Farmer Transaction Costs of Participating in Federal Conservation Programs: Magnitudes and Determinants

I. INTRODUCTION

Continuing water quality problems associated with non-point-source pollution from agricultural production imply that increased adoption of best management practices (BMPs) is needed. Governments have typically relied on voluntary BMP adoption by farmers to address the issue and have created programs that provide both technical and financial support. The USDA offers this support through programs such as the Conservation Reserve Program (CRP), the Environmental Quality Incentive Program (EQIP), and the Conservation Stewardship Program (CSP).

The fundamental nature of non-pointsource pollution—including the difficulty of measuring emissions, time lags, the large number of actors involved, and heterogeneity of landscapes—implies that transaction costs associated with efforts to address the problem will be relatively high (McCann 2013). Garrick, McCann, and Pannell (2013) document the steadily increasing interest by the profession in transaction costs associated with environmental policy from a low of 1 publication in 1990 to over 50 per year from 2007 to 2012. Previous studies in Europe have found the magnitudes of farmer transaction costs of participating in agri-environmental schemes to be substantial. To our knowledge, no studies have measured the transaction costs of farmers applying for conservation programs in the United States or examined whether transaction costs might affect participation in those programs.

This research consists of three components: (1) determine whether perceived high transaction costs were an important barrier to applying for conservation programs by asking nonapplicants about the reasons for nonparticipation, (2) estimate the magnitudes of ex ante and ex post transaction costs for producers who have applied for participation in government conservation programs, and (3) analyze the farm and farmer determinants of these costs. The main focus of this work is thus on transaction costs rather than the analysis of reasons for participation or nonparticipation in these programs, which is an important issue.

Analysis of barriers to conservation program participation, including farmer transaction costs, could aid in the redesign of programs and program application processes and/ or improve producer outreach at the federal, state, and local levels. Data on producer perceptions and transaction costs would also enhance research on conservation practice adoption, conservation program participation, and additionality in conservation programs. For example, including data on perceived barriers to conservation program participation in an analysis of these programs could show how these perceptions affect the likelihood of program application. These data may thus lead to more specific insights on program design and implementation. This may increase the likelihood that land with higher environmental benefits is enrolled. Estimates of transaction costs of participants could indicate which programs and which aspects of the process are most costly and whether there is a discrepancy between perceived costs and measured costs.

Researchers have identified a number of factors that affect adoption of conservation practices, as well as participation in government programs. One factor typically assumed in economic studies is whether adopting or participating increases profits; therefore, the research question often relates to the optimal level of compensation. Falconer (2000) discussed a study of farmers in several European countries that showed that one-third of nonparticipating farmers said the compensation was too low. However, compensation is not the only relevant factor. Ma et al. (2012) hypothesized a two-stage decision process and showed that farm and farmer characteristics affected the willingness to even consider environmental programs, but the participation decision related to characteristics of the programs such as payment rates. Mishra and Khanal (2013) found that participation in both CRP and EQIP was negatively affected by higher debt-to-asset ratios. Size of farm is typically associated with increased likelihood of participation (e.g., Mishra and Khanal 2013; Lambert et al. 2007) and adoption of more management-intensive practices (Lambert et al. 2007).

Sheeder and Lynne (2011) found that, while economic factors played a role, adoption of conservation tillage was also affected by other-regarding factors such as empathy with downstream water users. A fairly large literature finds that environmental attitudes affect adoption (see reviews by Prokopy et al. 2008 and Pannell et al. 2006). Ma et al. (2012) found that the expected off-farm environmental performance had a positive effect on both consideration of government programs and likelihood of participation. Reimer, Gramig, and Prokopy (2013) examined variation in state program application rates and found it was positively related to the number of impaired streams in the state. Sociopolitical attitudes may also affect participation. Kraft, Lant, and Gillman (1996) examined the Water Quality Incentives Program, the predecessor to EQIP, and found that attitudes toward government wetland policy affected participation. Studies often find that previous participation affects future participation (Ma et al. 2012), which may be related to environmental values or a latent variable relating to attitudes toward interacting with government agencies. Lambert et al. (2007) found that many practices were adopted by farmers who did not participate in government programs. They hypothesize that the practices may increase profits in the short or long term, but it may also be that farmers with environmental or other-regarding preferences may not want to be involved with government conservation programs.

Demographic factors such as age and education are often found to significantly affect adoption of practices and participation in programs, but the effects may vary. For example, Mishra and Khanal (2013) found that while education increased adoption of both EQIP and CRP, increasing age increased CRP participation. A wide range of barriers to participation in conservation programs have thus been identified in the literature.

One potential barrier to participation that has not been directly examined in the U.S. context is the magnitude of transaction costs, real or perceived, of applying for and complying with government programs. In related research, Reimer and Prokopy (2014), in a study of Indiana farmers, found that complexity of the U.S. conservation program system may limit participation. They also found that while knowledge of CRP was fairly high, knowledge of EQIP was quite low. Falconer (2000) found that 21% of European farmers said the application was too costly and 49% said they didn’t know enough about the schemes, both of which relate to transaction costs, rather than abatement costs. Learning about conservation schemes or programs would represent an information cost. Buckley and Chapman (1997), examining firm management decisions, found that perceived transaction costs were used in decision-making and that little effort was made to quantify these types of costs, contrary to changes in production costs or revenues.

A definition of transaction costs is needed to put our research on the measurement of these costs in context. For typical market transactions, a definition such as that by Demsetz (1967) is appropriate: transaction costs are defined as “the cost of exchanging ownership titles.” With respect to environmental and natural resource issues, a more broad definition is needed since both property rights and the “good” are usually not well defined. McCann et al. (2005) define transaction costs for these contexts as “the resources used to define, establish, maintain, and transfer property rights.” In addition, typologies of transaction costs have been developed to facilitate transaction cost analysis and measurement. These may include research and information, enactment or litigation, design and implementation, support and administration, contracting, monitoring/detection, and prosecution/ enforcement (McCann et al. 2005).

Conservation program transaction costs are borne by government agencies and farmers. A small (but growing) empirical literature shows that transaction costs borne by program agencies can be large (McCann and Easter 1999, 2000; Falconer, Dupraz, and Whitby 2001). Studies in the United States and Great Britain indicate that these costs, which include conservation planning and technical assistance, may be over 30% of the total cost of conservation programs.

Measuring farmer transaction costs is more recent and the magnitudes are lower, on the order of 15% to 20%, and appear to vary greatly by type of program (e.g., European programs that assist with conversion to organic farming tend to have higher transaction costs than those that pay for reduced tillage) (Falconer 2000; Rorstad, Vatn, and Kvakkestad 2007; Mettepenningen, Verspecht, and Van Huylenbroeck 2009). Mettepenningen, Verspecht, and Van Huylenbroeck (2009) indicated that farmer transaction costs, measured using weekly information registration procedures, were 15% of total implementation costs but did not include transaction costs associated with application activities. To date, efforts to measure farmer transaction costs have been confined to Europe, where a change in regulations allows countries to compensate farmers for these costs. To our knowledge, this effort is the first attempt to measure farmer transaction costs for government-sponsored conservation programs in the United States.

II. LITERATURE REVIEW

Following Coase (1937) and Williamson (1985), there is a large literature on how transaction costs affect the behavior of firms. Characteristics of the transaction affect transaction costs and thus whether firms will produce a needed item themselves, contract with another firm to produce it, or buy it on the spot market. Williamson (1985) suggests that three characteristics are crucial: asset specificity, frequency, and uncertainty. Asset specificity relates to whether an investment in physical or human assets is associated with one or a limited set of trading partners, which will generally increase transaction costs compared to the case where the same asset could be used with many other trading partners. Higher frequency of transacting allows the development of routine procedures, thus decreasing transaction costs. Rorstad, Vatn, and Kvakkestad (2007) indicate that for agri-environmental programs, these two characteristics are unlikely to be correlated, that is, transactions with high asset specificity tend to be infrequent. Williamson’s last characteristic relates to uncertainty of the behavior of trading partners (such as opportunism), as well as price or physical uncertainty, all of which would tend to increase transaction costs. For infrequent transactions that involve high asset specificity and uncertainty, one is likely to observe either contracting or hierarchy rather than spot market transactions (Williamson 1985). Rorstad, Vatn, and Kvakkestad (2007) and Coggan et al. (2013) indicate that agrienvironmental issues tend to have these three characteristics, in part because of the degree of heterogeneity across farms and landscapes. It is thus not surprising that farmers and government agencies contract for the provision of ecosystem services such as wildlife habitat or improved water quality. Farmers want to ensure that their specific investments, which would have few alternative buyers, will be compensated. Government agencies want to ensure that the public is receiving the environmental benefits they paid for. In addition, measurability is a salient characteristic of agri-environmental issues and fundamentally affects the potential for contracting for environmental services (Bougherara, Grolleau, and Mzoughi 2009). In both Europe and the United States, farmers are usually paid for installation of specific practices rather than for environmental performance.

There are thus a number of reasons to suspect that transaction costs involved with addressing non-point-source pollution issues will be substantial, and this has generally been supported by the empirical studies that have been conducted (McCann and Easter 1999, 2000; Falconer, Dupraz, and Whitby 2001; Mettepenningen, Verspecht, and Van Huylen-broeck 2009; Rorstad, Vatn, and Kvakkestad 2007; Vernimmen, Verbeke, and Van Huylen-broeck 2000). Studies have typically shown that farmer transaction costs are lower than those borne by government agencies. The likely high transaction costs of point-nonpoint source water trading policies are perceived to be a barrier for these programs. In one case, the transaction costs eliminated any gains from trade for a point-nonpoint source trade in Minnesota (Fang, Easter, and Brezonik 2005). However, Ribaudo and McCann (2012) found that other aspects of the design of the trading program in Pennsylvania were likely more limiting than the transaction costs.

McCann (2013) surveys the literature on factors affecting transaction costs of environmental and natural resource policies in order to develop design recommendations. As mentioned above, asset specificity due to heterogeneity may lead to contracting as the form of governance for agri-environmental issues, but transaction costs are still likely to be high. For farmers, they may contract only once every few years, and thus there is little scope for learning by doing, which has been shown to decrease agency transaction costs (Falconer, Dupraz, and Whitby 2001). Intermediaries may be able to lower some of these costs, particularly if transactions are infrequent and complex (a simple example is buying or selling a house using a real estate agent). McCann (2009) found that few farmers prepared comprehensive nutrient management plans themselves; most were prepared by Natural Resource Conservation Service (NRCS) staff or technical service providers. Vernimmen, Verbeke, and Van Huylenbroeck (2000) found that farmers were more likely to outsource complicated tasks. Complexity of the program design, application procedures, or environmental criteria used to evaluate bids would thus be likely to increase transaction costs. Nevertheless, farmers who have more experience with conservation programs would be expected to have lower transaction costs of applying for new programs, although the extent to which this is true will depend on the similarity of the past and current programs. If there are a larger number of operators, this would imply that some could specialize in tasks such as applying for conservation programs, and this would be expected to reduce transaction costs. An alternative hypothesis would be that these farms would have more capacity and thus may be less time-constrained and would spend more time applying for programs.

Transaction costs are typically positively correlated with abatement costs or the magnitude of the change involved (Garrick and Aylward 2012; Krutilla and Krause 2011; McCann and Hafdahl 2007; Rorstad, Vatn, and Kvakkestad 2007). In Europe, some environmental practices are mandated, and contracting in agri-environmental schemes (AESs) is for practices that go beyond these minimum standards (Mettepenningen, Verspecht, and Van Huylenbroeck 2009), so one would expect higher transaction costs than for programs that are entirely voluntary and thus may involve some practices that are low cost or even win-win. In the case of USDA programs, we would expect that the transaction costs of programs involving a higher level of environmental performance, such as CSP, would have higher transaction costs.

More complex farming systems, such as farms with both crops and livestock, may have higher transaction costs. Ducos, Dupraz, and Bonnieux (2009) hypothesized that on farms with more animals, transaction costs would be higher, and thus there would be lower AES participation rates, but they found no significant effect. More complex landscapes involving hilly rather than flat terrain, such as farms with land designated as highly erodible, would also be expected to have higher transaction costs. We note, however, that most highly erodible land is subject to highly erodible land conservation requirements that tie eligibility for many USDA programs to soil conservation on highly erodible cropland. So, much of the conservation planning required on highly erodible cropland has already been done.

The adoption literature finds that farmers with larger operations are more likely to adopt new technologies and also more likely to participate in conservation programs, partly because any fixed costs are spread over a larger output. A number of studies have found that there are fixed transaction costs involved with agri-environmental programs, and thus there are economies of scale related to these costs in addition to the economies of scale involved in production and abatement (Ducos, Dupraz, and Bonnieux 2009; Falconer 2000; McCann 2009). This may partially explain the lower participation rates of small farmers in Europe (Ducos, Dupraz, and Bonnieux 2009; Falconer 2000) and the United States (Lambert et al. 2007; Mishra and Khanal 2013). Value of production would be expected to slightly increase the magnitude of transaction costs but lower the per acre transaction costs.

The adoption literature often finds that new farming practices are more likely to be adopted by farmers with higher education levels. This is also typically found for conservation practice adoption and participation in government programs (Prokopy et al. 2008). We would expect that higher education levels would be associated with participation in more complex programs but could be associated with higher or lower transaction costs. Farmers with more education may be able to understand and fulfill application requirements more quickly than other producers. Given that USDA conservation program applications are typically ranked using complex benefit-cost indices, more highly educated farmers may spend more time on conservation program applications because they perceive a comparative advantage in developing applications that meet their needs and have high probability of being accepted. Retired farmers, who may have extensive experience in dealing with government programs, may also perceive a similar comparative advantage and may have the flexibility to spend time considering various participation alternatives. In many studies, farmer age is correlated with lower likelihood of adoption of environmental practices or participation in government programs (e.g., Prokopy et al. 2008). For those who do participate, the effect of age on transaction costs is ambiguous since experience with other programs may lower transaction costs but, similar to those with more education, they may realize that the programs are competitive and that application quality will be a factor in application acceptance.

III. BACKGROUND

The USDA offers a broad suite of voluntary payment programs to help farmers address conservation and environmental issues in agricultural production. Because our data are derived from a survey of soybean producers relating to a specific field planted to soybeans, we consider only relevant programs. EQIP can support a wide range of practices, applied narrowly within a single field or throughout the farm. CRP, through continuous signup for high priority practices, supports a subset of these practices including grass waterways, field-edge filter strips, or other “partial-field” practices that take very little land out of production but are typically used in or adjacent to fields in crop production. As with EQIP, CRP-funded practices can be applied to all or only a small part of the farm.

CSP can support a broad range of practices but, unlike EQIP and CRP, requires participants to (1) achieve a minimum level of conservation practice adoption before enrolling, (2) enroll all eligible land in the entire farm (most cropland, pasture, range, and forest land),¹ and (3) agree to further improve environmental performance by adopting additional practices over the five-year life of the contract (which can be extended for another five years). In exchange, farmers can receive payments that support ongoing conservation effort (not available from any other USDA conservation program) as well as payments for new practices. Unlike other programs, CSP payments can exceed the cost of installing, adopting, or maintaining practices.

The whole-farm approach and stewardship requirements embodied in CSP are likely to increase transaction costs because CSP applicants must provide extensive documentation of land use and land management practices throughout their farm, in addition to the contract, conservation plan, and related forms required by EQIP and other USDA-NRCS conservation programs.² CSP requires farmers to provide documentation of land use (a page-long form for each land use), general agricultural practices (one page), and specific practices on each type of land. On cropland, that includes documenting crop rotations, tillage, and other residue management practices; measures to reduce soil compaction; and a wide range of individual nutrient management, pest management, and irrigation management practices, for each crop rotation used on the farm. The documentation form for cropland alone is five pages long and requires farmers to respond to 70 different questions and subquestions. Farmers who also have land in other uses would be required to provide many more pages documenting practices on pasture land (six pages), rangeland (two pages), and forest land (four pages). Overall, farmers with several types of land could be required to fill in as many as 20 pages of forms.

IV. DATA

Data on farmer transaction costs are from the field-level (phase 2) portion of the 2012 USDA Agricultural Resources Management Survey (ARMS).³ The ARMS is a national survey, administered every year by the National Agricultural Statistics Service and other USDA agencies, including a farm-level and a crop-specific field-level survey, which focused on soybean production in 2012.⁴ Survey respondents were asked to provide extensive information on production practices, conservation practices, and conservation program participation for a specific field selected at random from fields that were planted to soybeans in 2012. A total of 3,555 farmers in 19 states that account for more than 90% of soybean production were selected for the survey, and 2,492 provided usable responses. Farmlevel and demographic data are from the ARMS phase 3 follow-on survey of each individual who responded to the phase 2 survey. A total of 1,807 farmers provided usable responses to both surveys.

ARMS respondents who were not enrolled in a conservation program (or had not applied for enrollment during the past four years) were asked about perceived barriers to participation. Given response options of agree, neutral, and disagree, survey participants were asked about the following factors: (1) lack of awareness of programs, (2) lack of awareness of environmental problems on the field, (3) payments being too low, (4) government standards being more expensive than necessary to solve the problem, (5) perception that the application would not be accepted, (6) the application process being too complex or timeconsuming, and (7) documentation of compliance being too complex. Items 6 and 7 are related to transaction costs.⁵

To examine reasons for nonparticipation, we first eliminated respondents who indicated the same response to all questions, since these were not viewed as credible, and were left with 1,010 observations (Table 1). Responding to the questions in that way may indicate that they do not care about conservation programs in general. The most common “agree” response (after lack of awareness of a problem on the field, 63%), was that government practice standards were more expensive than necessary to solve the problem (34% agreement), followed by documentation of compliance (31%) and a complex application process (29%). These latter options represent perceived transaction costs so these are a relatively important barrier. The application process seemed to be more of a barrier among U.S. soybean farmers compared to European farmers; Falconer (2000) reported that only 21% said that the application was too costly. Less important barriers were thinking the application wouldn’t be accepted (23%) and the payments being too low (20%). The latter result is somewhat surprising since this is often suggested as a solution to low participation, and Falconer (2000) found about one-third of European farmers gave this reason. Only 15% of respondents agreed with the statement “I was not aware of conservation programs.” However, farmers may not be aware of the full range of programs available to them, or the characteristics of those programs. Reimer and Prokopy (2014) found there was little knowledge among Indiana farmers of conservation programs available, other than CRP. McCann and Núñez (2005) found that only 53% of Iowa and Missouri farmers were aware of EQIP. Falconer (2000) found that 49% of European farmers said they didn’t apply because they didn’t know enough about the programs. Being aware there are programs, and knowing enough about them to be interested in applying are different questions, but both relate to information costs and thus point to transaction costs as a potential barrier.

Using two-tailed tests (data not shown), we found farmers with highly erodible land were more likely than other farmers to disagree that the application and documentation processes were too complex (21% vs. 12%, and 25% vs. 10% of farmers, respectively, p < 0.10). Because these farmers are subject to conservation compliance requirements that require the application of soil conservation plans on highly erodible land as a condition of farm program participation, they are very likely to have experience documenting compliance with a conservation plan even if they have not participated in a conservation program like EQIP or CSP. Farmers who had past experience with conservation programs were significantly less likely to agree (19% vs. 30%) and more likely to disagree (27% vs. 12%) that the application process is too complex compared to those without experience. Perceptions of transaction costs thus appear to be more of a barrier than actual transaction costs, which is in line with Falconer’s (2000) and Buckley and Chapman’s (1997) results. The presence of a disparity between perceived and actual transaction costs may itself be an indication of the transaction costs associated with information gathering. Commercial-size farms were significantly less likely than smaller farms to agree that documentation of compliance was a barrier (29% vs. 38%). It should be noted that the list of potential barriers is not exhaustive and in particular does not include environmental attitudes, other-regarding preferences, or aversion to government programs. We also do not have data on the extent to which these farmers are knowledgeable about the programs.

ARMS respondents were also asked whether the surveyed field was included in a current conservation program contract or had been included in an unsuccessful conservation program application during 2009–2012 (the period covered by the 2008 Farm Act), and which program they were participating in or had applied to. Choices included EQIP, CSP (or its predecessor, the Conservation Security Program, 2004–2007), CRP, or other programs.⁶ A total of 149 respondents indicated that the surveyed field was currently enrolled in a conservation program, while 20 indicated that they had applied but had not been accepted into a conservation program during the previous four years. These 169 respondents were asked to report the number of hours spent on tasks typically involved in conservation program applications. Based on the literature (particularly Mettepenningen et al. 2007) and consultation with NRCS staff, questions were included to capture hours spent (1) learning about conservation programs, (2) planning conservation activities (to develop specific proposals need for the application), (3) collecting documents, (4) filling out forms, and, if accepted for participation, (5) understanding and signing the contract and (6) documenting compliance. Other studies of farmer transaction costs have also used time spent as a measure of transaction costs (Mettepenningen, Verspecht, and Van Huylen-broeck 2009; McCann 2009).

Table 2 provides descriptive statistics for ex ante (before application acceptance) and ex post time spent, with CSP separated from EQIP, CRP, and other programs because CSP requires broader documentation of existing conservation practices and may require broader conservation treatment. We define ex ante transaction costs as the sum of learning about programs, planning conservation, collecting documents, and filling out forms. Ex post transaction costs are the sum of understanding/signing the contract and documenting compliance. On average, CSP applicants spent more than 20 hours on ex ante tasks and almost 8 hours on ex post tasks. In contrast, applicants for other programs spent only 8 hours on ex ante and less than 2 hours on ex post tasks, on average. At the median, CSP hours are 11 and 3 for ex ante and ex post tasks, respectively, while non-CSP hours are 6 and 2 for ex ante and ex post tasks, respectively. Pair-wise t-tests confirm that average time spent on both ex ante and ex post tasks is significantly higher for CSP applicants/participants than for EQIP, CRP, and other programs, which did not differ significantly among each other. Therefore, in subsequent analyses of determinants, these programs are combined. The fact that many non-CSP applicants spent very little time planning or designing (median = 1 hour) and collecting information (median = 0 hours) suggests that many applicants are seeking payments to help defray the cost of conservation plans that have already been developed.

The magnitudes estimated for U.S. conservation program applicants/participants are much lower than the magnitudes found for European farmers applying for AESs. Mettepenningen, Verspecht, and Van Huylenbroeck (2009) found ex ante costs of 7.2 days for information gathering, 7.3 days for field maps and soil samples, 3.3 days for consulting with the agency, and 2.6 days for filling out the application form. They found that the standard deviations of transaction costs were very high. When the survey was conducted, few farmers knew that they might be eligible for payments to compensate them for transaction costs,⁷ so the result does not seem to relate to strategic behavior. One difference is that the ARMS question asked for hours spent, while the European survey asked for days spent, and responses may have included partial days. The European study also asked for transaction costs relating to the whole farm, rather than to applications relating to one field/parcel. The U.S. survey also required recall of time spent up to four years in the past, which may have reduced the time spent estimates. It is also the case that the European programs are more analogous to CSP since compliance with mandatory practices is required, and for CSP, farmers can enroll only after a minimum level of conservation effort has been achieved.

Broad variability in time spent across farms suggests that transaction costs may also vary with farm characteristics, the demographic characteristics of the farmer, and the practices to be installed or adopted. Unfortunately, the ARMS data are not directly linked to current conservation program contracts, so information on specific practices funded is unavailable. We do, however, have information on a selected set of conservation practices used in the field and information on the farm and farmer. An additional source of variation may be the amount of assistance provided by NRCS staff in individual counties, or the services provided by consultants, but this information was not available in the ARMS data.

V. MODEL OF TRANSACTION COST DETERMINANTS

The list of explanatory variables and the descriptive statistics for non-CSP and CSP programs are found in Table 3. Human capital available to develop conservation program applications is measured by the level of producer education, producer age (which is also highly correlated with experience), whether the survey respondent is active or retired from farming, and the number of operators on the farm. Producer education is the approximate number of years of education obtained by the primary operator.⁸ Farmers applying to CSP are more highly educated (14.3 years), on average, than farmers applying to the other programs (13.6 years). CSP applicants were slightly younger than applicants to other programs (54 years versus 56.4 years). The respondents’ active or retired status is described using a binary variable that equals 1 if the respondent retired from active farming (although these individuals are still involved in farm management, share production risk, and may pay some production expenses). Farmers applying to CSP were slightly less likely than others to indicate that they were retired from farming (10% vs. 14%) and report a larger number of operators (an average of 1.8% vs. 1.5% for farms that applied to other programs).

To account for farm size and complexity we use the value of total agricultural production, total farmland acreage, and the proportion of value derived from livestock. The total value of production from crops and livestock is developed for Economic Research Service (ERS) farm income estimates and are based on producer responses regarding crop and livestock production in the farm-level portion of the ARMS survey. The value of production is higher for farmers who applied for CSP, $1.05 million versus $0.76 million with large variability in both cases. Those applying for CSP had a slightly higher proportion of their income from livestock, which would indicate a more complex farming system. CSP applicants also operated larger acreages, averaging 1,890 acres per farm versus 1,210 acres for non-CSP program applicants. A binary indicator of highly erodible land is also included to capture the fact that conservation on these acres may be complicated by steep slopes and that conservation compliance requirements may apply and thus could affect producer eligibility for conservation programs.⁹ A lower proportion of farmers applying for CSP had highly erodible land (19% vs. 26%).

Because CSP requires previous conservation action, we define three binary indicators of previous conservation performance that serve as a measure of early stewardship. Farmers who had a written soil conservation plan, a written comprehensive nutrient management plan, or an integrated pest management plan by 2004 (preceding the original CSP, which held its first signup late in 2004) are more likely to be eligible for CSP and may have had some advantage in competing for enrollment.¹⁰ Although some farmers may have adopted/installed practices after that time, farmers who indicated participation in CSP may have adopted practices after CSP enrollment, as the CSP enrollment date is not known and a contract can last for up to 10 years with the optional five-year extension. Therefore, practices that were in place before CSP began enrolling participants can be viewed as an indicator of these farmers’ underlying stewardship ethic (Chouinard et al. 2008). Surprisingly, farmers applying to CSP were less likely to have a soil conservation plan, perhaps due to a lower percentage having highly erodible land. (A discussion of related regression results is found in the next section.) On the other hand, as expected, a larger proportion of CSP applicants had a comprehensive nutrient management plan or an integrated pest management plan. We note that formal, written plans are not required for entry into the CSP, even though farmers who do have these plans appear to be more likely to be accepted as participants—perhaps serving as a signal that conveys information about the producer’s conservation effort. Farmers who manage soil and nutrients effectively and can document their practices (as part of the CSP application) can meet the stewardship threshold. To control for regional variation, we have included dummy variables for the ERS farm resource regions. We use the ERS regions because they are formed by grouping counties with relatively similar land resources, climate, and crops. The majority of applications came from the Heartland region, a major soybean-producing area.

Given the differences between CSP and other conservation programs, we estimate determinants for CSP and other programs in separate equations. Because participation in USDA conservation programs is voluntary, ordinary least squares (OLS) models may be biased due to producer self-selection for program application. An example of an unobserved variable that may result in selection bias could include a nearby stream that is very polluted. To account for self-selection, we use an endogenous switching model (Maddala 1983; Abdulai and Huffman 2014). [1] [2] where T_0j is the transaction cost for farm j, given application to a program other than CSP; β₀ is a vector of parameters to be estimated; x_j is a vector of explanatory variables for farm j; ε_0j is an error term that is assumed to follow a standard normal distribution (N(0,1)); and D_j = 1 for producers who applied for CSP, and 0 otherwise. Equation [2] variables are defined identically, except that subscript “1” refers to CSP participants or applicants. Selection bias arises when the producer choice of program is correlated to the level of realized transaction costs. To test for and correct selection bias, we estimate a binary probit model of the decision to participate in CSP, along with the transaction cost equations: [3]

Selection bias, if present, will lead to nonzero covariance among errors for the CSP application and transaction cost equations: where = variance of error for non-CSP transaction cost equation, = variance of error for CSP transaction cost equation, = variance of error for CSP application equation, σ₀₁ = covariance for CSP and non-CSP transaction cost equations, σ_1u = covariance between the CSP transaction cost error and the CSP choice error, and σ_0u = covariance between the non-CSP transaction cost error and the CSP choice error. We note that σ₀₁ cannot be estimated because there are no observations with data on both CSP and non-CSP transaction cost data, and the error variance in the binary probit equation can be estimated only up to a scale factor (Maddala 1983).

Given correlation between ε_1j and u_j, OLS estimation implies E(ε_1j) ≠ 0. To correct for bias, the transaction regression equations are adjusted to account for the decisions to apply for CSP or non-CSP enrollment: [4] [5] which suggest new regression equations:

The model can be estimated using a two-step procedure where α is estimated with binary probit then β₀,β₁,σ_0u, σ_1u are estimated by OLS. Then and can be estimated from residuals, corrected for bias.

Finally, identification requires that at least one variable in z_j be excluded from x_j. We exclude the stewardship variables. While they indicate a history of stewardship and suggest which farms are more likely to be eligible for CSP or are more likely to be enrolled, they do not relieve farmers of documenting land use, production, and conservation practices in the process of applying for CSP enrollment.

VI. REGRESSION RESULTS

Parameters are estimated separately for hours spent on ex ante and ex post activities (Table 4). Each model includes a probit regression for CSP participation versus participation in other programs and regressions to identify factors affecting the magnitude of transaction costs for non-CSP and CSP programs. Both transaction cost regressions include bias correction based on the probit model. Selection bias is indicated only in the CSP equation for ex ante transaction costs. Estimated error correlation is positive (0.539)¹¹ and significant (p < 0.01) implying that producers who participate in CSP also have higher transaction costs. For the other equations, OLS models would be unbiased.

For both sets of regressions, human capital had a significant effect on CSP participation. Years of education and a higher number of operators are both associated with a significantly increased likelihood of CSP participation. Farmer age and retirement status do not appear to make a difference in CSP participation. Farm size also appears to affect CSP participation, although coefficients are significant in the ex ante model, but not in the ex post model. The probability of CSP participation increases as total farmland area rises, but declines with increases in the value of production. These results may reflect the focus of CSP on land-based practices. Farms with large land area relative to the value of production are most likely to be enrolled in CSP. We had expected that having a soil conservation plan, a nutrient management plan, or a pest management plan prior to 2004 would increase the likelihood of CSP participation. While this was true for nutrient and pest management plans in both probit regressions, having a soil conservation plan actually reduced the likelihood of applying for CSP versus other USDA programs examined in this study. The negative sign on the soil conservation equation may reflect the fact that soil conservation plans are required on highly erodible land for producers who receive income support, disaster, or conservation payments from USDA programs. These plans, however, may not fully address soil erosion (i.e., reduce erosion to levels that sustain soil productivity and minimize sedimentation). Under the highly erodible land conservation provisions of the 1985 Farm Act, better known as Sodbuster, producers were allowed to implement plans that were less expensive than plans that would have fully protected the soil (Claassen et al. 2004). As a result, these conservation plans may not indicate a level of stewardship that satisfies CSP requirements. As shown in Table 3, CSP participants are less likely than non-CSP conservation program participants to have formal soil conservation plans (although they would have had to address soil quality concerns to be eligible for CSP) and less likely to report that the surveyed field is designated as highly erodible.

Turning to the factors affecting transaction costs/hours, a number of variables are significant for ex ante costs, but they differed by type of program. For non-CSP programs, more education may indicate a high level of skill in developing plans on paper, and this may be viewed by applicants as a comparative advantage in competing for conservation program money, particularly in programs like EQIP where applications typically exceed funding and complex benefit-cost indices are used to rank applications. We also find that retired producers spend more time on non-CSP conservation program application. These producers may find comparative advantage due to experience in dealing with older government programs or may perceive lower opportunity costs in allocating time to conservation program applications. Farmers who are primarily retired may be farming on a relatively small scale or have passed many of the production and management tasks on to a younger family member. Education and retirement status do not appear to be significant factors in CSP transaction costs.¹² Only one factor significantly affected the magnitude of transaction costs for both non-CSP and CSP program applicants: the number of operators increased transaction costs, ceteris paribus. On multiple-operator farms, specialization on the part of individual operators may also lead these producers to perceive a comparative advantage to spending more time on conservation program applications. On these farms the overall level of management input may also be larger, reducing the time constraint. For non-CSP applicants, farm size, measured by value of production, increased the time spent, in line with expectations, although farm size measured by total farmland acreage reduced the amount of time spent. So, for any given level of production (value), transaction costs tend to be higher on farms with relatively small land area. That is, transaction costs are higher when the value of production per acre is higher, implying that production is more intensive. The literature indicates that transaction costs and abatement costs are positively correlated, which may explain this result. Complexity of the farming system, as measured by the proportion of the farm receipts from livestock, did not affect transaction costs nor did having highly erodible land.

Regarding ex post transaction costs for non-CSP programs, farmer age significantly affected the costs of signing the contract and documenting compliance. The estimated parameters imply that program participants spend more time on these activities as they age, although the rate of increase declines as the producer ages.¹³ The time spent on ex post transaction activities for both CSP and non-CSP programs was significantly and negatively related to value of production, perhaps due to time constraints, holding number of operators constant. They may also have proposed practices that were more easily documented, but these data are unavailable. In line with theory, farmers with a higher proportion of value from livestock, and thus more complex farming systems, had higher ex post CSP transaction costs.

VII. CONCLUSIONS

The magnitudes of transaction costs of farmers who actually applied to conservation programs do not seem particularly onerous and are lower than the transaction costs that have been measured for European AESs. The data strongly suggest that farmers spend considerably more time applying for the CSP, although time costs are still modest. Successful CSP participants spend 28.5 hours, on average, on developing applications, signing the contract, and documenting compliance. Half of CSP participants spent 14 hours or less on all of these activities, combined. We note that our data measure only the farmers’ own time and do not capture other costs such as consultants who may be hired to help develop conservation program applications or assistance received from NRCS staff. Because of this, and because the survey required recall of activities up to four years in the past, we expect that these estimates represent a lower bound of actual transaction costs. Because our sample size is modest, particularly for individual non-CSP programs, differences in time spent on applications and so forth for those individual programs could not be determined.

Nonetheless, our data on program participation barriers among nonparticipants suggest that many farmers perceive transaction costs as an important barrier to conservation program participation, more important than payment levels. This adds to the existing literature on barriers to participation. Farmers who had experience with conservation planning (e.g., conservation plans for highly erodible land) or conservation programs were less likely to identify transaction costs as a barrier, suggesting that the perception of transaction costs is worse than reality. Further research on actual versus perceived transaction costs is thus warranted; the disparity itself may be indicative of information acquisition as a transaction cost. In the meantime, outreach efforts, designed to educate farmers on the actual process of applying for conservation programs, may be helpful in encouraging broader interest in program participation by lowering information costs. These efforts, however, are more likely to improve conservation program performance if they are targeted to producers who can address locally important resource concerns in a cost-effective manner. An educational module designed for college farm management classes might also be helpful to increase familiarity with the programs and learn application procedures.

Our econometric results confirm that CSP applications/participants spend more time on application and compliance activities than do farmers in other programs. These results also indicate that increased management capacity and demonstrated stewardship increase the likelihood of participating in a more complex and demanding program, CSP. Given the farmers’ choice of program, there are some variables that are associated with increased time spent on applying for and, if accepted, complying with program requirements. As far as other factors that are hypothesized to affect transaction costs, no variables were significant for all types of programs and across both ex ante and ex post activities. There is some evidence for farm complexity increasing ex post costs, particularly compliance documentation costs.

While our initial analysis provides some valuable insights on transaction costs and participation barriers, more research is needed, using more complete information on conservation program offers and contracts. Producers who propose a single practice to address a narrowly defined problem on a single field will probably spend very little time on a conservation program application, while broader, more complex proposals will require more effort. At present, however, the survey data that underlie our analysis cannot be linked to contract or offer data for any conservation program. In the future, this information could be obtained inexpensively by linking conservation program offer and contract data to ARMS survey data, or, at greater expense, by expanding the set of survey questions that capture conservation practice adoption and conservation program participation.

Our research suggests that while the magnitudes of farmer transaction costs associated with conservation programs in the United States are not high, they do seem to represent a barrier to program participation. Efforts to simplify forms and procedures may be a costeffective way to increase participation, particularly among farmers who have not traditionally participated in government-sponsored conservation programs, while also reducing the transaction costs borne by government agencies such as NRCS.