The Incidence of U.S. Agricultural Subsidies on Farmland Rental Rates

Transcription

1 The Incidence of U.S. Agricultural Subsidies on Farmland Rental Rates Barrett E. Kirwan * University of Maryland May 2007 ABSTRACT Each year U.S. farmers receive more subsidies than needy families receive through welfare assistance or post-secondary students receive through student aid grants. Yet, who benefits from agricultural subsidies is an open question. Economic theory predicts the entire subsidy incidence should be on the farmland owners. Since non-farmers own nearly half of all farmland, this implies that a substantial portion of all subsidies accrue to non-farmers while a significant share of all farmers receive no benefits. Using a complementary set of policy quasiexperiments, I find, contrary to the standard theory, that tenant farmers receive 75 percent of the subsidy. The competitive predictions fail to hold because the farmland rental market appears to be less than perfectly competitive. In settings where the market appears more competitive, the incidence is larger for landowners. * I thank David Autor, Michael Greenstone, Jonathan Gruber, Melissa S. Kearney, James Poterba, Michael Roberts, and participants at MIT s Labor and Public Finance Lunches for their many helpful comments and suggestions. I also would like to thank James Breuggen and Jamison Burt at the USDA National Agricultural Statistics Service for their generous assistance in accessing the data. This work commenced while I worked as an intern at the USDA Economic Research Service, whose support I gratefully acknowledge. Financial support from the National Science Foundation is also gratefully acknowledged. Contact: bkirwan@arec.umd.edu

2 1 Introduction The primary goal of U.S. agricultural policy over the past century has been to increase farmers income. Since 1973, direct payments to agricultural producers have been a vital instrument for supporting that goal. 1 Whether agricultural subsidies actually benefit farmers, however, is an open question. Nearly half of all farmland in the United States is rented, almost all of it from non-farmers. Standard economic theory predicts that subsidies accrue entirely to land owners. Agricultural subsidies may not benefit farmers if non-farmer landlords are able to adjust rental rates to capture the subsidies paid to agricultural producers. In the United States, agricultural subsidies are a significant transfer program. Between 1998 and 2004, farmers received, on average, $17 billion annually. In comparison, federal welfare assistance to needy families averaged $12.5 billion, while federal aid to post-secondary students averaged $16.1 billion. On a per-recipient basis, agricultural subsidies are one of the largest income support policies. In 2002, the average farm subsidy recipient received $6,947. Compare this to $1,956 annually per recipient household in food stamps; an average total unemployment compensation claim of $4,369; or $6,540 per eligible individual in annual benefits from SSI, an income support for the needy aged, blind, and disabled. The size of the farm subsidy program alone emphasizes the importance of understanding whether the stated policy goals are being met. The standard model of agricultural subsidy incidence predicts that, due to the extremely inelastic supply of agricultural land, landowners receive the entire benefit of the subsidy. This model is widely used by economists to explain the effects of agricultural subsidies (e.g., Shultze, 1971; Gardner, 2001). 2 According to this model, non-farmers, who own 94 percent of rented 1 See Orden, Paarlberg, and Roe (1997) for a history of agricultural policy. 2

3 farmland and 42 percent of all U.S. farmland, reap a substantial share of the benefits from agricultural subsidies. Surprisingly little evidence substantiates this model. In spite of a large literature examining the capitalization of subsidies into farmland values, the proportion of the marginal subsidy dollar captured by landowners remains unknown. Much of the work investigating the effects of agricultural subsidies assumes full subsidy incidence on the landowner. Empirical approaches based on a full incidence assumption range from cross-sectional models (Herriges, et al, 1992; Barnard, et al., 1997), to time series models (Weersink, et al., 1999; Featherstone and Baker, 1988), to models using pooled cross-sections over time (Goodwin and Ortalo-Magne, 1992; Goodwin, Mishra, and Ortalo-Magne, 2003). Although these studies demonstrate a positive correlation between land values and subsidies, they fail to account for unobservable characteristics that influence both land values and subsidies. For instance, farmland productivity is a key component of subsidy calculation and a primary determinant of land values, but it is typically unobserved. Failure to account for unobservables wrongfully attributes productivity driven land value increase to subsidy payments. This paper investigates subsidy incidence by examining the relationship between farmland rental rates and agricultural subsidies using a panel of farm-level production data, which can be used to address the unobserved heterogeneity problem. Few others have focused on rental rates (Lence and Mishra, 2003; Roberts et al., 2003). As per-period prices, farmland rental rates respond primarily to innovations in the current period, including exogenous, policy-induced subsidy changes. Using one such change, the 1996 Federal Agricultural Improvement and Reform (FAIR) Act, combined with farm-level data from the 1992, 1997, and 2002 Census of Agriculture micro files, I identify the effect of agricultural subsidies on farmland rental rates. I exploit an exogenous 2 Gardner, speaking before the House Committee on Agriculture in February, 2001 stated, one of the things I think is fairly well-agreed upon that these [subsidy] programs do is that the way in which they affect farmers income is not through everything that happens on the farm, but in the particular way of holding up land values & land rental rates. 3

4 subsidy change while controlling for unobserved heterogeneity and regional shocks using farm fixed effects and time-varying county effects. I use instrumental variables techniques to overcome possible bias due to the difference between the expected subsidy in the spring and the actual subsidy received in the fall of the base year. I use the legislated, pre-determined 1997 subsidy payments as an instrumental variable to address this expectation error in the 1992 subsidies. The analysis finds that only about 25 percent of the marginal subsidy dollar is reflected in increased rental rates. This stands in sharp contrast to the standard model s predictions. This is an important finding because by assuming perfectly competitive land markets and a fixed supply of land, agricultural economists typically assume that the landlord captures all Ricardian rents, including subsidies. A closer look at the farmland rental market suggests that the standard assumptions of fixed supply and perfect competition may not hold. In particular, the farmland rental market may not be perfectly competitive. The evidence indicates that the incidence increases as competition increases. Based on these findings, economic theory suggests that subsidies may have effects beyond the farmland rental market. Because farm subsidies are factor-specific subsidies to land, the marginal subsidy dollar effectively lowers the rental rate by 75 cents. The lower rental price of land induces substitution away from other variable inputs toward rental land, while the lower marginal cost might result in greater output and increased variable input use. The ultimate effect on other variable factors of production is an empirical question. The analysis finds a significant, positive response of expenditure on nearly all variable inputs. Overall, the marginal per-acre subsidy dollar increases per-acre expenditure by about 55 cents. 4

5 The increased output ultimately results in a gain to tenant farmers. The analysis shows that tenant farmers gain 75 cents of the marginal subsidy dollar, supporting the notion that in fact the farm operator benefits from the marginal subsidy dollar and farm policy largely achieves its aims. The paper proceeds as follows. Section 2 details the institutional facts about the farmland rental market and subsidy policy. Section 3 describes the data. Section 4 lays out the empirical strategy employed in this investigation, emphasizing the identifying assumptions. Section 5 presents the incidence on landlords and provides evidence of the robustness of the estimated incidence. Section 6 provides a plausible explanation of the results. Section 7 examines the subsidy s effect on non-land factors of production. Section 8 provides evidence that the incidence on tenants is complementary to the incidence on landlords. Finally, section 9 interprets the findings in light of existing assumptions and policy objectives and suggests directions for future research. 2 Institutional Background 2.1 Overview of the farmland rental market Renting farmland is a common practice in U.S. agriculture, where more than 45 percent of the 917-million farmland acres are rented (USDA 2001b). A typical tenant rents 65 percent of the land he farms, paying either in cash or in shares of production. According to the 1999 Agricultural Economics and Land Ownership Survey (AELOS), 3 60 percent of rented farmland is paid in cash, 24 percent in shares of production, and 11 percent in a cash/share combination. In 1999, the average cash-rented acre went for $32. Subsidy recipients paid a slightly higher $36 per acre for the land they cash rented. In the United States, non-farmers own 94 percent of the rented land, or 340 million acres twice the size of Texas (USDA, NASS 2001b). 3 The Agricultural Economics and Land Ownership Survey is a follow-up survey to the Census of Agriculture conducted every 10 years. This survey asks questions of both tenants and land owners. 5

6 The rental market is characterized by short-term contracts in long-term relationships. Rental contracts commonly take the form of year-to-year handshake agreements. When formal, written rental contracts exist they also usually last one year. In spite of frequent renewal, these arrangements are typically found in long-term landlord-tenant relationships. Allen and Lueck (1992) report an average landlord-tenant relationship length of 11.5 years in Nebraska and South Dakota, and Sotomeyer, et al. (2000) report 11.3-year average landlord-tenant relationships in Illinois. The parties enter into the rental agreements early in the year, typically by March 1 st. A rental rate negotiated in the spring is based on the expected returns from farming during the upcoming year. Among the expected returns are government payments, which typically are made after the harvest. The mistiming between the setting of the rental rate and the realization of government payments results in forecast error. The forecast error involved in the agreed upon rental rate may serve to confound the relationship between subsidies and rental rates if not adequately addressed. Below I detail the instrumental variables (IV) strategy used to address the issue. 2.2 Agricultural Subsidy Policy A key component of domestic agricultural policy is the support of farmers income. The U.S. farmers income is supported by a two-tier system: 1) price supports and 2) income supports, i.e., subsidies. Price supports have existed since the farm program began in the 1930s. For most of the century, they were the chief mechanism for transferring money to the agricultural sector. Prices were supported by the government s promise to purchase the commodity at a predetermined price. Although the government once took direct possession of these purchases, today it generally augments the price the farmer receives from sales to a third party. 6

7 Income supports became a distinct objective with the introduction of production subsidies in Since their inception, subsidies have been calculated in a consistent way. An acre of land used to grow one of seven crops wheat, corn, sorghum, barley, oats, rice, and cotton could qualify to be subsidized. A qualifying acre receives a subsidy equal to a national subsidy rate times an output yield assigned to that acre. Between 1985 and 2002 the assigned yield, called a program yield, approximately equaled the farm-average yield between 1980 and An acre that qualifies for a subsidy is called a base acre. A farm s total crop-specific subsidy is the product of the subsidy rate, the program yield, and the total number of base acres for that crop. That is, (1) subsidy = subsidy rate * program yield * base acres. Prior to 1996, the subsidy rate was the difference between a legislated target price and the national average price received. As described below, after 1996, Congress legislated the subsidy rate, and it no longer depended on prices. It is important to note that program yield and base acreage are specifically tied to a qualifying acre of land. As an acre changes ownership, these parameters are transferred with the land. As such, agricultural subsidies are factor-specific subsidies to land The Federal Agricultural Improvement and Reform Act The 1990s saw an end to supply control measures and the completion of a process to decouple subsidies from contemporaneous production. In a dramatic move, the Federal Agricultural Improvement and Reform (FAIR) Act froze base acres at their 1995 level and removed all planting restrictions. With the exception of certain fruits and vegetables, producers were given complete planting flexibility, while they still received subsidies based on their 1985 program yield 4 Prior to 1985 the program yield was the 5-year Olympic average of the farm s annual yield. The current program yield is the average of the program yields. 7

8 and their 1995 acreage base. At the same time, the subsidy rate was no longer conditioned on the commodity s price, as it was before 1996, but it was exogenously determined by the policy. Two aspects of this reform are significant to the analysis below. First, by freezing the farmspecific program parameters (program yield and base acres), the policy divorced farmer behavior from the subsidy. As described below, this unique feature of the legislation allows me to control for simultaneity that would otherwise serve to confound the incidence estimates. Second, the 1996 legislation removed the uncertainty of subsidy payments by establishing a schedule of annual payments from 1996 to I exploit the post-1995 known subsidy payments as an IV for pre ex ante uncertain subsidy payments. Failure to address this issue would result in attenuation bias of the incidence estimate The Farm Security and Rural Investment Act of 2002 Legislation signed into law in May 2002 reintroduced a connection between subsidies and output prices. In the years after the FAIR Act of 1996 market prices decreased significantly, but there was no automatic subsidy response. Instead, Congress passed emergency legislation that increased subsidies by 50 percent in 1998 and 100 percent in 1999 and Consequently, Congress did not end agricultural subsidies in 2002, which was ostensibly the goal of the 1996 legislation. Instead, Congress increased subsidies and again conditioned them on output prices. Each of these pieces of legislation exogenously changed the per-acre subsidy rate, providing a second quasi-experiment. An important caveat for this experiment is that although the legislation became law after planting decisions were made in 2002, the provisions of the law were made retroactive in order to cover the 2002 crop year. Hence, 2002 may best be considered a policy transition year rather than a new policy regime. 8

9 3 Data The primary data source used in the analysis described below is the U.S. Census of Agriculture, a quinquennial census of those who produce at least $1,000 of agricultural goods. These data are confidential micro files accessed under an agreement with the USDA Economic Research Service and the USDA National Agricultural Statistics Service (NASS). The data are available at NASS in Washington, DC. 5 The Census of Agriculture contains farm-level information such as total acres farmed, total acres rented, and total government payments received. It also contains detailed information on the number of acres harvested and the total production of each crop. The value of production is reported for 13 crop groups and for all livestock, and the Census collects information on the corporate structure of the operation as well as demographic information on the primary operator. Additionally, the USDA collects detailed information on the financial structure of the farm, including total cash rent expenditure, 6 by employing a stratified sampling procedure in order to select a subset of farms to receive the Census long-form. 7 This sampling procedure over samples large farms. 3.1 Sample Creation In the main analysis I use two years of the Census of Agriculture, 1992 and 1997, to create a balanced panel of farming operations. Each year, roughly 1.6 million farms respond to the Census. About one million farms are observed in consecutive census years, and approximately three- 5 Any interpretations and conclusions derived from the data represent the author s views and not necessarily those of NASS. 6 Long form recipients report production expenses in 15 cost groups, the estimated value of the farm s land and buildings, the type and amount of equipment used, the number of workers hired, and the use of agricultural chemicals and fertilizer. 7 Farms are stratified based on expected revenue. All farms with very high expected revenue are selected, and approximately one-in-four of all other farms also receive the Census long form. I use Census-provided sample weights when appropriate to account for this sample composition. 9

10 quarters of these farms grow a subsidized crop in either year. The first two columns of Table 1 contain the summary statistics for the population of farms that grow subsidized crops. Nearly half of farms growing subsidized crops cash rent some land. The summary statistics reported in Table 1 columns three and four reveal that cash-renting farms are slightly larger and more profitable than the average farm. In 1992, cash-renting farms were 200 acres larger (about 30 percent) on average and earned on average $4 more per acre (about 7 percent) in net returns. Cash renters were also more likely to receive subsidies (64 percent versus 55 percent) and received a 50- cent higher subsidy per acre than the average farm in 1992 ($16.22 versus $15.73). I construct the sample for the final analysis by limiting the sample to farms that report paying cash rent in both years and paid less than $300 per acre in rent. 8 The summary statistics of the final analysis sample, contained in the last two columns of Table 1, reveal that the final analysis sample is representative of the population of cash renters. The only substantial difference between the population of cash renters and my sample is in the mean rental rate, reflecting the presence of large outliers in the population; note that the median rental rates, reported in brackets, are very similar between the sample and the population. The final analysis sample consists of 57,791 farms observed over two years Sample Selection Bias Farms used in the final analysis reported paying cash rent in both 1992 and This poses a problem if the propensity to rent farmland in 1997 is influence by the 1996 policy change. This source of endogeneity could bias the incidence estimate downward if farms experiencing a high subsidy pass-through stopped renting, but those experiencing low incidence continue to rent. I investigate the potential sample selection bias using a two-step Heckman selection correction. In 8 Limiting the per acre rent at $300 removes farms that are primarily renting structures. Fewer than 0.5-percent of reported rental rates exceeded $300 per acre in the 1997 June Agricultural Survey, a nationwide survey that specifically asked for cropland cash rental rates. The results are substantively the same with a $500 per acre rental rate cutoff. 10

11 addition to the control variables used in the main analysis below, the first step includes characteristics of the farm and farm operator that might affect the decision to rent, but are not likely to affect the rental rate paid, e.g., farm organization, age, race. 9 Based on these exclusion restrictions, the coefficient on the inverse Mill s ratio when included in the main regressions below is not statistically significant, a result consistent with no sample selection bias. 10 The incidence estimate remains the same as the values reported below, demonstrating its robustness to this specification. 3.2 Variable Creation Dependent Variable The Census of Agriculture does not report the per-acre rental rate, however respondents do report the total amount paid in cash rent. The total acres rented on a cash, share, or free basis also are reported. From these two variables, I create the per-acre rental rate by dividing total cash rent by total acres rented. Admittedly, the resulting rental rate will be too small for farms that cash rent part of the land and share rent another part. 11 The resulting measurement error is not classical measurement error. Since the calculated rental rate is less than or equal to the true cash rental rate, the expected value of the measurement error must be less than zero. As long as the non-classical measurement error is uncorrelated with the regressors, only the intercept will be confounded. However, if the measurement error is positively correlated with the magnitude of government 9 The demographic covariates are: whether the farm is a family farm, whether it is a partnership, whether the farmer resides on the farm, whether farming is the principal occupation, number of days worked off farm, farming experience, experience squared, age, age squared, race. 10 The coefficient on the inverse-mills ratio is with a standard error of According to the 1999 AELOS, 17.5 percent of farms that rent some land do so using both cash and share leases. 11

12 payments, the estimated effect of government payments on rental rates will be biased upward. Evidence from the 1999 AELOS demonstrates that if an upward bias exists, it is likely to be small Independent Variables The subsidy variable is constructed from the reported government payments variables. Every agricultural producer is asked to report non-price support payments received from the government. Producers are asked to report both total payments received and payments received from the Conservation Reserve Program (CRP). By subtracting CRP payments from the reported total payments, I construct an approximate measure of subsidy receipts. In 1992, subsidies accounted for 68 percent of direct government payments net of Conservation Reserve Program payments. 13 The remaining 32 percent are from disaster relief (17 percent) and other programs (USDA, NASS, 1996). In 1997, subsidies account for virtually all (96 percent) of non-crp direct government payments (USDA, NASS, 2001). The remaining covariates are constructed directly from variables contained in the Census of Agriculture. All regressors are measured on a per-acre basis Empirical Strategy Identification 4.1 Unobserved Heterogeneity 12 I investigate this source of bias using the microfiles of the 1999 AELOS, which asked for detailed information about the type of lease arrangement, the total cash rent paid, and the value of the share of production paid as rent. Using these data I construct an estimate of the non-classical measurement error, and I estimate the relationship between the measurement error and government payments. Appendix Table 1 reports the estimates. The coefficient represents the potential bias. With state fixed effects, the incidence estimate is 0.03 greater than the truth. However, with county fixed effects, I find no relationship between government payments and the measurement error, implying no bias in the primary analysis. Details of this analysis are available from the author upon request. 13 This measurement error provides a second justification for the IV strategy outlined below. Measurement error is unlikely to be driving the findings below since the analysis accounts for the entire subsidy dollar going to either the landlord or the tenant. Measurement error would attenuate both estimates, and less that the full dollar would be accounted for. 14 Analysis performed using log-levels produces very similar results. The per-acre analysis is presented in order to maintain a more natural interpretation of the coefficient. 12

13 If subsidies were randomly assigned, then the parameter identified by a regression of the rental rate on subsidy per acre would be the proportion of each extra subsidy dollar per acre reflected in higher rental rates. However, subsidies are not randomly assigned. They are most likely correlated with unobservable determinants of the rental rate, and the resulting OLS incidence estimate will be biased. Many farm characteristics influence both subsidies and farmland rental rates, yet they cannot be observed by the econometrician. The subsidy is a function of yield and crop choice which, in turn, are determined by farm-level soil properties and farmer human capital and entrepreneurial skill. Transient shocks, such as drought or pests, also may affect rental rates and government subsidies. Using farm-level data allows me to control for permanent farm-level characteristics that might cause the incidence estimate to be inconsistent. The positive correlation between government payments and the unobserved factors that influence productivity will result in an upward bias to the estimated effect of subsidies on rental rates incidence. Including farm and time-varying county fixed effects allows me to overcome this source of bias. The main estimation equation is: (2) rit = + git + X it + fi + Ct + it where r it is the average rental rate for farm i in year t. The average subsidy per acre is denoted with g, and f i is the fixed effect for farm i. The parameter C t is a time-varying county effect that allows for shocks, such as weather or pests, that impact everyone within a localized region. X it is a vector of observable covariates such as yield, selection and production of crops, occurrence of irrigation, farm size, revenue, and expenditures. The estimating equation used in this study is obtained from equation (2) by first differencing the data to absorb the farm effect, resulting in 13

14 (3) ri = C + gi + X i92 + i. The first difference of the control variables is not included because the 1997 level of these variables are potential outcomes influenced by the exogenous subsidy change. Instead, the 1992 values of these variables are included in the estimating equation. 4.2 Simultaneity and Expectation Error The remaining problems to be addressed are simultaneity bias and expectation error. Simultaneity bias occurs in 1992 because prices determine both the subsidy and the rental rate. High prices mean a lower subsidy and higher rental rate, resulting in a spurious negative relationship. The 1996 FAIR Act eliminates simultaneity bias in 1997 by divorcing prices from subsidy determination. I address the 1992 simultaneity by controlling for 1992 revenue and using the exogenous 1997 subsidy as an instrument to isolate the fundamental, non-price determined, components of the 1992 subsidy. Expectation error causes attenuation bias. As detailed earlier (see Section 3.1), rental rates are set according to expected receipts, including expected subsidy payments. Prior to the 1996 FAIR Act, subsidy payments were conditioned on the market price and thus were unknown until after the harvest, while rental rates were agreed upon before planting in the spring. This mistiming results in expectation error, the difference between expected and actual subsidies, becoming part of the error term in the estimating equation. Assuming the expected subsidy and the expectation error are uncorrelated implies that realized government payments are correlated with the error term. The effect on the coefficient of interest is the same as classical errors in variables, namely attenuation bias. The 1996 FAIR Act reduces the complexity of the problem by eliminating expectation error in Recall that in 1996 the subsidy rates were exogenously predetermined for the next seven 14

15 years. Because of this feature of the legislation, there was no expectation error in 1997 The expected government payments for 1997 and 1992, respectively, are: * (4) g i97 = gi97, (5) * g g i92 g i 92 i92 =. Substituting (7) and (8) into equation (2) and first differencing results in (6). g ri = C + g i + i i92 An instrumental variables strategy overcomes the attenuation bias induced by the expectation error. The 1997 subsidy level is a candidate instrument because it is uncorrelated with the composite error term in equation (6). By assumption, conditional on the fixed effects, g i97 is uncorrelated with both i92 and i97 thus with i. Furthermore, this variable is uncorrelated with the second error term, g i92, due to the absence of expectation error in This allows one to write g * the orthogonality condition as ( g ) 0 E. This condition holds if the subsidy shock in 1992 i92 i97 = contained no information for the expected subsidy in 1997, a reasonable assumption. Since the 1997 level of government payments is uncorrelated with the composite error, it is therefore a good instrumental variable insofar as it is correlated with the change in government payments, a proposition explored below. 5 Estimation and Results 5.1 Ordinary Least Squares A Cross-Sectional Approach The approach taken by this paper has the advantage of controlling for unobserved characteristics of the farm, such as the operator s entrepreneurial skill and the productivity of the land. Estimates that fail to account for these characteristics likely suffer an upward bias as 15

16 productivity and skill are positively correlated with the subsidy. Estimation methods that fail to adequately control for these variables will attribute rental rate variation to the subsidy that should be attributed to productivity and skill. I explore the consequences of unobserved heterogeneity by ignoring the panel nature of the data and estimating the incidence in each year separately and in the pooled cross-section. Table 2 contains the results of this exercise. Panel A reports the estimates for 1992, panel B reports the estimates for 1997, and panel C reports the estimates obtained by pooling the two cross-sections. The incidence estimates decline steadily from the bivariate specification in column one until ultimately controlling for county fixed effects, measure of productivity, crop mix, returns, and farm characteristics in column four, dropping from 0.47 to 0.24 in 1992, from 0.92 to 0.33 in 1997, and from 0.64 to 0.27 in the pooled regression. This parameter instability causes one to wonder how much bias continues to exist because of other unobservable and excluded covariates A New Approach Table 3 contains the incidence estimates when I control for unobserved heterogeneity, as delineated in equation (3) above. Columns one and two report the results of the fixed effects model including farm and time-varying county effects. Column one contains no covariates, while column two includes the 1992 level of the covariates listed above. As noted earlier, the 1997 values of these covariates are potential outcomes affected by the changing subsidy and their inclusion in this specification would serve to confound the incidence estimate. Compared to the estimates in Table 2, the incidence estimate is lower once one accounts for unobserved farm characteristics, signifying a decrease in the upward bias. Another important feature of columns one and two is the stability of the incidence estimates. Once farm fixed effects are included, additional covariates do little to change the estimate, suggesting that farm fixed effects account well for potential sources of bias. 16

17 The most notable characteristic of Table 3 is the low incidence estimate; the estimated incidence is only 0.18, and it is significantly different from one. In other words, 18 cents of the marginal subsidy dollar is reflected in higher rental rates. As discussed earlier, conventional wisdom and simple economic theory suggest that the landowner should capture all of the Ricardian rents, including the subsidy. The evidence presented here suggests that the truth lies far from perfect incidence; the landlord is only able to capture 18 cents of the marginal subsidy dollar. One may be concerned that this estimate suffers from attenuation bias caused by measurement error or expectation error. The next section details the instrumental variables strategy used to account for these sources of bias. Section 9 provides further evidence against attenuation bias by fully accounting for the remaining portion of the subsidy dollar. These pieces of evidence demonstrate that the conclusions reached above are not likely to be driven by attenuation bias. 5.2 Instrumental Variables As discussed earlier, the measurement error and expectation error concerns apply to the 1992 government payments and can be addressed using the 1997 subsidy payment as an instrumental variable. The first panel of Table 4 reports the first stage of a two-stage least squares estimation strategy. Column one reports the coefficients and the F statistic when no covariates are included in the fixed effects specification. The instrument is a significant predictor of the change in subsidies. At a value of 0.85 and 20,755, Shea s partial R 2 and the F statistic indicate a strong instrument. The second panel of Table 4 reports the IV estimates. The IV estimates, and 0.263, are slightly higher than the OLS estimates, revealing some attenuation. By accounting for the potential sources of bias, namely unobserved heterogeneity, simultaneity bias, and expectation error, I have found that the landlord captures only about 25 cents of the marginal subsidy dollar. Economic theory predicts that if land is the only specific factor of 17

18 production and if markets are perfectly competitive then the landlord will capture the entire marginal subsidy dollar. These findings call into question the validity of these assumptions in the short run. 5.3 The Heterogeneity of Rental Rate Incidence across Region and Farm Size One might be concerned that the results presented mask variation in response across region or the size of the farm operation. Regions within the U.S. differ substantially in the crops grown and the predominant lease contract type. Noting that each crop is subsidized separately, one might worry that the incidence differs according to crop and subsidy regime. Farm size might also influence the size of the incidence. For instance, perhaps large farms are better able to negotiate for lower rental rates, and they are driving the relatively low incidence. I explore the robustness of the results by estimating the incidence separately for different regions 15 and different farm sizes. 16 The incidence appears very stable across resource regions and sales class. The point estimates for six of the nine resource regions lie between 0.2 and 0.3, and only one of the remaining three is distinguishable from zero at When estimated across sales classes, the estimates more tightly cluster around the main result, ranging from 0.24 to The uniformity of incidence estimates spatially and across firm size gives further confidence in a twenty-five percent incidence Further Evidence Land Values Panel C of Table 4 provides another check on the plausibility of the incidence estimate by focusing on the marginal dollar s effect on farmland value. Unlike the rental rate incidence, which 15 The USDA has established 9 resource regions corresponding to predominant crop mix and farming practices. 16 Following USDA classification, e.g., 17 Details available in the working version of this paper. 18

19 reflected the incidence of the contemporaneous marginal subsidy dollar, the land value incidence reflects the incidence of the contemporaneous subsidy dollar and information about future subsidy dollars. If the estimated incidence on the rental rate is a permanent feature of the farmland market, a reasonable discount rate should reconcile the rental rate incidence with the incidence on land values. Columns 5 and 6 in Table 4 report the results of fixed effects instrumental variables estimation using land values reported in the 1992 and 1997 Censuses. Again the 1997 subsidy instruments for the subsidy change. The estimates, which differ only slightly when covariates are included, show that farmland values increase by about $2.81 with the marginal subsidy dollar. Using a traditional present value model, one can calculate the implied discount rate from the estimated incidence on rental rates and on land values. Specifically, the change in land value, dl, equals the ratio of the change in the rental rate, dr, divided by the discount rate, : (7) dr dl =. Equating the estimated incidence on land values, 2.81, with the ratio of the rental rate incidence, 0.26, and the discount rate results in an implied discount rate of 9 percent. This discount rate is very similar to that calculated by others, e.g., Weersink, et al. (1999) and Lamb and Henderson (2000), lending further credence to the 25-percent incidence estimate The Food Security Reform Act of 2002 The time period provides yet another quasi-experiment with which to explore the subsidy incidence. When farmers made planting decisions and signed rental contracts for the 2002 crop year, the 1996 FAIR Act was in place and Congress was reconciling the House and Senate versions of the 2002 farm bill. The Food Security Reform Act of 2002 subsequently became law on May 13, 2002, and its provisions covering the 2002 crop year were implemented as soon as 19

20 practicable after the date of enactment of this Act. The 2002 farm bill changed both the subsidy level and the relative subsidy rate, thereby impacting the farm-average subsidy, depending on the farm s crop mix. This source of exogenous variation in the subsidy rate provides a second setting in which to identify the impact of subsidies on cash rental rates. The potential uncertainty surrounding 2002 subsidies at the beginning of the 2002 crop year warrants implementation of the previously used IV strategy in order to account for potential expectation error. Table 5 contains the estimates from the quasi-experiment. In parallel fashion to the main analysis, the sample is selected to be those potentially eligible for subsidies, i.e. growing program crops in 1992 or 1997, and cash renting in both 1997 and The results from using the 1997 subsidy to instrument for the subsidy change are presented in columns 3 and 4. The Shea s partial R 2, reported on the last row of columns 3 and 4, show that the instrument is weaker for the change than it was for the change (0.378 versus 0.878). Nevertheless, the F statistic, which has a value of 7,545, demonstrates that it is still a strong instrument in this setting. Instrumenting for the change in subsidies results in slightly larger incidence estimates. The IV estimate with a full set of controls suggests that in 2002 landlords gained 10 cents of the marginal subsidy dollar. These findings reinforce the main results that landlords extract a small proportion of agricultural subsidies, and they underscore the relevance of imperfect incidence as a fundamental feature of the farmland rental market. 6 Interpretation Conventional wisdom holds that competitive farmland markets enable landlords to extract the entire marginal subsidy dollar, resulting in perfect subsidy incidence on farmland. The less than full subsidy incidence found here warrants reconsideration of the fundamental assumptions. One possible explanation is that the pattern of these incidence estimates reflects frictions in the farmland 20

21 rental market. Farmland rental markets may adjust slowly to subsidy innovations, as reflected in a 10-percent incidence in the first year of a subsidy change (2002) and a 25-percent incidence two years after the previous subsidy change (1997). Imperfectly competitive rental markets provide another possible explanation of these findings. In a market with many landlords and few renters, the landlords may implicitly share the subsidy dollar in an attempt to attract tenants. Landlords may have low reservation rents since agricultural land has few alternatives outside farming. Historically, only active farmers could receive subsidies, landlords could not. Therefore, in a less-than-perfectly competitive rental market, the landlord might share the surplus in order to attract a tenant, resulting in less than perfect incidence. Although landlords qualified to receive subsidies for the first time with the 1996 FAIR Act, they would still forego Ricardian rents unless they found a tenant, resulting in similar surplus sharing incentives. In order to examine the hypothesis that tenants might have market power, I create four measures of rental market concentration. The first measure is the proportion of farmers in a county who rent some land, and the second is the proportion of farmland in a county that is rented. I also calculate two Herfindahl indexes to measure rental market concentration. One Herfindahl defines market share over total county rental expenditure, the other defines market share over the total number of rented acres. I interact these measures with the change in government payments in order to determine whether the marginal effect of subsidies changes as rental markets become less concentrated. The results from this exercise are found in Table 6. Every measure of concentration returns a sign and magnitude that is consistent with decreasing subsidy incidence as the rental market becomes more concentrated. For instance, the first row of Table 6 indicates that if all farms in a 21

22 county rented land the incidence would increase to about 33 percent. Notably, in this specification the main effect is very close to zero; the effect works through the interaction term, suggesting that rental market concentration is an essential element of the subsidy incidence. These findings suggest that rental market imperfection is a plausible explanation for the incidence finding reported above. 7 Variable Factors of Production Agricultural subsidies are de facto specific subsidies to land. As such, they serve to reduce the rental price of land and change the relative prices of the factors of production. The altered relative prices will have an indeterminate impact on the use of non-land inputs. Because the relative price of non-land inputs has increased, producers will substitute away from them toward the relatively cheaper land. At the same time, since total costs are lower, the farmer may expand output and demand more non-land variable inputs. The net result on the non-land factors of production is theoretically ambiguous. What follows addresses the broader ramifications of agricultural subsidies. 7.1 Land The subsidy will directly affect the demand for land. Facing a lower rental rate, farmers will rent more acres. The first row of Tables 7 and 8 support this implication. Since the number of rented acres will have a mechanical, negative relationship with the subsidy per acre, the log-rented acres are regressed on the log-total subsidy. One should thus interpret the coefficient as the elasticity of rented acres with respect to the marginal subsidy dollar rather than the marginal subsidy dollar per acre. The results of this estimation suggest that increasing the subsidy by one dollar per acre (a total increase of $860 for the average farm) leads to about one and a half more acres rented according to the IV estimates. This result supports the theoretical prediction about the direct effect of subsidies. 22

23 7.2 Non-land Variable Factors of Production I measure the indirect effect of subsidies on non-land variable factors of production by applying the empirical strategy developed above to expenditures on 13 variable factors of production. The data on variable input expenditures are reported by those returning the census long form. Using the same sample of cash renters as above, I treat the expenditure on each input as the dependent variable and estimate the effect of subsidies using equation (3). The appropriate specification in this setting must adequately account for the secular trend in expenditures. The time-varying county effect not only captures local shocks, but also accounts for the county-specific secular trend. If the expenditure trend is proportional to the farm size then the per-acre transformation is sufficient for the county effect to capture the change. However, if the secular trend is proportional to total expenditures and not to farm size then a log-log specification would be preferred. It stands to reason that expenditure growth of many factors, e.g. livestock and feed, will not be proportional to the acres farmed. Therefore, the log-log specification is the appropriate specification for this analysis. Table 7 reports the OLS coefficient on government payments for each of the 13 factor expenditure regressions and for a total expenditure regression. All of the regressions include a complete set of covariates as specified earlier. For comparison purposes, column one reports the results from the per-acre specification. The largest response comes from expenditures that are least likely to grow in proportion to farm size, such as livestock and feed. Column 2 of Table 7 reports the elasticity estimates from a log-log specification. Columns 3 and 4 report the incidence, which is calculated by evaluating each elasticity at the mean and median of the dependent and independent variables. Using OLS, I estimate that the marginal subsidy dollar results in a 35 cent increase of total variable factor expenditures. 23

24 Expenditures might suffer from expectation error in much the same way as rental rates. The subsidy also continues to be measured with error. To account for this, I instrument for the change in government payments using the 1997 government payments. Table 8 reports the IV estimates. The estimated response of total expenditures to the marginal subsidy dollar increases to 55 cents after instrumenting for the subsidy change. The higher estimate is consistent with attenuation bias in the OLS estimator caused by expectation and measurement error. An important caveat to these findings is the difficulty of disentangling the effect on quantity purchased from the effect on price in this analysis. There are little data on the quantity of inputs purchased by individual producers, but data are available on the number of acres treated with fertilizer and chemicals on each farm. Analysis of these data reveals a positive, statistically significant response in the number of acres treated with fertilizer and/or chemicals. 18 Further analysis reveals no response in the fertilizer/chemical expenditure per acre treated. 19 Together these findings suggest that the increased expenditures are primarily driven by increased quantity demanded. Ultimately, the indirect effect of the subsidy increases expenditure on variable factors by about 55 cents, revealing a positive production response to the factor-specific subsidy. This response is consistent with the cost of farming an additional acre and a half of land. As noted above, increasing the subsidy payment by one dollar per acre for the average farm would increase the total subsidy payment by $860 and result in about one and a half more acres rented. The average cost of farming an acre of land, according to the summary statistics in Table 1, is about $254. The expenditure results imply that an extra $860 subsidy results in $473 more dollars spent 18 Details on this analysis are available from the author upon request. 19 Although this analysis does not distinguish between a price effect and n intensity effect, it seems unlikely that these effects would perfectly offset each other. 24

25 on other variable factors of production, an amount consistent with the cost of farming about one and a half additional acres. 8 Tenant s Incidence The above analysis found landlords capturing 25 cents of the marginal subsidy dollar, and tenants increasing expenditures on variable inputs by 55 cents. Yet the question of how the subsidy ultimately affects the tenant remains. If variable inputs are perfectly elastically supplied, and the farmer is a risk-neutral profit maximizer, then one would expect the farmer s net returns to increase by at least 84 cents on average since in the sample farmers rent 67-percent of their farmland on average. The production response should not dissipate the subsidy incidence on the tenant. Using an empirical strategy similar to the one detailed in section 5, I measure the tenant s incidence by regressing net returns on government payments. Table 9 reports the results of this analysis. Columns 1-4 emphasize the point made earlier that, for some farms, expenditure growth (hence growth in net returns) is not proportional to farm size. Using all farms in the sample, column 1 reports that tenant farmers gain more than dollar-fordollar from the subsidy, in spite of forfeiting 25 cents to landlords on rented land. Median, robust, and trimmed regressions in columns 2, 3, and 4, respectively, report a more reasonable cent gain. 20 Instrumenting for possible measurement error results in little change, narrowing the gain to cents in columns 5 and 6. The robust IV estimates indicates that the tenant ultimately receives about 84 cents of the marginal subsidy dollar, a finding consistent with a tenant/landlord split of the marginal subsidy dollar on rented land. Accounting for the entire marginal subsidy dollar bolsters the low subsidy incidence found in Section 6 and eliminates concern about measurement error induced attenuation bias. In this 20 The main analysis is robust to these specifications, e.g., estimating the rental rate incidence with the trimmed sample used in Table 9 columns 4 and 5 leaves the results virtually unchanged. 25