs in the U.S. have not specifically
targeted carbon sequestration as an objective. For example, the CRP is geared toward
reducing soil erosion, and while tree planting is a component of this program, the same
incentives and programmatic rules and guidelines that lead to optimal soil stabilization may
not be consistent with carbon sequestration. By the same token, the Forest Service's
reforestation programs, such as the SIP and the Forestry Incentives Program (FIP), primarily
focus on improved forest management and a variety of environmental objectives, such as
preserving wildlife habitat and reducing overland run-off. While these objectives are clearly
compatible with carbon sequestration, they do not ensure that these types of programs will
also be effective or efficient vehicles for sequestering carbon.
RCG/Hagler Bailly
-------
INTRODUCTION » 1-2
There are basic programmatic variables that can be targeted in designing efficient program for
seauesterine carbon. These are:
» Regional Distribution. Some programs, such as FIP and SIP, have used
regional allocations to "share" out scarce resources. However, a more efficient
approach would be to adopt a least-cost (per unit of carbon) allocation scheme,
whereby the program would seek to allocate funds by moving up an enrollment
supply curve, without regard to where the most cost-effective acres might be
located.
> Funding Mechanism. The FIP and SIP programs employ cost-sharing
arrangements, whereby a certain proportion of the establishment cost is paid by
the program in a lump sum, while the rest is paid by the farmer. This approach
tends to reward actions that will lead to short-run actions to improve yields
over one rotation. An alternative to this is to pay the landowner the full
opportunity cost of his actions, including not only "production costs", but any
rents that may be foregone as a result of using the land to sequester carbon
instead of its best alternative use.
» Land Type. Most afforestation and reforestation programs have focused on a
single land use, such as forestland, cropland, or pastureland. A more
economically efficient approach is to select acres for carbon sequestration
based strictly on least-cost per unit of carbon, regardless of land type. Barring
this, it is important to be sensitive to the variations in carbon capture and cost
as a result of targeting a specific land type.
» Program Size To date, afforestation and reforestation programs have been
relatively small and really have not tapped the low-cost carbon sequestration
opportunities in the U.S. In this regard, it is important to investigate not only
larger programs, incrementally, but also to integrate these programs with
existing ones, such as the CRP. Furthermore, this variable enables us to study
the effects of policy-making processes on carbon sequestration programs
because the program sizes for several of the scenarios are based on proposed or
actual federal funding levels for tree planting programs. Consequently, we can
compare carbon sequestration results for a program as it was funded with the
results for its proposed funding level.
» Enrollment Constraints. In analyzing larger carbon sequestration programs,
one must take into account resource constraints that are sometimes not included
in supply curves. For example, available nursery capacity may limit the number
of acres that can be planted in any region. Also, while programs can be
designed to fully compensate land owners for their opportunity costs, this is no
assurance that all of the land available for planting trees in a region will be
RCG/Hagler Bailly
-------
INTRODUCTION 1-3
convened to trees, either because farmers are unwilling to bear the risk (in
terms of new investments) that a program may discontinue before they can
recoup their investment, or else fanners may have nonmarket reasons for not
converting all their land into forests.
It is important to study how these variables affect a model of carbon sequestration, and how
the variables interact with one another. We have used the following approach to analyze these
issues.
Initially, rather than construct a simple least-cost curve for analysis across all types of land,
the analysis in this study is divided into four sets of scenarios, which are based on land-type.
> timberland scenarios similar to the SIP
» cropland scenarios similar to the CRP
» pastureland scenarios patterned after the CRP
» wetland scenarios created to include bottomland acreage in the analysis.
This approach allows us to make carbon sequestration and program cost comparisons across
tree planting programs on different types of land. Then we vary enrollment and funding
mechanism assumptions within the sets of scenarios. For example, some scenarios are
enrolled on a least-cost basis, while others are enrolled proportionally to existing regional
distributions of CRP acreage. This variation enables us to determine whether a least-cost basis
reduces costs compared to an alternative enrollment policy. Detailed descriptions of the
scenarios are provided in Chapter 2, but we present an overview of the types of scenarios in
each of the four sets below
First, this study explores opportunities for increasing the acreage in SIP, or some similar
program, on privately owned, nonindustrial timberland from the current level of funding to
levels in the neighborhood of $20 and $70 million, annually. These funding levels are
selected for analysis because they are based on actual federal budget figures. The $70 million
funding level approximates recent requests made by the USDA for annual funding for tree
planting programs. The $20 million funding level in our analysis approximates the annual
funding that Congress has granted for tree planting programs, which is some S50 million
below the requested amount. Alternative assumptions about cost-sharing arrangements and the
proportion of funding allocated to tree planting, timber stand improvement, and other program
components permit comparisons across different program designs.
Furthermore, this study examines the effects of increasing the size of tree planting programs
on cropland through accelerated tree planting, from the cunent acreage of 2.4 million to
levels of 3 7, 8.7 and 18 7 million acres The current 2 4 million acres are in the CRP, and the
first of the three targets represents a 1.3 million acre increase, which would make the total
acreage in cropland programs equal the original CRP goal of 3.7 million acres. The other two
targets represent additions of 5 million acres and 10 million acres, respectively, to the 3.7
RCG/Hagler Bailty
-------
INTRODUCTION > 1-4
million acres. The study also considers expanding the scope of agricultural tree planting
programs to include pastureland, as well as cropland. In so doing, the study looks at tree
planting programs of 7.5 million acres, 15 million acres, and 25 million acres.
Finally, this study analyzes the cost and carbon consequences of taking bottomland that is
currently in agricultural production and planting this land in trees as a part of a wetlands
reserve tree planting program. According to recent estimates by the Center for Agricultural
and Rural Development (CARD), there are almost 5 million acres of wetlands, which could
be reclaimed and planted in trees. This would not only restore some of the important
environmental services originally provided by these converted wetlands, such as flood
protection and wildlife habitat, it would also provide a useful sink for storing carbon in
bottomland hardwood forests.
This report is divided into five chapters. Following the introduction in this chapter, Chapter 2
reviews, in detail, the assumptions used in the twenty-two tree planting scenarios that form
the basis of this study. Chapter 3 describes the methodology used to estimate both the amount
of carbon sequestered in each of the scenarios and the cost associated with the program,
including not only program costs, but also the opportunity costs of land-owners, as well.
Chapter 4 presents the results of the scenarios in both graphic and tabular form and discusses
the results. Finally, Chapter 5 summarizes the information from Chapter 4 and provides the
main conclusions of the study.
RCG/Hagler BaiUy
-------
CHAFFER 2
SCENARIOS
This chapter describes the tree planting scenarios that were analyzed. Four sets of scenarios
were included in this analysis. The first set looked at expanding tree planting and improved
management on nonindustrial private lands. The second set of programs involved increasing
the amount of cropland on which trees are planted, while the third looked at increasing
planting on pastureland. The fourth set examined the costs and carbon consequences of
planting trees in wetlands convened to agriculture. The scenarios are described in much more
detail in Table 1 and in Sections 2.1 through 2.4.
Within each of these four sets of scenarios, the characteristics of any given scenario varies
along three dimensions:
» type of funding mechanism
» regions included in program
» enrollment schedule
Table 1 provides an overview of the scenarios, showing how each scenario varies according
to these four characteristics A key that describes the funding, region, and enrollment codes
used in Table 1 can be found on a separate page, following the table The scenario labels,
which are used to identify the individual scenarios throughout this report, are explained in the
descriptions of the scenarios in Sections 2 1 through 2.4.
2.1 TREE PLANTING INITIATIVE (TPI) ANALYSES
This section describes the assumptions used to simulate the carbon consequences of increasing
tree planting and the intensity of management on private nonindustrial timberland These
scenarios are patterned, variously, after the Stewardship Incentive Program (SIP) and the Bush
Administration's America the Beautiful Program (ATB), and the Clinton Administration's
Trees for the Future Initiative The SIP program shares with farmers the cost of planting trees
and improved management through timber stand improvement. It does not compensate
timberland owners for the opportunity cost of their land in the form of rental payments. The
payments made for tree planting under the SIP program have been in the range of $5 to $10
million, annually The ATB program was designed to meet several ends: to increase planting
on less productive forestland and to expand tree planting in suitable, nonforested areas such
as those in cities and urban fringe areas
RCG/Hagler Bailly
-------
SCENARIOS *> 2-2
Table 1
Funding, Region, and Enrollment Options for Tree Planting Scenarios
Scenario
Funding
Region
Enrollment
Tree Planting On Timberland
TPI20CS
TPI20
TPI70CS
TPI70
Fl
F2
Fl
F2
RO
RO
RO
RO
El
El
El
El
Tree Planting on Cropland
CROP1RC
CROPiREC
CROP5RC
CROP5EC
CROPIORC
CROP10EC
CROP 10
F3
F3
F3
F3
F3
F3
F3
R2
R2
R2
Rl
R2
Rl
Rl
El
E2
El
E2
El
E2
El
Tree Planting on Pastureland
PAS7
PAS7EC
PAS15
PAS15EC
PAS25
PAS25EC
PAS25ECM
E3
F3
F3
F3
F3
F3
F3
Tree Planting on Wetlands
WET5
WET5U
WET10
WETIOU
F3
F3
F3
F3
Rl
Rl
Rl
Rl
Rl
Rl
Rl
El
E2
El
E2
El
E2
E3
RO
R3
RO
R3
El
El
El
El
RCG/Hagler Bailly
-------
SCENARIOS » 2-3
Key to Table 1
Funding
Fl
F2
F3
As in current SIP program 50/50 cost share. 45% of money will be spent on
regeneration (planting) and 20% on timber stand improvement (TS1).
CRP-type structure - Fanners are paid full planting cost + die opportunity cost of
their land (i.e., rent), amortized at 4% over 10 years. If the program is greater than
$20 million, there is a three-year start up period, during which trees are phased in
gradually. A least cost approach is used in each individual region to determine
whether a portion of funds is spent on TSI, so this proportion varies across regions.
Like F2, except no phase-in period and no TSI.
Region
RO
Rl
R2
R3
Trees are planted\managed based on minimization of the carbon cost (as defined in
Fl, F2, F3). No regions are excluded.
No land is allowed to be eligible in the Rocky Mountains or Northern Plains.
The regional distribution of acreage is the same as the current distribution in CRP.
The regional distribution of planted acreage is the same as the distribution of
eligible acreage.
Enrollment
El
E2
E3
No restrictions.
a. Max. total enrollment is limited to 25% of available cropland in any
region.
b. Max. total enrollment is limited to 50% of available pastureland in any
region.
c. Max. rate of annual enrollment is 10% of available land in specified land
use for each of first three years.
d. Max. rate of annual enrollment is 20% of available land in specified land
use after the third year.
Same as E2 for a., c., d.
b. Max. total enrollment is limited to 66.67% of available pastureland in any
region.
RCG/Hagler Bailly
-------
SCENARIOS »2-4
In the TPI scenarios, tree planting programs are assumed to be funded at two annual levels:
$20 million and $70 million. The $20 million scenario is the approximate level at which tree
planting programs have been funded by Congress over the past several years, while $70
million is the approximate amount rsaifislfid of Congress by USDA over the past several
years. Two funding mechanisms are used:
» One that compensates fanners for 50% of their planting/management costs in
the year the trees are planted; 45% of total funding is spent on tree planting
while 55% is spent on various support activities including TSI
* One that is analogous to the CRP, which compensates farmers for both tree
planting costsannualized over ten years, and land rental costs for ten years
following enrollment; the proportion of total funds that pays for TSI is
determined on a least cost basis within each region, so it differs across regions.
Every scenario, here and in Sections 2.2 through 2.4, contains the assumption that trees
remain in the simulated program for 10 years following their enrollment. After that, the
number of enrolled acres in each "cohort"1 drops by a constant rate until only 95% of the
acres that were originally enrolled remain in the program in 2035.
The four alternative TPI scenarios are described in Sections 2.1.1 through 2.1.4. The scenarios
are identified by the following labels: TPI20CS, TPI20, TPI70CS, and TPI70. In addition to
the TPI designation, these labels indicate whether the scenario includes federal expenditures
of $20 million or $70 million, and whether the scenario involves cost-sharing (CS).
2.1.1 TPI20CS
All of the TPI scenarios increase tree planting on nonindustrial, private timberland. Under this
scenario, there is one planting in each of the 10 years, starting in 1993. In each of these ten
years, the federal government pays one-half of the treatment cost and the land owner pays the
other half of the treatment cost. Under this scenario, the government does not reimburse the
landowner for any other costs, such as the rent associated with the highest alternative use of
the land. Thus, total expenditures in any year are twice the amount of the federally funded
program level (i.e., federal program costs are $20 million annually and fanner costs are also
$20 million annually, for 10 years). Of the total amount expended (including both the public
and private cost shares), 45% is assumed to be spent on regeneration, while 20% of the total
funding is devoted to timber stand improvement (TSI). With this constraint in place, acres are
enrolled into the program on a least cost carbon basis.
1 The term cohort, or regeneration cohort, refers to a group of trees all of which are planted in the
same year.
RCG/Hagler Baffly
-------
SCENARIOS » 2-5
2.1.2 TPI20
Under this scenario, there is a single enrollment period in 1993. The federal government then
makes an annual payment to land owners for ten years, equal to the annualized value of their
planting and TSI costs plus the opportunity cost of the land on which the trees are planted
(i.e., a land rent). Total federal expenditures are $20 million per year. The amortization period
is 10 years, so that the federal government pays land owners the full opportunity cost of the
trees they have enrolled during that time. Acres are enrolled on a least cost carbon basis.
2.1.3 TPI70CS
This scenario is exactly like TPI20CS with the exception that the amount of money spent by
the federal government in each year (for ten years) is $70 million. Land owners also pay $70
million in each year.
2.1.4 TPI70
This scenario is similar to TPI20 with the following exception. Annual expenditures reach a
maximum of $70 million following a phase-in period of four years. Expenditures are
constrained to $20 million in the first year, and increased by $20 million in year two and
again by $20 million in year three. In the fourth year, a $10 million increase brings total
annual expenditures to $70, where they remain through year ten. This ramp-up period is
accompanied by a four-year enrollment schedule. Acreage that accounts for approximately 2/7
of the total program cost is planted in each of the first three years and acreage accounting for
the final approximate 1/7 of total program cost is planted in the fourth year. This scenario
was designed to show the effects of ramping-up to the targeted program level. Landowners
are compensated, over time, in the same way as in TPI20.
2.2 CROPLAND PROGRAMS
This section describes the assumptions in a series of analyses that are intended to estimate the
costs and the carbon consequences of planting trees on cropland, both under the existing CRP
program and under several hypothetical tree planting alternatives. The scenarios, taken as a
whole, assume that one begins with a currently enrolled base of about 2.4 million acres of
trees that are in the CRP. Then, additional cropland tree planting takes place in increments of
1.3, 5 and 10 million acres. The scenarios are designed so that these additions can be
aggregated to raise the level of tree planting on cropland acreage from 2.4 to 3.7 (the original
CRP goal), 8.7, and 18.7 million acres of trees, successively. This aggregation is
straightforward for the scenarios having the El constraint. However, the E2 constraint must
RCG/Hagler Bailly
-------
SCENARIOS » 2-6
be relaxed somewhat to aggregate the scenarios with that constraint; this point is discussed in
the descriptions of the relevant scenarios.
All of the scenarios use a CRP-type funding structure, with payments lasting for ten years. In
addition, it was assumed that once trees are planted in any given year, that cohort of trees
will remain in the program for 10 years. Then the number of enrolled acres in each cohort
drops by a constant rate until only 95% of the acres, originally enrolled, remain in the
program in 2035.
The labels for the eight scenarios in this section are: CROPBASE, CROP1RC, CROP1REC,
CROP5RC, CROP5EC, CROP10, CROP10EC, and CROP10RC. CROPBASE represents the
currently enrolled base of 2.4 million acres. Labels for the other scenarios include an element
(either a 1, 5, or 10) that indicates the additional acreage for the scenario. The RC and EC
designations indicate whether the scenario contains a region constraint (R2) or an enrollment
constraint (E2), respectively. Scenario CROP1REC contains both constraints, and scenario
CROP 10 contains no constraints.
2.2.1 CROPBASE
It is assumed that there are approximately 2.4 million acres of trees in the tree portion of the
CRP. The carbon sequestered by these trees has already been procured. For aging purposes,
these acres are treated like a cohort planted in the first year of the program.
2.2.2 CROP1RC
Under this scenario, 1.3 million acres of trees are added during the three years, 1993-1995, so
that the total cropland tree acreage in 1996 is approximately 3.7 million acres. The
distribution of the new trees in each year is assumed to be proportional to the current
distribution of tree acreage in CRP (R2); this distribution is shown in Table 2. Subject to this
constraint within each region/state, enrollment is on a least cost basis. In all of the cropland
scenarios, annual payments are made to land owners for ten years. That is: once a farmer
plants trees, she or he is paid an annual sum equal to the land rental + annualized treatment
cost. (As in the TPI analysis the amortization period is ten years.)
2.2.3 CROP1REC
The only change between this analysis and CROP IRC is in the enrollment rate. Instead of
being entirely on a least cost carbon basis, there are constraints applied to the amount of land
that can be enrolled in each year. The enrollment constraints (E2) are as follows:
RCG/Hagler Bailly
-------
SCENARIOS » 2-7
Table 2
Current Distribution of Tree Acreage in the
Conservation Reserve Program by Region
Region
Northeast
Lake States
Combelt
Northern Plains
Appalachian
Southeast
Delta States
Southern Plains
Mountain
Pacific
Total
* Total may
Acreage (1,000s)
14.1
120.8
88.2
5.6
147.3
12S8.6
696.8
22.3
4.6
5.8
2364.2*
differ due to rounding.
Percent of Acreage
0.6%
5.1%
3.7%
0.2%
6.2%
53.2%
29.5%
1.0%
0.2%
0.2%
100.0%
» Max. Total Enrollment: 25% of total cropland in each region
» Max. Annual Enrollment: 10% of wet or dry cropland in each of the first three
years; 20% of wet or dry cropland in each year subsequent to the third year.
The effect of these constraints, among other things, is to control the period over which trees
are planted. (However, as it turned out, this constraint was not binding and the CROP IRC
and CROP1REC scenarios were effectively the same).
2.2.4 CROP5RC
Under this scenario 5 million acres of trees are added to those already in CROP1RC and
CROP1REC. This makes a total of approximately 8.7 million acres of trees in a cropland
program. The additional trees are planted in proportion to the existing baseline acres (i.e., the
2.4 million acres). Within each region, the choice between wet or dry cropland acres is made
RCG/Hagler BailJy
-------
SCENARIOS > 2-8
on a least cost carbon basis. Trees are phased in over 10 years, starting in 1996, such that
one-tenth (1/10) of the additional 5 million acres of trees are planted in each of the years
1996-2006, or 500 thousand acres per year.
2.2.5 CROP5EC
Under this scenario, 5 million acres of trees are planted in a cropland program. No trees are
planted in the Rocky Mountain and Northern Plains regions (region constraint Rl).
Enrollment is on a least cost carbon basis, constrained according to the E2 constraints (See
2.2.3 above). Given the imposition of the E2 constraint, the trees are phased in fairly quickly:
all 5 million acres could be planted in a single year, 1996.
These S million acres can be added to the 2.4 million baseline acres and the 1.3 million acres
in CROP1REC, bringing the total cropland acreage to 8.7 million. This aggregation is
possible because both scenarios, CROPSEC and CROP1REC, are constrained to enrollment
levels of 10% of the available land in any region because all acres are planted in the first
year. Costs are uniform within a region, so there is no overlap in the land planted by the two
scenarios. Aggregating the two scenarios effectively violates the 10% enrollment rate
constraint. However, enrollment rates for CROP1REC are below 5% of the available acreage
in all regions because enrollment was also constrained to match the existing CRP distribution
(the R2 constraint). Consequently, the actual aggregate enrollment rate is below 15%, and the
E2 constraint for total enrollment in a region, which is a maximum of 25% of available
acreage, is still satisfied. Had the analysis applied the 10% acreage constraint to a single-year
enrollment of all 6.3 million additional acres (1.3 million + 5 million acres) it would have
shifted some of these acres to higher cost regions, thereby increasing total program cost and
reducing total carbon sequestration.
2.2.6 CROP10RC
In this scenario, 10 million acres of trees are added on top of the acres in CROP5RC and
CROP5EC, bringing the total acreage in the program to 18.7 million acres. Trees are phased
in over 10 years, starting in 1996, such that one-tenth (1/10) of the additional 10 million acres
are planted in each of the 10 years, 1996-2006, or one million acres per year. Enrollment each
year is proportional to the current distribution of tree acreage in CRP (R2).
2.2.7 CROP10EC
10 million acres of cropland are enrolled, excluding acreage in the RM and NP regions (Rl).
Timing of plantings and payments is determined by the E2 constraints (See 2.2.3 above). In
this case, it was possible to satisfy the E2 constraints and still plant the additional 10 million
RCG/Hagler Bailly
-------
SCENARIOS * 2-9
acres, of trees in a single year, 1996. These acres can be added to CROPBASE, CROP1REC,
and CROPSEC to bring total cropland acreage to 18.7 million. Aggregating the scenarios as
described results in a violation of the first-year enrollment constraint, i.e., 10% of available
acreage in each region. The causes and effects are the same as those discussed in 2.2.5.
Consequently, the aggregate constraintthat no more than 25% of the total available acreage
be enrolledis not violated because less than 5% of the acreage is enrolled in CROP1REC,
so aggregate acreage must be less than 25% given that enrollments in CROPSEC and
CROP10EC are constrained to 10% of available acreage within a region. Similarly, had the
10% enrollment rate constraint been applied to a single-year enrollment of 16.3 million acres
(1.3 million + 5 million + 10 million), then acreage in more costly regions would have been
planted, raising costs and reducing carbon sequestration.
2.2.8 CROP10
In this scenario, 10 million acres of trees are added to the acres in CROPSRC and CROPSEC,
bringing the total acreage in the program to 18.7 million acres. None of the additional trees
(10 million acres) are planted in the RM and NP regions (Rl). As in CROPSRC, these trees
are added over a 10 year period, starting in 1996, such that one million acres of new trees are
planted in each year.
2.3 PASTURELAND PLANTING
This section describes the assumptions in a set of scenarios that were used to examine the
costs and the carbon consequences of planting trees on pastureland, using a CRP-type
payment structure. Three funding levels were considered:
» $110 million
* $220 million
» $500 million.
The first funding level was based on the $110 million funding proposed by the Bush
Administration for the America the Beautiful Tree Planting Program. The second two levels
represent more aggressive funding levels, essentially doubling and quadrupling (then rounding
up to an even $500 million) the proposed amount. These funding levels correspond to acreage
levels of 7.5 million acres, 15 million acres, and 25 million acres, respectively. We will refer
to the acreage levels rather than the funding levels throughout the analysis.
All of the scenarios used a CRP-type funding structure, with annualized planting and rental
payments lasting for ten years. In addition, it was assumed that once trees are planted in any
given year, that cohort of trees will remain in the program for 10 years. Then the number of
RCG/Hagler Bailly
-------
SCENARIOS * 2-10
enrolled acres in each cohort drops by a constant rate until only 95% of the acres, originally
enrolled, remain in the program in 203S.
The labels for the seven pastureland planting scenarios include the PAS designation and a
number that indicates the level of acreage planted (either 7, IS, or 25). A label includes the
letters EC if the scenario has an enrollment constraint (E2).
2.3.1 PAS7
No land from RM and NP regions is used in this scenario (Rl). Trees are phased in over 10
years, such that one-tenth (1/10) of the additional 7.5 million acres are planted in each of the
10 years, starting at Jan. 1, 1996 (750K acres/yr). This gives 10 regeneration cohorts. Farmers
are paid the opportunity cost of planting these trees annualized over a ten year period, plus
annual rental payments for ten years.
2.3.2 PAS7EC
Trees are planted on 7.5 million acres of land, excluding land in the RM and NP regions
(Rl). Farmers are paid the opportunity cost of planting these trees annualized over a ten year
period, plus annual rental payments for ten years, for ten years. The E2 constraints dictate the
timing of planting. Enrollment is on a least cost carbon basis, constrained according to (E2).
The pastureland constraints are as follows:
» Max. Total Enrollment: 50% of total pastureland in each region
+ Max. Annual Enrollment: 10% of wet or dry pastureland in each of the first
three years; 20% of wet or dry pastureland in each year subsequent to the third
year.
In the analysis, it was possible to plant 7.5 million acres in two years, starting in 1996, and
still satisfy the above constraints.
2.3.3 PAS15
This scenario is the same as PAS7, except that 15 million acres are planted.
RCG/Hagler Bailly
-------
SCENARIOS »2-ll
2.3.4 PAS15EC
This scenario is the same as PAS7EC, except that IS million acres are planted. In this
analysis, it took four years to plant the IS million acres of trees and still satisfy the
constraints mentioned in Section 2.3.2.
2.3.5 PAS25
This scenario is the same as PAST, except that 25 million acres are planted.
2.3.6 PAS25EC
This scenario is the same as PAS7EC, except that 25 million acres are planted. The maximum
number of trees that could be planted in this scenario was about 17 million acres, over a four
year period, because of the enrollment constraint. Increasing the planting period did not affect
this result.
2.3.7 PAS25ECM
This scenario is the same as PAS2SEC, except that the £2 constraint was modified to increase
the maximum acreage from 50% to 66 67% of the available pastureland. The ECM
designation within the label indicates that the enrollment constraint is modified. Under the
modified constraint, all 25 million acres could be planted in a period of five years.
2.4 BOTTOMLAND HARDWOOD PLANTING
This set of scenarios involves planting trees on about 4.9 million acres of land that is in the
bottomland hardwood reserve category. Table 3 shows the distribution of the acreage by
region. All of the acreage will be planted under each of the four enrollment strategies. Cost
estimates are based on the cost of treatment, which is annualized over the 5 or 10 year
enrollment period, and the opportunity cost of land. As with all other programs, we assume
that 95% of the acreage will remain planted in 2035.
The bottomland hardwood scenarios were selected to compare cost-effective enrollment across
regions with a proportional enrollment that matched the regional distribution of available
lands. The proportional enrollment essentially plants a uniform percentage of regional acreage
in each enrollment period, e.g., 20% of the available acreage in a region is planted each year
for a five-year enrollment strategy. The five and ten-year enrollment schedules allow us to
specifically consider the effect of shorter enrollment schedules on unit cost.
RCG/Hagler Bailly
-------
SCENARIOS » 2-12
Table 3
Bottomland Hardwood Reserve Acres by Region
Region
Northeast
Lake States
Combelt
Appalachian
Southeast
Delta
TOTAL
Acres (1000s)
51.6
959.8
2104.3
381.5
112.8
1388.1
4998.1
Labels for these scenarios are: WETS, WET5U, WET10, and WET10U. The numerical
component indicates whether the enrollment period lasts for 5 or 10 years. The two scenarios
that constrain the annual enrollment to be a uniform percentage across regions include a U to
distinguish them from the least-cost enrollment scenarios.
2.4.1 WETS
In this scenario, the acres are planted over a five-year period, based on a least cost carbon
approach. Each year, 1/5 of the total 4.9 million acres are planted, starting in the low cost
regions. This planting regime results in 5 regeneration cohorts. The planting costs are
annualized over the 5-year program. Land owners receive annual payments for 5 years equal
to the sum of the annualized planting cost and annual land rent.
2.4.2 WET5U
This scenario assumes that trees are planted over a five-year period, beginning in 1996. A
uniform planting schedule is utilized within each region during the program phase, so 1/5 of
the acreage in each region is planted each year.
RCG/Hagler Bailiy
-------
SCENARIOS »2-13
2.4.3 WET10
This scenario is similar to WET5, except that a 10 year planting period is used instead of a 5
year planting period. Thus, there are ten regeneration cohorts, and costs are annualized over
10 years.
2.4.4 WET10U
This scenario is similar to WET5U, except that a 10 year planting period is used.
RCG/Haglcr Bailly
-------
CHAPTER 3
DATA AND METHODS
This chapter outlines the procedures used to estimate the cost and quantities of carbon
sequestered under each of the twenty-two scenarios, described above. It also describes the
data used in these analyses.
3.1 DATA
The data used to estimate the cost and quantities of carbon sequestered in the various
scenarios is described in Moulton and Richards (1990) and Richards (1992). This section
briefly summarizes the data used in this study for:
» treatment cost and land rents
» timber and carbon yields
> eligible acreage.
3.1.1 Treatment Cost Data
The cost of establishing trees and the land rental rates used in this study are based on
estimates in Moulton and Richards (1990). They provide cost estimates for six different land
categories in the ten USDA farm production regions:1 wet and dry cropland, wet and dry
pastureland, planting timberland, and active management (timber stand improvement) on
timberland. Establishment costs ranged from a low of around $60/acre to establish trees on
cropland in the Southern Plains and the Southeast to a high of about $200/acre to establish
forests on pastureland in the Northeast. Land rents also varied considerably, both by type of
land and by region. Rental rates on forestland were as low as $4.35/acre in the Pacific
Northwest and as much as $12.1 I/acre in the Northeast. Rental rates on cropland varied from
a high of $8I/acre in the Cornbelt to $45/acre in the Mountain region. Pastureland rental
rates, typically, ran from about one-half to one-third the value of cropland rental rates.
Costs were originally annualized in Moulton and Richards (1990) assuming a 40 year
amortization schedule at a real discount rate of 10 percent. However, for this study, it was
assumed that the programs would basically last for just 10 years. Also, the opportunity cost of
1 Northeast, Lakes States, Combelt, Appalachian, Southeast, Delta, Northern Plains, Southern Plains,
Mountain, and Pacific.
RCG/Hagler Bailly
-------
DATA AND METHODS * 3-2
capital was assumed to be much lower than in Moulton and Richards: a real rate of 4% was
used to discount costs in this analysis. This reflects a nominal rate of return in the area of
7%, with an expected rate of inflation around 3%. Decreasing the discount rate and the
amortization period, simultaneously, had the effect of slightly increasing the average total cost
of sequestering carbon on both a per acre and per ton basis by about 2-5% for all land
categories.2
3.1.2 Timber and Carbon Yield Data
The timber and carbon yield data used in the sensitivity analysis part of this study are from
Richards (1992) and from unpublished growth yield tables made available by Birdsey
(1992).3 Moulton and Richards (1990) estimated regional timber and carbon yields for the
same six land categories identified above. However, the estimated carbon yields contained
systematic errors that biased these yields upward in the neighborhood of 20%. Richards
(1992) has recently revised this earlier set of yields in Moulton and Richards (1990) to be
fully consistent with the methods developed by Birdsey (1991).4
3.1.3 Eligible Acreage
The eligibility data for all but the wetlands cases is from Moulton and Richards (1990). They
provide estimates of the land available in the ten farm production regions for the six broad
land use categories defined above. Eligible cropland and pastureland include highly credible
land, lands in land capability classes V-Vm, and lands classified as wet soils. In all, they
identify about 225 million acres of eligible cropland in the U.S. and about 31 million acres of
eligible pastureland. Timberland eligibility was determined based on suitability to treatment
type In all about 80 million acres of timberland were deemed to be eligible for treatments
under SIP-type programs. Eligible acreage for the wetlands cases was shown, previously, in
Table 2. These data were supplied by CARD.
2 Decreasing both the discount rate and the amortization period increased the capital recovery factor
by about 20%, with a corresponding effect on the annualized value of treatment costs; however, rental
rates are not affected by these computations, so the net effect on average total cost is less than 20%.
3 Personal communication, Richard Birdsey, U.S. Forest Service, July 2, 1992.
4 Some confusion about Birdsey's methodology led to a systematic over-estimate of carbon yields in
Moulton and Richards (1990). The source of this confusion and the revised procedure is outlined in
Richards (1992).
RCG/Hagler Bailly
-------
DATA AND METHODS » 3-3
3.2 METHODS
We had intended to design a computer model that would enable us to estimate the carbon and
cost consequences for this study. However, the final scenarios that were selected for analysis
varied so greatly that it was not possible to develop a single model to generate the cost and
carbon sequestration results for all twenty-two scenarios using the same model and
methodology. Nevertheless, it is possible to define a common methodology that was applied
in broad terms to all of the twenty-two scenarios. The steps used in this methodology are
presented in the following sections.
3.2.1 Determine Enrollment Distribution/Schedule During the Planting Period
The first step involved allocating acres to the hypothetical tree planting program over time
and space, consistent with the specifications in a given scenario. From a methodological
standpoint, the major consideration was whether or not the acreage allocation was
proportional to the current acreage (as in some of the cropland cases). If the acreage to be
allocated was based on the existing distribution of acreage in CRP, then this regional
distribution was applied to the eligible acres in each region, as determined by Moulton and
Richards. The resulting acreage was removed from the eligible pool of acres in each region,
consistent with this distribution and the temporal schedule in the scenario. A mass balance
was kept for each region to ensure that the amount of acreage enrolled in each period from
the eligible acres could not exceed the amount remaining. This never occurred.
For all of the remaining cases, the allocation of acres over time and space was accomplished
with a computer program developed for this study. The program developed supply curves for
land and carbon, consistent with the land use specification and the funding mechanism, region
and enrollment constraints in Table 1. It kept mass balances on the acreage in each of the
land use categories to ensure that the amount of acreage enrolled in each period from the
eligible acres could not exceed the amount remaining. If no land was available, then the
model moved to the acreage that was least costly from a carbon production perspective (i.e.,
least cost carbon). The output of the program was the number of acres that could be planted
in each region during each of the planting periods and the amounts of carbon in each region,
in each year of the planting schedule, consistent with the acreage allocation.
The initial computer program for constructing the supply curves contained information about
average annual carbon sequestration over entire rotations in each region. This was modified in
the process of the research to include the growth and yield information supplied by Birdsey.
We matched the species distribution from Moulton and Richards (1990) by region and land
type with the growth and yield tables, and constructed weighted yield tables for each land
type in each region. For each of the component yield tables, we created a ratio of the carbon
increment at each age to the average increment. By multiplying the average annual carbon
increment derived from the least-cost model in each year by the corresponding (i.e., by age,
RCG/Hagler Bailly
-------
DATA AND METHODS »3-4
region, and land type) carbon increment-to-average carbon ratio, we were able to develop
carbon yield curves for each scenario that approximated sigmoidal growth patterns.
3.2.2 Calculate Carbon Sequestration in Each Year of the Program
The first step of the analysis provided the basic information about how much acreage was
planted in each year and how much carbon could be sequestered during each year of the
planting. However, unless all of the planting could take place in a single year, the model was
not able to account for the effect of "staggering" the regeneration cohorts over time on the
amount of carbon sequestered across cohorts in each year. Thus, the acreage and carbon
profiles for each regeneration cohort were taken from the previous step and read into another
program which staggered each of these profiles, according to the planting schedule.
Next, the carbon in each cohort was reduced by a decay function, beginning ten years after
the planting date, such that in the terminal year of the scenario only 95% of the acreage (and
carbon) was left in that cohort. This was accomplished using a function in the form of
C,*exp{ln(0.95)*n/D), which simplifies to C^O.951"0, and where C, is the amount of carbon
in the current year (t); D is the total number of years in the decay period, and n is the current
year in the decay period (i.e., the number of years elapsed from the start of the decay period
plus one). For example, the remaining carbon in the first year of the decay period for a cohort
planted in year 1 would be C,*0.95>/33, and the remaining carbon in the second year of the
decay period would be Q'O^S2733. The remaining carbon in the first decay year for the
second cohort would be C,*0.95I/32 because the second cohort has only 32 years to decay to
95%. The resulting carbon data were in the form of an N by M matrix, where N (rows) is the
number of years in the scenario, from the planting of the first regeneration cohort to the
terminal period, and M (columns) is the number of regeneration cohorts. A similar matrix was
constructed containing the participating acreage data.
For example, if the number of regeneration cohorts is 10 and the scenario length is 42 years,
then the first 10 entries in the first column would show the annual incremental carbon
associated with the growth of the trees in the first cohort over the first ten years of the
program. The values of the annual increments in the remaining entries in that column would
be slightly less than they would originally have been to reflect the gradual acreage loss. In the
next column, trees would be planted in year two, so the first row entry would be zero, and the
next ten entries (from year 2 to year 11) would indicate the annual incremental carbon
associated with the planting of that cohort. Then the entries from year 11 to year 42 would be
the annual incremental carbon amounts minus some small amount that represents the gradual
acreage loss. The rate of acreage loss for the second cohort would be slightly faster than for
the first cohort. Each column in this example matrix would show this "staggered" pattern,
until column 10, when planting would take place in year 10 and acreage loss would take
place from year 20 to year 42.
RCG/Hagler Bailly
-------
DATA AND METHODS » 3-5
Finally, given the structure of the above matrix, the carbon amounts were summed across
each cohort in each year to give the amount of carbon sequestered in each year of the
program. These total annual carbon increments were graphed. The cumulative amount of
carbon sequestered in each year was also calculated by summing the increments over time.
3.2.3 Cost Calculation
Costs were also calculated using the information produced in the previous step. Costs were
calculated for program years and for out-years. Program costs included the net present value
of the planting/management costs plus the present value of land rental costs for the trees that
were planted during the program period. At that period in time, the land owner was fully
compensated for his or her use of land to plant carbon.1 Out-year costs include the present
value of the land rental payments, once the land owner or farmer is no longer in the program.
These costs are included in the calculation of total costs, based on the assumption that land
owners are paid to remain in the programs. Alternatively, one could assume that land owners
would voluntarily remain in the program without receiving compensation for the opportunity
cost of their land. In that case, the out-year costs could be deducted from the total costs.
Costs for the program years were calculated by combining our estimates of per acre and per
ton carbon costs with the data in the acreage and carbon matrices for each scenario.
5 Note that we implicitly assume that the trees are not harvested, because landowners are
compensated not to do so through rental cost payments in program out-years. If landowners were not
compensated, then they might harvest the trees on their land and the revenues from harvesting would offset
the costs of producing the carbon by some amount. Thus, landowners, probably would not need to be paid
the full opportunity cost to be compensated for "growing" carbon. However, this phenomenon can not be
assessed with a least cost carbon model. What is needed is a model that can estimate both the benefits
from harvesting trees and the costs of planting them.
RCG/Hagler Baffly
-------
CHAPTER 4
RESULTS
This chapter presents the results for the twenty scenarios in graphic and tabular form, and
analyzes the following for each of the four sets of scenarios in Sections 4.1 through 4.4:
» incremental carbon sequestration
* cumulative, or total carbon sequestered
» total cost, and cost components
» regional enrollment distributions
* unit costs.
4.1 TPI RESULTS
The annual increments of carbon sequestration are shown for each scenario in Figure 1. These
increments (and those in the rest of the figures) are derived from underlying growth and yield
tables contained in Richards (1992). These curves reflect the fact that trees sequester carbon
slowly, at first, and then more rapidly, until they reach a point where both annual growth and
carbon sequestration slow down. This pattern is affected in some scenarios by staggered
enrollment patterns (i.e., the staggered enrollment has a dampening effect on the aggregate
growth rates) and by the general decay rate assumption that reduces the number of enrolled
acres to 95% after a cohort reaches the age of 10 (i.e., the decay rate offsets increasing
growth rates or compounds decreasing growth rates). The large differences between the results
for TPI20CS and TPI20 and between TPI70CS and TPI70 are due to fundamental differences
in the structure of the two programs, as will be discussed.
Figure 2 shows cumulative carbon sequestration for the period 1993-2035 for each scenario;
these totals are also shown in Table 4. The two cost-share programs (TPI20CS and TPI70CS)
sequester a total of about 240 and 700 million tons, respectively, over the period 1993-2035,
while the two corresponding programs with opportunity cost structures (TPI20 and TPI70),
i.e., where rental payments are made during the program to cover the opportunity cost of the
land, sequester less than half these amounts TPI20CS and TPI70CS sequester more carbon
largely because the 50% cost share assumption means that land owners will spend S20 million
and $70 million, respectively, planting trees and the government will also spend that same
amount. Thus, about twice as many trees could be planted compared to the TPI20 and TPI70
scenarios. Moreover, because the cost of planting trees is less than the true opportunity cost
of the trees (which includes both the planting costs and the displaced rents), more trees can
RCG/Hagler Bailly
-------
RESULTS 4-2
Figure 1
Annual Carbon Sequestration by TPI Scenario
24
20 -
c
O
18
l2
4)
3 B
cr
01
W
c
O *
X]
u
O
e-
B
A
1TI20CS
TFIZO
TPI70CS
TPI70
" A..,.
H
-©
800
CO
O
O 600
500
T3
U 400
en
% 300
cr
03
O
to
200
100 -
mm
«:&:&
TPI20CS
TPI20 TPI70CS
TPI Scenario
TPI 70
-B-
1990 199S 2000 2005 2010 2016 2020 2025 2030 2035
Year
Figure 2
Total Carbon Sequestration by TPI Scenario
(1993-2035)
RCG/Hagler BaUly
-------
RESULTS » 4-3
Table 4
Tree Planting Initiative Analysis
Carbon Sequestration and Costs
TPI Scenario:
TPI20CS
TPI20
TPI70CS
TPI 70
Carbon Sequestration (million tons)
Total (1993-2035) 237.2 100.6 705.1 330.6
Total by 2000 0.5 1.1 1.7 18
(Annual Increment in 2000) (0.3) (0.6) (1.0) (12)
Total by 2010 27.2 22.2 88.2 624
(Annual Increment in 2010) (5.5) (3.0) (17.3) (101)
Cost Estimates (billion $)t
Program Cost $0.16/50.16 $0.16/-' $0.57 / $0.57 $0.48/-'
(Program/Fanner)
Out-Year Cost $0.38 $0.13 $1.13 $0.33
Total Cost $0.70 $0.29 $2.15 $081
Unit Costs ($/ton)t
Program Unit Cost S0.67/S0.67 $1.59/-* $0.81/50.81 S1.45/-'
Total Unit Cost $2.95 $2.88 $3.22 $245
All costs paid by program
t Program costs include federal funds for planting and any rental payments made to landowners during
the program, out-year costs represent land rents between the time a cohort leaves the program and the
year 2035; total costs equal program costs plus out-year costs.
J Program unit cost equals program costs divided by total carbon; total unit costs equals total costs
divided by total carbon.
be planted and more carbon can be sequestered per dollar spent in TPI20CS and TPI70CS
than in the other two scenarios, which allocate some funding to pay land rental opportunity
costs.
Table 4 also shows the cumulative amounts of carbon sequestered by the years 2000 and
2010 By the year 2000, TPI20 and TPI70 have sequestered more total carbon than their
counterparts This occurs because these two scenarios have substantially shorter enrollment
periods, e.g., all trees are planted in one year for TPI20 compared to a 10-year enrollment
schedule for TPI20CS. However, by the year 2010, the two cost-sharing scenarios have
surpassed the smaller programs in terms of total carbon sequestration due to the greater
number of acres planted in TPI20CS and TPI70CS.
RCG/Hagler Bailly
-------
RESULTS > 4-4
Several cost estimates of the TPI scenarios are shown in Table 4. As stated in Section 3.2.3,
all costs are in present value terms. Costs for each scenario are divided into two parts:
program costs and out-year costs. Program costs include the present value of the treatment
costs plus any rental cost payments during the life of the program (e.g., ten years for
TPI20CS). Program costs for the two cost-sharing scenarios are further divided into federal
funding costs and farmer costs. All program costs are paid by the federal government in
TPI20 and TPI70. Out-year costs comprise rental payments to land owners to encourage them
to retain the trees on their land. This convention is consistent with social cost accounting,
assuming that land owners are paid not to harvest the trees on their land. If one assumes that
land owners would remain in the program without receiving compensation, then out-year costs
can be deducted from total costs. Finally, total costs are the sum of all program and out-year
costs.
To obtain a rough measure of the cost-effectiveness of the TPI scenarios, we calculated unit
costs from total costs and the program cost component. Total unit cost equals total cost
divided by the total number of tons sequestered from 1993 to 2035. Program unit costs are
derived by dividing program costs by the total amount of carbon sequestered during the
period 1993-2035. These ratios are shown at the bottom of Table 4.
Total unit costs range from around $2.45/ton for TPI70 to $3.22/ton for TPI70CS. Cost-
effectiveness comparisons are somewhat inconclusive because several of the characteristics
that affect unit costs change simultaneously across the scenarios. However, it appears that
total unit costs generally increase as program size increases. This happens because the
marginal cost of carbon sequestration is rising under the least-cost enrollment structure. The
exception is that total unit cost falls from $2.88 for TPI20 to $2.45 for TPI70. Both scenarios
use a least-cost approach for enrolling acreage across regions. The larger program is able to
enroll acreage in the Pacific region, while the smaller program cannot. This difference can be
seen in Table 5, which shows the regional enrollment shares for the TPI scenarios. The
Pacific region has fairly high carbon yields, but it also has high planting costs, which gives it
a slightly higher unit cost than the regions in TPI20. However, rental costs in the Pacific
region are very low, so out-year expenses per ton of carbon are much lower in TPI70 than
they are in TPI20. Furthermore, program unit costs for TPI70 are lower because it has more
planting cohorts than TPI20, which has only one. This effectively reduces the present value of
program costs because they are prorated over a longer time horizon.
In general, we would expect that the scenarios that constrained the type of management and
that enrolled land based on treatment costs (i.e., TPI20CS and TPI70CS) would be less cost-
effective than the scenarios that did not constrain management and enrolled land based on the
total cost of carbon (i.e., TPI20 and TPI70). This turned out to be the case. As Table 4
shows, the cost per ton of TPI20 ($2.88/ton) was slightly less than TPI20CS ($2.91/ton),
while the cost of TPI70 ($2.45/ton) was more than $0.70 per ton lower than the cost of
TPI70CS ($3.22/ton).
RCG/Hagler Bailly
-------
RESULTS » 4-5
TablcS
Tree Planting Initiative Analysis
Regional Mil of Enrolled Acreage by Scenario
Scenario:
Region:
Lake States
Combelt
Northern Plains
Delta States
Southern Plains
Pacific
Total
TPI20CS TPI20 TPI70CS
48.9% 15.7%
25.7% 92.4% 18.7%
34% 7.6% 1.1%
10.7%
7.9%
22.0% 45.9%
100% 100% 100%
TPI70
73.4%
4.3%
22.3%
100%
4.2 CROPLAND RESULTS
Figure 3 displays estimates for the annual increments of carbon sequestered by each of the
cropland scenarios, individually, over the period 1993-2035.' The two smallest scenarios,
CROP1RC and CROP1REC, both sequester about 3 million tons of carbon/year on average,
once the acres in these programs are fully enrolled. CROP5RC and CROP5EC, both of which
involve additions of 5 million of acres of trees, have average sequestration rates of about 10
to 13 million tons of carbon/year. For the three cases involving the addition of 10 million
acres of trees (CROP10RC through CROP 10), the average annual increments range from
about 21 million to 25 million tons of carbon/year. These averages are consistent with
average annual carbon increments in the range of 2 to 2.5 tons/year, as reported in Richards
(1992).
Figure 3 illustrates that the length of the enrollment period has important effects on carbon
sequestration in the 43-year period. The annual incremental carbon sequestration curves for
CROP5EC and CROP10EC have earlier and larger peaks than their counterparts, CROP5RC
and CROP10RC. This is a direct result of the shorter enrollment periods of CROP5EC and
CROP10EC.
1 Cases CROP1RC and CROP1REC are shown together because their results were identical. This
convention is followed throughout this section. A combined label, CROP1R/E, identifies these scenarios on
some of the figures.
RCG/Hagler Bailly
-------
RESULTS *4-6
I
c
o
-*->
c
o
TJ
V
h
a)
0)
3
in
c
o
-a
Figures
Annual Carbon Sequestration Above CRP Baseline
by Cropland Scenario
28
4
ft
e- -
e- -
+- -
o-
"?
- CROP1RC.REC
- CROP5RC
CROP10RC
> CEOP10BC
CROP 10
.<&
A-J
/ .''/* ^.E
' '_^ ^ *
/
/
/
/ /v
^ //
J'!*...,,.:
V
£..- ..
^>:""'
-
7 '
1 /^
...-A-"
...yCJ..
X
;
j
\
^ ''
".
' .
'x
vv
>J
«
'". N
'^..
" -^ ->
-e ~
N
s N.
sx >
'"'. s
' ' '^".
-
s
x^lr
""
^4".-
e .
1990 1995
2000
ZOOS
2010 2016
Year
zozo
2025
Z030
Z035
Figure 4 shows the total amount of carbon sequestered as of 2035 in each of the individual
cropland scenarios; these amounts are also reported in Table 6.2 These totals again show that
the shorter enrollment periods sequester larger amounts carbon within the analysis period.
Table 6 also reports total carbon sequestration for each scenario from its inception to the year
2000 and the year 2010. Again, the shorter enrollment periods for CROP5EC and CROP10EC
sequester substantially higher amounts of carbon than CROP5RC and CROP10RC or
CROP10, respectively. This is particularly true for the year 2000, where the carbon
sequestered by CROP5EC and CROP10EC is roughly five times greater than their counterpart
scenarios.
Figures 5 and 6 show the aggregate amounts of carbon that could be stored by combining
similar scenarios. For example, if one combines the carbon stored over the period 1993-2035
in CROPBASE with CROP IRC, CROP5RC and CROP10RC, total storage is in the range of
2 These amounts do not represent aggregate sequestration across like scenarios. For example,
CROP5RC sequesters a total of 426.4 million tons by itself, and CROP IRC sequesters a total of 131
million tons, so if they were combined to increase total CRP acreage from 2.4 million acres to 8.7 million
acres, then aggregate sequestration would be 557A million tons.
RCG/Haglcr Bailly
-------
RESULTS
Figure 4
Total Carbon Sequestration by Cropland Scenario
(1993-2035)
c
o
c
o
1200
1000
- 800
4)
t-
VI
0)
3
cr
-------
RESULTS »> 4-8
Cropland
Scenario:*
CROP
BASE
CROP1RC A
CROP1REC
CROPSRC
CROPSEC
CROP10RC
CROP10EC
CROP10
Table 6
Conservation Reserve Program and Cropland Analysis
Carbon Sequestration and Costs
for Individual Cropland Scenarios
Carbon Sequestration (million tons)
Total (1993-2035)
Total by 2000
(Annual Increment in 2000)
Total by 2010
(Annual Increment in 2010)
Cost Estimates (billion $)t
Program Cost
Out-Year Cost
Total Cost
Unit Costs (S/lon)J
244.0
46.3
(5.8)
103.9
(57)
131.0
5.8
(1.4)
36.2
(4.2)
$0.63
$0.79
S1.42
426.4
2.8
(1.2)
53.8
(9.5)
S1.8S
S2.31
S4.16
541.6
13.4
(4.5)
120.5
(17.8)
Sill
S2.92
$5.03
852.9
5.5
(2.4)
107.6
(19-0)
$3.75
$4.71
$8.46
1,028
25.4
(8.5)
228.7
(33.8)
$4.60
$6.26
$10.86
992.5
6.6
(2.8)
125.4
(22.0)
$3.38
$4.24
$7.62
Program Unit Cost
Total Unit Cost
$4.81
$10.84
$4.34
$9.76
$3.90
$9.29
$4.40
$9.92
$4.47
$10.56
$3.41
$7.68
Results are shown for individual scenarios, e.g.. results for CROP5RC do not incorporate the results
shown for either CROP IRC or CROPBASE.
t Program costs include federal funds for planting and any rental payments made to landowners during the
program; out-year costs represent land rents between the time a cohort leaves the program and the year
2035; total costs equal program costs plus out-year costs.
J Program unit cost equals program costs divided by total carbon, total unit costs equals total costs divided
by total carbon
bottom portion represents the present value of the program cost over the defined period, 10-
19 years.3 The present value of program cost includes the cost of planting the trees and the
rental value of the land during the program period. Thus, program cost measures the
opportunity cost of the program, if it were ended after the program period. However, in all of
3 The length of the program period depends on how quickly the acreage is enrolled. If all of the
program acreage can be enrolled in a single year, then the program period is ten years. However, it mt
take as long as ten years (and 10 regeneration cohorts) to plant all of the acres, depending on the
assumptions made in each scenario. In the 10 cohort case, the program period is 19 years long, because
each cohort is staggered by a year. It is this feature of the programs that makes estimation of carbon
capture and cost somewhat complicated.
RCG/Hagler Bailly
-------
RESULTS »4-9
Figure 5
Total Carbon Sequestration for Combined Cropland
Scenarios with Regional Constraints
(1993-2035)
2.000
.2
1
1,500
o
3 i.ooo
B
I
t/i
i
500
2.4 3.7 8.7 18.7
Total Acreage for Combined Scenarios (million acres)
Figure 6
Total Carbon Sequestration for Combined Cropland
Scenarios with Enrollment Constraints
(1993-2035)
2,000
§ 1.500
O
5 1,000
o
1)
in
o
£>
a
U
500
2.4 3.7 8.7 18.7
Total Acreage for Combined Scenarios (million acres)
RCG/Haglcr Bailly
-------
RESULTS * 4-10
Figure 7
Net Present Value of Costs by Cropland Scenario
(1993-2035)
~ 12
C
o
10
en
in
O
U
v
"5
v
m
0)
z
B program costs
D oul-y««r co»ti
CKOPBASE CROP1R/E CROP5RC CROP9EC CKOP1ORC CROP10EC CROPIO
Cropland Scenario
the cases, it is assumed that the trees remain in the ground until 203 5.* The difference
between the height of each bar and the height of the bottom segment of each bar represents
the out-year cost of each scenario/program. This is the present value of the rental payments
required to keep the program acreage enrolled during the period after the program payments
expire and 2035. If landowners were wealth maximizers, then these additional payments
would be required to keep them from harvesting the trees before 2035.
Unit costs, which are shown at the bottom of Table 6, facilitate cost-effectiveness
comparisons. Because a number of different characteristics were varied across the scenarios,
these results will be somewhat inconclusive. However, we can derive the following insights
by comparing related scenarios. Consider the scenarios CROP1RC, CROP5RC and
CROP10RC, which are similar except for the amount of acreage planted. As acreage
increases, total unit cost initially declines from $10.84/ton for CROP1RC to $9.76/ton for
CROP5RC, and then rises to $9 92Aon for CROP10RC. Two factors are responsible for these
changes in unit costs. First, total unit cost initially fell because of the effects of discounting.
The CROP5RC program does not begin until 1996, so costs do not begin to accrue until then.
Effectively, those costs are discounted an additional three years from 1996 to 1993, which
makes the unit costs appear smaller. Second, unit costs rose between CROP5RC and
CROP10RC due to rising marginal costs within the regions.
4 Subject to the assumed drop-off enrollment, down to the 95% level in 2035 in each case.
RCG/Hagler Bailly
-------
RESULTS »-4-ii
In contrast, unit costs rose between CROP5EC and CROP10EC because the enrollment
constraint forced the regional mix of acreage toward higher cost regions as the number of
acres in the program increased. Table 7 shows the regional mixes for the scenarios. The
changes between the regional proportions in the existing CRP base and the CROPSEC
scenario are indicative of a shift in favor of acres in low-cost regions because CROPSEC used
a least-cost approach across regions. Unless the enrollment constraints are binding, the
distribution for CROP10EC should be similar to CROPSEC. Changes in the regional mix
between CROPSEC and CROPlOEC-particularly the decrease in the share of Delta States
acreage from 50% to 25%-suggest that the enrollment constraints are binding, and therefore
shift the regional mix toward more costly regions.
Table?
Conservation Reserve and Cropland Analysis
Regional Mix of Enrolled Acreage by Scenario
Cropland Scenario:
Region:
Northeast
Lake States
Combclt
Northern Plains
Appalachian
Southeast
Delta States
Southern Plains
Mountain
Pacific
Total
CROPBASE
CROP1RC
CROP1REC
CROP5RC
CROP10RC
0.6%
5.1%
3.7%
0.2%
6.2%
53.2%
29.5%
1.0%
0.2%
0.2%
100%
CROPSEC CROP10EC CROP10
21.7%
15.9%
22.7% 23.8%
50.5% 25.2% 39.5%
26.9% 13.4% 60.5%
100% 100% 100%
RCG/Hagler Bailly
-------
RESULTS > 4-12
We can compare the cost-effectiveness of different enrollment options and regional constraints
by comparing unit costs for scenarios that have the same level of acreage enrolled. CROP5EC
was more cost-effective than its counterpart, CROP5RC. There are two reasons for this. First,
the proportionality constraint for CROPSRC increased unit costs by forcing the regional mix
away from least-cost regions. Consider the regional mix for CROP10 in Table 5. CROP10
had the least constrained enrollment schedule and consequently the lowest unit cost. That
scenario's regional mix suggests that the least-cost region is the Southern Plains, followed by
the Delta States. The regional mix for CROPSEC includes a higher proportion of these
regions than CROPSRC does, which implies a lower unit cost. The second reason is that the
all of the acreage was planted in one year in CROPSEC. Enrollment took ten years in
CROPSRC. Earlier enrollment increase the amount of carbon sequestered, which helps reduce
unit costs.
Conversely, CROP10RC is more cost-effective than CROP10EC. This time, the differences in
regional mixes shown in Table 7 suggest that the enrollment constraints on CROP10EC
forced enrollment into higher cost regions compared to CROP10RC.
4.3 PASTURELAND RESULTS
Figure 8 shows incremental carbon sequestration over time for the seven pastureland
scenarios. PAS7 and PAS7EC are based on planting 7.S million acres of trees to sequester
carbon on pastureland. Incremental carbon capture for these two programs averages about 17
to 18 million tons/year, after all acres are enrolled. Scenarios PAS15 and PAS15EC both are
based on doubling this amount of acreage. Incremental carbon capture for these two programs
is about 35 million tons/year. The last three scenarios are based on planting 25 million acres
of pastureland in trees. Of these, only PAS2S and PAS2SECM could accomplish this
objective. Carbon capture for these two scenarios is about 55 million tons/year on average,
following complete enrollment. These estimates are consistent with slightly lower carbon
productivity on pastureland than on cropland.
Figure 9 shows how much carbon is stored by each of the seven pastureland scenarios from
1996 to 203S. These results are also shown in Table 8. Total carbon storage ranges from
about 0.5 billion tons for the 7.5 million acre scenarios, to about 2 billion tons for 25 million
acre scenarios. Total carbon sequestration differences between the scenarios within each of the
acreage target levels (7.5, 15, and 25 million acres) are relatively minor, with the exception of
the PAS25EC scenario, which was infeasible given the enrollment constraints. However, the
amounts of carbon captured by the years 2000 and 2010 indicate appreciable short-run
disparities. As with the cropland scenarios, these disparities are due to enrollment schedule
differences. At each acreage level, the E2 enrollment constraint resulted in a shorter
enrollment periods compared to the 10-year enrollment schedules for PAS7, PAS 15, and
PAS2S. Consequently, trees were planted more quickly in PAS7EC, PAS 1 SEC, and
PAS25ECM, and subsequently captured more carbon by 2000 and 2010, compared to their
counterparts
RCG/Hagler Bailly
-------
RESULTS » 4-13
Figure 8
Annual Carbon Sequestration by Pastureland Scenario
e- PAS?
PXS7EC
PiSIB
+- - PAS15EC
<3> PAS26
PAS25EC
PAS25ECU
1995 2000 2005 2010 2015 2020 20Z5 Z030 2035
Year
Figure 9
Total Carbon Sequestration by Pastureland Scenario
(1996-2035)
C
c
-*->
c
o
3400
zooo
1800
OJ
O)
cr
oo
c
o
1200
BOO
400
&
PAS7 PAS7EC PAS15 PAS15EC PJLS25 PXS25EC PAS25ECU
Pastureland Scenario
RCG/Hagler Bailly
-------
RESULTS > 4-14
TableS
Paitureland Analysis
Carbon Sequestration and Costa
Pastureland Scenario:
Carbon Sequestration
Total (1996-2035)
Total by 2000
(Annual Increment
Total by 2010
(Annual Increment
in
in
(million
2000)
2010)
PAS?
tons)
630
2.4
(1.0)
59.2
(12.0)
PAS7EC
640
14.3
(5.2)
132.3
(20.0)
PAS1S
1,230
3.6
(1.5)
104.7
(23.1)
PAS1SBC
1,260
16.0
(6.6)
211.0
(36.0)
PAS2S
1,940
5.3
(2.2)
159.9
(36.2)
PAS25EC PASZSECM
1,530
26.2
(11.6)
295.1
(44.1)
2,000
28.1
(13.4)
363.9
(55.3)
Cost Estimates (billion S)t
Program Cost
Out- Year Cost
Total Cost
SI.09
$0.88
$1.97
S2.20
$2.30
$4.50
$2.83
$2.43
$5.23
$4.11
$4.06
$8.17
$5.42
$4.58
$10.00
$5.35
$5.47
$10.82
$6.77
S6.42
$13.19
Unit Cost ($/ton)J
t
Program Unit Cost
Total Unit Cost
Program costs
include
$1.73
$3.13
$3.44
$7.03
$2.30
$4.25
$3.26
$6.48
$2.79
$5.15
federal funds for planting and any rental payments made
program; out-year costs represent
t
203S. total
costs equal
land rents
between the
$3.50
$7.07
to landowners
time a cohort leaves the program and
program costs plus out-year costs.
Program unit cost equals program
divided by
costs divided by total
carbon; total
unit costs
$3.39
$6.60
during
the
the year
equals total costs
total carbon.
The graph in Figure 10 displays the cost results for the pastureland cases. These costs are also
shown in Table 8. Figure 10 is identical in construction to Figure 7, which contained the
cropland scenarios' costs separated into program and out-year costs. Figure 10 shows that
PAS7 is much less costly than PAS7EC, although total carbon capture is almost identical.
Total unit costs, shown in the last row of Table 8, reveal that the unit cost for PAS7EC is
more than twice as high as PAS7. This outcome is due to the enrollment constraint for
PAS7EC. Table 9 shows the enrollment regional mixes for the pastureland scenarios. PAS7
enrolls acreage on a least-cost basis, with the constraint that no acreage in the Mountain and
Northern Plains regions is enrolled. Most of the acreage for this scenario is in the Southern
Plains region. The enrollment constraint (E2) for PAS7EC shifts the regional mix toward
costly regions such as Appalachia and the Combelt. The same is true for PAS 1 SEC, which
has higher unit costs than PAS 15, and PAS25ECM, which has higher unit costs than PAS25
Unit costs increase as program size increases for the PAS7, PAS 15, and PAS25 scenarios.
This trend indicates rising marginal costs. However, the scenarios with the E2 enrollment
constraints do not exhibit the same trend. Unit cost declines from $7.03 to $6.48 when the
acreage increases from 7 5 to 15 millions acres. The regional enrollment mixes in Table 9
RCG/Hagler Bailly
-------
RESULTS * 4-15
J3
CO
.!_>
o
U
c
01
OJ
BU
_j
0)
Figure 10
Net Present Value of Costs by Pastureland Scenario
(1993-2035)
~. 14
I 121-
10
B program coats
H out year costs
PAS7 PAS7EC
PAS15 PAS1 SEC FASZ5 PASZ5EC PASZSCCU
Pastureland Scenario
Table 9
PastureUnd Analysis
Regional Mix of Enrolled Acreage by Scenario
Scenario:
Region:
Northeast
Lake States
Combelt
Appalachian
Southeast
Delta States
Southern Plains
Pacific
Total
PAS7 PAS7EC
3.4%
7.0%
27.2%
21.3%
12.4%
18.9% 9.7%
81.1% 16.2%
2.7%
100% 100%
PAS15 PAS1SEC
5.1%
0.9% 6.9%
20.4%
22.3% 22.3%
16.6% 10.4%
19.8% 12.1%
40.5% 20.3%
2.6%
100% 100%
PAS25
10.4%
8.9%
32.0%
12.4%
11.9%
24.3%
100%
PAS25EC
7.2%
5.7%
28.8%
20.7%
7.7%
9.2%
17.1%
3.6%
100%
PAS25ECM
6.4%
7.0%
27.3%
21.4%
8.3%
9.7%
16.3%
3.5%
100%
RCG/Hagler Bailly
-------
RESULTS > 4-16
suggest that this decrease in unit cost can be due to a lower cost mix for PAS 1 SEC with
higher enrollment shares in the Southern Plains and Delta States than PAS7EC. Increasing the
program to 25 million acres shifts the regional mix back towards the PAS7EC mix, and
thereby increases unit costs.
4.4 BOTTOMLAND HARDWOOD RESULTS
Figure 11 shows incremental carbon sequestration for the four bottomland hardwood
scenarios. All of the scenarios sequester carbon at a rate of slightly more than 12 million
tons/year by the end of the study period. The two scenarios that have 5-year enrollment
periods, WETS and WET5U, sequester more carbon than the two that have 10-year
enrollment periods, WET10 and WET10U. One interesting feature is that the least-cost carbon
scenarios (WETS and WET 10) tend, initially, to have slower carbon sequestration rates than
the scenarios that have the constrained enrollment (i.e., uniform percentages within regions).
This happens because the tree growth rates for the regions that have low unit carbon costs
(i.e., the regions that would be planted first in a least-cost enrollment pattern) tend to be
lower than rates in the other regions shortly after planting, but then become relatively higher
in later years.
Figure 11
Annual Carbon Sequestration
by Bottomland Hardwood Scenario
IB
18
c
o
^ 14 I-
c
o
12 -
10
-o
91
Ix
0)
;
VI
o>
3
cr
0)
in
c
o
-0
u
03
U
Q
D
A
4-
WETS
WETSU
WET10
WET10U
1995 2000 2005 3010 2015 2020
Year
2025
2030
Z035
RCG/Hagler Bailly
-------
RESULTS » 4-17
Total carbon is pictured in Figure 12 and Table 10 reports total carbon through the year 2035.
All of the scenarios sequester between 460 and 500 million tons of carbon. Table 10 shows
the amounts of carbon captured by the years 2000 and 2010. Carbon sequestration for the
year 2000 demonstrates that the least-cost regions tend to start sequestering carbon at a slower
rate. However, by the year 2010, the least-cost enrollment scenarios have surpassed their
uniform enrollment counterparts, and overall sequestration is greater for the scenarios with the
least-cost enrollment patterns.
Table 10 also shows the costs for each scenario. These costs do not include any costs that
might be associated with converting wetlands; only the establishment and rental costs for tree
planting are reported as program costs. Total costs range from $4.19 to S4.26 billion for the
two five-year programs and from $3.61 to $3.73 for the two ten year programs.
Unit costs, reported in Table 10, show that the least-cost methods, WETS and WET 10, are
more cost-effective than the uniform planting methods, WET5U and WET10U. Although
scenarios WET5U and WETS plant the same number of acres per year, total costs for
WET5U exceed those of WETS by more than $70 million. Similarly, the costs of WET10U
exceed those of WET 10 by almost $120 million.
Figure 12
Total Carbon Sequestration
by Bottomland Hardwood Scenario
(1996-2035)
600
m
c
o
c
o
-a
01
u
OJ
en
0)
a
cr
-------
RESULTS » 4-18
Table 10
Bottomland Hardwood Analysis
Carbon Sequestration and Coits
Bottomland Hardwood Scenario:
WETS
WET5U
WET10
WET10U
Carbon Sequestration (million tons)
Total (1996-2035)
Total by 2000
(Annual Increment in 2000)
Total by 2010
(Annual Increment in 2010)
Cost Estimates (billion S)t
Program Cost
Out-Year Cost
Total Cost
Unit Costs (S/ton)J
497.4
7.4
(3.3)
99.0
(15.1)
$0.30
$3.89
$4.19
494.6
8.4
(3.6)
92.2
(13.2)
$0.33
S3.93
S4.26
468.1
3.7
(1.5)
69.4
(12.9)
$0.22
S3.39
$3.61
4624
4.2
(1.8)
65.9
(10.4)
$0.25
S3.48
$3.73
Program Unit Cost
Total Unit Cost
$0.60
S8.42
$0.67
$8.61
$0.47
S7.71
$0.54
$8.07
Program costs include federal funds for planting and any rental payments made to
landowners during the program; out-year costs represent land rents between the time a
cohort leaves the program and the year 2035; total costs equal program costs plus out-year
costs.
Program unit cost equals program costs divided by total carbon; total unit costs equals total
costs divided by total carbon.
Longer enrollment schedules reduce total costs because the program costs are distributed over
a longer period of time, which reduces their net present value and also reduces the initial size
of the out-year cost stream. Thus, both program costs and out-year costs appear smaller for
the WET10 and WET10U scenarios rejative to their five-year counterparts. The unit costs for
WET10 and WET10U are smaller than the unit costs for WETS and WET5U, respectively,
which indicates that the cost savings associated with the longer enrollment schedule has a
larger effect on unit costs than the accompanying reduction in carbon sequestration.
Program unit costs for this set of scenarios are lower than program unit costs for the TPI
scenarios, even though the TPI scenarios have the lowest total unit costs. Thus, if only
program costs are considered, then the bottomland hardwood scenarios are the most cost-
effective, followed by the TPI scenarios.
RCG/Hagler Bailly
-------
CHAPTER 5
SUMMARY AND MAJOR CONCLUSIONS
This chapter summarizes the carbon sequestration and cost results from Chapter 4 in Section
S.I. Section 5.2 provides conclusions about the effects on unit costs of different assumptions
regarding regional planting mixes, program sizes and enrollment patterns.
5.1 SUMMARY
Table 11 summarizes the results in Chapter 4 by showing the carbon sequestration and cost
ranges for each set of scenarios. Note that within a single set of scenarios, the values shown
in the table might represent several scenarios, so they will not necessarily be consistent along
the rows (e.g., the minimum total cost value of $2.45 for the timberland scenarios is from the
TPI70 scenario, while the minimum total cost value of S0.29 billion and minimum total
carbon sequestration value of 101 million tons are both from the TPI20 scenario).
Overall cost-effectiveness comparisons across different types of programs can be made using
the total unit cost values in the final row of Table 11. These ranges indicate that the
timberland scenarios tend to be the most cost-effective alternatives, followed by the
pastureland scenarios. The wetland and cropland scenarios tend to be fairly comparable in
terms of overall cost-effectiveness.
However, if we consider unit costs that are calculated based solely on program costs, (shown
in the second-to-last row) then the cost-effectiveness ordering shifts. Now the wetland
scenarios are the most cost-effective, followed by the timberland scenarios. This change in
cost-effectiveness rankings raises an important issue: changes in funding strategies can affect
an existing program's cost-effectiveness relative to other programs.
5.2 CONCLUSIONS
The scenarios were designed to evaluate the effects of various tree planting strategies on
carbon sequestration and costs. The following conclusions are drawn from the results
presented in Chapter 4.
RCG/Hagler Bailly
-------
SUMMARY AND MAJOR CONCLUSIONS » 5-2
Table 11
Summary of Carbon Sequestration and Costs
by Scenario Sets
Scenario Sets:
Carbon Sequestration (million tons)
Total (1993-2035)
Total by 2000
(Annual Increment in 2000)
Total by 2010
(Annual Increment in 2010)
Cost Estimates (billion S)
Program Cost
Out-Year Cost
Total Cost
Unit Costs (S/ton)
Program Unit Cost
Total Unit Cost
* Amount includes both Federal
Timberland
101-705
i-2
22-88
$0.16- $1.14
S0.13.S1.13
S0.29-S2.27
S1.34-S1.62
S2.45-S3.22
(50%) and landowner
Cropland
131-1,028
3-25
36-229
S0.63-S4.60
S0.79-S6.26
S1.42-S10.86
S3.41-S4.81
$7.68-$ 10.84
costs (50%).
Pastureland
630-2,000
2-28
59-364
S1.09-S6.77
S0.88-S6.42
S1.97-S13.19
S1.73-S3.50
S3.13-S7.07
Wetland
462-497
4-8
66-99
$0.22-S0.33
$3.3943.93
S3.61-S4.26
S0.47-S0.67
S7.71-S8.61
5.2.1 Regional Shares
The regional distribution of enrolled acreage is an important determinant of unit costs. This
was most evident in the cropland scenarios where least-cost regional distributions were
compared with the existing regional distribution of CRP tree acres. Currently, the majority of
CRP tree acreage is located in the Southeast region (53%) and the Delta States region (29%).
The analysis of the cropland scenarios showed that a regional mix with 60% of the acreage in
the Southern Plains region and 40% in the Delta States region is more cost-effective than one
that replicates the current distribution. Comparing CROP10 and CROP10RC, which were
similar except that CROP 10 had a cost-effective regional distribution of acreage, we see that
CROP10 had total unit costs that were 23% lower than unit costs for CROP10RC.
A comparable regional mix was also cost-effective for the pastureland scenarios. The most
cost-effective scenario that enrolls 7 S million acres has 81% of that acreage in the Southern
Plains region and 19% in the Delta States region. The larger programs, i.e., IS million acres
and 25 million acres, had less cost-effective regional mixes because they exhausted
enrollment opportunities in the least-cost regions, and subsequently enrolled acres in higher-
cost regions.
RCG/Hagler Bailly
-------
SUMMARY AND MAJOR CONCLUSIONS > 5-3
Results for the timberland scenarios indicated that the least-cost acreage was in the Combelt
and Northern Plains regions, followed by the Pacific and Lake States regions. Each of the
wetland scenarios planted the same 4.9 million acres, so we did not test the effects of regional
mix on cost-effectiveness.
5.2.2 Annual Carbon Increments
Incremental carbon sequestration appears to peak sometime around the year 20IS for all of
the scenarios. Annual carbon sequestration trends for most of the scenarios do not enter a
phase of rapid decline following this peak. In fact, annual carbon increments for most of the
timberland, pastureland, and wetland scenarios remained at or above their 2010 levels through
the year 2035.
There are two primary reasons for this outcome. First, the analysis period from 1993 through
203S is shorter than the high growth intervals for some of the tree species under
consideration. For example, annual increments for some of the tree species in the TPI
scenarios still exceed their respective average annual increments by the year 203 5, so their
growth rates are consistently high throughout the period. Second, the annual increment curves
are moderated by the staggered enrollments of some scenarios because the curves represent
weighted averages over ten years of growth increments, and consequently have smoother
characteristics such as smaller peaks and more gradual declines.
5.2.3 Increasing Unit Costs
For some of the scenario sets, the total unit cost results do not seem to indicate appreciable
increase in total unit costs as program size increases. In the timberland scenarios, total unit
cost increases from $2.95 for TPI20CS to $3.22 for TPI70CS, which is small relative to the
magnitude of program increase. TPI70CS is roughly three times larger than TPI20CS, yet
total unit cost increases by 9%. This outcome is interesting from a policy perspective because
it contrasts the potential results of the tree planting program had it been funded per recent
USDA requests of $70 million per year with the predicted results from the tree planting
program as it was actually funded by Congress ($20 million).1
The cropland results suggest that an increase of 1.3 million acres and a much larger increase
of 10 million acres have comparable total unit costs, if enrollment is constrained either to
match the current regional mix or to limit the proportion of enrolled regional acreage.
1 From a policy perspective, it is also interesting to note that annual carbon sequestration rates for
the years 2000 and 2010 were more than three times greater in the TPI70CS scenario than they were in the
TPI20CS scenario.
RCG/Hagler Bailly
-------
SUMMARY AND MAJOR CONCLUSIONS 5-4
Furthermore, a least-cost approach to enrolling 10 million acres has a total unit cost that is
30% lower than either of the 1.3 million acre enrollment scenarios.
Compared to the timberland and cropland scenarios, the pastureland scenarios demonstrate a
larger degree of increasing unit costs. As enrolled acreage doubled from 7.5 million acres to
15 million acres, total unit costs rose by about 36%. Similarly, a 67% increase in program
size from 15 million acres to 25 million acres produced a 21% increase in total unit cost.
There were no program size differentials in the wetland scenarios, so they cannot be used to
address this issue.
5.2.4 Enrollment Timing
In all scenarios, short enrollment schedules increase both the total amount of carbon
sequestered, and the amount of carbon sequestered by the years 2000 and 2010. The effect of
staggered enrollments are especially noticeable for the sequestration totals for the years 2000
and 2010, reducing these totals by a factor of 5 in some instances.
The net effect on total unit costs depends on which of the following conflicting effects is
stronger:
» early enrollment increases the total amount of carbon sequestered, which
reduces unit cost
» early enrollment increases the present value of costs, which increases unit cost.
The results regarding the net effect of early enrollment on unit costs are somewhat
inconclusive, but tend to suggest that the increase in present value of costs is stronger so unit
costs rise. This was seen in the wetland scenarios, where the only difference between WETS
and WET10 (or between WET5U and WET10U) is the length of the enrollment period. The
ten-year enrollment period was more cost-effective even though total carbon sequestration was
lower.
In the cropland and pastureland scenarios, the enrollment timing effect cannot be adequately
isolated from the enrollment constraint effect because the two occur simultaneously. In the
timberland scenarios, program size and payment structure change simultaneously with
enrollment timing, so the tatter's effect cannot be isolated.
RCG/Hagler Bailly
-------
SUMMARY AND MAJOR CONCLUSIONS » 5-5
5.2.5 Funding Mechanism
Finally, we note that absent information to the contrary, programs that cover the full cost of
sequestering carbon, including out-year land rental costs, will be more effective in
maintaining participation levels beyond the enrollment period than those which depend on the
good will of landowners to refrain from harvesting the trees on their land. This assumption is
based on economic logic rather than study results. The way costs were structured in this
analysis further reflect that logic. There are two caveats to this final statement. First, to the
extent that there is good evidence that private nonindustrial landowners incorporate nonmarket
values into their decision to manage timberland, they will not need to be compensated as fully
for holding trees as would an owner with strictly wealth maximizing objectives.2 Second, the
current SIP program is highly effective, even though the payments cover only pan of the
landowner's full opportunity cost. However, as the number of acres in a cost-sharing program
is expanded considerably, it is possible that quality of acres at the margin will decline. This
would happen if participants enroll their more productive lands first because they expect that
future returns on these lands will be high enough to offset their share of the costs. As more
acres are enrolled, the quality would decline until participants no longer expect future returns
to cover their share of the costs. At that point, the amount of money the government must pay
to induce farmers to participate would need to increase. We do not know at what point this
would become a issue. The Forest Service estimate that there are about 70 million acres of
nonindustrial private timberland in the United States with investment opportunities that would
yield at least 4% (USFS 1990). The relative small size of the TPI scenarios considered here
would suggest that they would not be affected by this phenomenon.
2 A recent study by Newman and Wear (1993) demonstrates this fact and quantifies this effect for
nonindustrial private timberland owners. These results have not been incorporated in this analysis.
RCG/Hagler Bailly
-------
CHAPTER 6
REFERENCES
Moulton, Richard and Kenneth Richards. 1990. Costs of Sequestering Carbon Through Tree
Planting and Forest Management in the United States. USDA Forest Service, GTR WO-58.
Washington, DC.
Newman, David and David Wear. 1993. "Production Economics of Private Forestry: A
Comparison of Industrial and NONindustrial Forest Owners." AJAE 75(3): 674-684.
Richards, Kenneth. 1992. "Derivation of Carbon Yield Figures For Forestry Sequestration
Analysis. Draft Paper Prepared for Office of Economic Analysis, U.S. Department of Energy.
January 8, 1992.
USDA Forest Service. 1990. An Analysis of the Timber Situation in the United States. 1989-
2040. General Technical Report, RM-199. USDA Forest Service, Rocky Mt. Forest and
Range Experiment Station, Ft. Collins, CO.
RCG/Hagler Bailly
------- |