The World’s Largest Open Access Agricultural & Applied Economics Digital Library This document is discoverable and free to researchers across the globe due to the work of AgEcon Search. Help ensure our sustainability. Give to AgE con Search AgEcon Search http://ageconsearch.umn.edu [email protected] Papers downloaded from AgEcon Search may be used for non-commercial purposes and personal study only. No other use, including posting to another Internet site, is permitted without permission from the copyright owner (not AgEcon Search), or as allowed under the provisions of Fair Use, U.S. Copyright Act, Title 17 U.S.C. Role of Ethanol Plants in Dakotas’ Land Use Change: Analysis Using Remotely Sensed Data Gaurav Arora Dept. of Economics & Center for Agricultural and Rural Development, Iowa State University Email: [email protected] Peter T. Wolter Department of Natural Resource Ecology and Management, Iowa State University Email: [email protected] Hongli Feng Department of Economics, Iowa State University Email: [email protected] David A. Hennessy Dept. of Economics & Center for Agricultural and Rural Development, Iowa State University Email: [email protected] Selected Paper prepared for presentation at the 2015 Agricultural & Applied Economics Association and Western Agricultural Economics Association Annual Meeting, San Francisco, CA, July 26-28. This research is part of a collaborative project supported by the United States Department of Agriculture- National Institute of Food and Agriculture, award number 2014-67003-21772. The views in this paper are attributable solely to the authors. Copyright 2015 by Gaurav Arora, Peter T. Wolter, Hongli Feng, and David A. Hennessy. All rights reserved. Readers may make verbatim copies of this document for non-commercial purposes by any means, provided that this copyright notice appears on all such copies. 1 Abstract: North and South Dakota have experienced rapid land-use changes in the past decade. Recent studies have shown that these land-use changes are mainly characterized by conversions of grasslands to crop production, especially corn and soybeans. Approximately 271,000 hectares of grasslands were lost to corn and soy production in 2006-2011 period, almost seven times the losses in 1989-2003. The implications of these changing land-uses range from reduced biodiversity and loss of habitat for waterfowl species to low agricultural productivity on drought- sensitive marginal lands. While progress has been made in characterizing regional land-use changes, formal analyses establishing causal relationships at the local level are lacking. We construct a spatially delineated dataset for the Dakotas and utilize a Difference-in-Difference (DID) model in conjugation with Propensity Score Matching to estimate the impact of an ethanol plant on nearby corn-acres. We hold the advent of an ethanol plant to be a treatment that influences land-use on surrounding agricultural plots. In our preliminary work, based on the Parallel Paths assumption of the DID, we find that the effect of ethanol plants on corn production varies by plants and a single point estimate for all ethanol plants in a region, as usually provided in the literature, can be highly misleading. Surprisingly, we find both positive as well as negative effects of ethanol plants on corn-acres that may be statistically insignificant. Negative estimates are irreconcilable to the economic incentives due to these corn-based ethanol plants. We find intensified corn production and reduced soybeans due to the ethanol plants. Our analysis also reflects a difference in opportunity of converting from wheat to corn and from grass to corn. We use placebo tests and pre-treatment trends in corn acres to examine the Parallel Paths assumption that identifies the DID estimates. We find that this assumption fails and propose to carry out this analysis by incorporating differentiated trends into the DID framework through more flexible assumptions in future. An important contribution of this paper is that it presents a unique research design that uses quasi-experimental techniques to evaluate the impact of a change/policy upon availability of spatially delineated datasets. To this extent, our results are to be viewed as preliminary. 2 Background and Motivation Characterizing the Dakotan Land Use Change Recent research findings suggest rapid land use changes in North and South Dakota, where grasslands have been lost to corn and soybean cultivation. Wright and Wimberly (2013) characterize conversion rates from grass to corn and soybean in the U.S. Western Corn Belt (WCB) from 2006 to 2011. The authors attribute expanding biofuels production and increased crop prices as potential factors driving higher production of these crops and therefore, such land use changes. The WCB spans five states: North Dakota, South Dakota, Nebraska, Iowa and Minnesota. A total of 271,000 hectares of net grassland losses in the Dakots out of 528,000 hectares in all of the WCB’s five states imply that conversions during this period were predominantly in the Dakotas. Spatial characterization of land use changes in these two states, using U.S. Department of Agriculture (USDA) Cropland Data Layer (CDL), finds westward expansion of the Corn Belt in regions east of Missouri River that intersect with the Prairie Pothole Region (PPR). Johnston (2014) provides a longer-term perspective on cropland expansion in the Dakotas, utilizing USDA National Agricultural Statistical Service (NASS) state-level cropping acres from 1980 to 2011, along with USDA CDL spatial imagery from 2006 to 2012. She reports that land attributed to corn or soybean production almost tripled between 1980 and 2011, where in 1980 it accounted for only 5% of the total area in the two states. Author also characterizes land use transitions among various categories such that probability of corn/soy being re-planted to corn/soy increased from 68% in 2006-07 to 80% in 2011-12. On the other hand, such probability for grasslands decreased from 81% in 2006-07 to 74% in 2011-12. In addition, corn and soybeans replaced multiple land uses such as wheat and other small grains that were 3 historically predominant in this region due to their climatic tolerance. Technological advancements yielding drought and cold resistant corn and soybean varieties are reported to be potentially driving such land use conversions. One other study by Stephens (2008) estimates the probabilities of grassland conversion conditional on amounts of surrounding grasslands, slope and soil productivity. Their annual estimate of the probability of grassland conversion was 0.004 for the Dakotas from 1989 to 2003, amounting to 36,450 hectares of grassland conversion for the period of study. However, they find that probability of conversion is not uniform across all lands of high biological value. Thus, conservation policies for such lands should be prioritized based on the probabilities of conversion, conditional on their location and other land attributes. A 2015 study by Lark, Salmon and Gibbs evaluates the types, amounts and locations of converted lands for cultivation in the conterminous U.S from 2008 to 2012. North and South Dakota are found to have experienced greatest increase in new cultivated land around all U.S. states during this period, predominantly east of the Missouri river. However, northwestern and southeastern North Dakota experienced contraction of croplands in 2008-2012 period. To evaluate conversion rates on native prairies they utilize long-term trend analyses from U.S. Geological Survey spanning 1972-2002. For the Dakotas, they report 14-25 acres of previously native prairies converted per 10,000 acres of land on the east of Missouri river and 10-14 acres converted west of the river. Overall, the Dakotas stood out with highest conversion rates on lands previously attributed to native grasses. Soybeans were found to be the first crop planted upon conversion during 2008-2012 period on east of the Missouri river, whereas west of the river spring and winter wheat were the first crop planted upon conversion in North and South Dakota respectively. 4 Although Dakotas’ native grasslands are a natural resource of national importance, most is under private ownership. Hence, the observed land use changes reported in the recent literature are an aggregate outcome of private decisions by individual landowners. These decisions could be a result of change in many factors including climatic conditions, technology, the local business environment, infrastructure, commodity prices, government payments towards conservation and crop insurance etc. For instance, Claassen et al. (2011) provide evidence that federal crop insurance subsidies have intensified cropping practices by reducing related risks. They conclude that the 2008 Sodsaver provision that restricts such subsidies could reduce grassland conversions by up to 9% in the PPR. These land use decisions have not only permanently or temporarily change the overall landscape of these states, but would also have long term impacts on the welfare of local farmers in the Dakotas. Related Concerns and Policy Implications Land use changes in the Dakotas raise many ecological, agronomic, environmental and economic concerns and related policy implications. The aforementioned study by Wright and Wimberly (2013) acknowledges the threat to existing wetlands and supported biodiversity from rapid agricultural conversions in the PPR, since wetlands are critical nesting and habitat sites for regional waterfowl species. Increased corn and soybean acres on originally native grassland imply loss of ecosystem services. Reduced populations of game species, when such conversions are in close proximity to the wetlands in the area, augment these losses (Wright and Wimberly, 2013; Johnston, 2014; Stephens et al. 2005). Another finding of Wright and Wimberly (2013) that raises concerns as well as interests to policymakers is that, in the Dakotas, corn/soybeans has replaced pasture and hay for livestock production on high quality lands (Land Capability Class II, explained hereafter in the Data section). First, higher production of corn and soybeans means 5 fewer opportunities for livestock production. This may be due to an imbalance in incentives towards intensive cropping through reduced risks with insured crops and investments into developing tolerant genetically-engineered seed varieties. Second, rapid increase in corn and soybeans in the region would tailor the socio-economic structure of the region towards more crop-based infrastructure, thereby making crops even more attractive to farmers. Agronomic issues arising from grassland conversions relate to reduced soil quality and increased soil erosion. Shifts from grass-based agriculture to crop-based agriculture reduce the water holding capacity of the soils, reduce soil ecosystem functions and decrease soil carbon thereby reducing soil productivity. Erosion due to intensified row cropping practices, especially corn, degrades soil quality and pollutes water streams in the region (Wright and Wimberly, 2013; Johnston, 2014). Degraded soils ultimately affect land productivity due to elevated vulnerability to drought due to less suitable climates of this region (Wright and Wimberly, 2013). Further intensification of agricultural activity and prolonged periods of extreme weather events like droughts in this region are considered serious threat to mostly ephemeral wetlands. Further, loss of stored carbon from uprooting the native grasses accounts towards environmental impacts of conversion (Johnston, 2014). Among the policy suggestions, Johnston (2014) calls for policies that incentivize farmer behavior towards sustainable agricultural practices in light of detrimental environmental and soil-quality implications of intensive corn/soy production on these marginal lands. Further, whereas Stephens (2008) suggests conservation policies to prioritize land with higher chances of conversion based on their location and attributes, Wright and Wimberly (2014) suggest regulating location of biorefineries, deemed responsible for higher corn production in their study. Lark et al. (2015), while recognizing the broad economic and environmental impacts of land use 6 conversion, point to the need for reformed policies aimed towards conserving natural ecosystems. Even though the new Renewable Fuel Standards program (RFS2) mandated procurement of grains for ethanol production only from lands under cultivation prior to December 2007, their study finds substantial increase in croplands in the United States. Further, the authors recognize the importance of the new Sodsaver provision in the 2014 U.S. Farm Bill. This provision, applicable in the PPR states including the Dakotas, dis-incentivizes conversion of native sod for agriculture after January 2014 through reduced crop insurance subsidies. Based on their analysis, the authors recommend a nationwide Sodsaver provision that covers forests and native ecosystems other than grasslands. Our Contribution: Moving from Characterization towards Explaining Land Use Changes The above studies characterize the rate and extent of land use conversions in the Dakotas at various spatial and temporal scales. They also speculate on potential factors that driver these land use changes in the region. However, detailed analyses to identify various phenomena that drive land use changes in Dakotas are lacking. We take a first step in understanding this phenomenon by evaluating the impact of ethanol plants on land use changes for these states. All Dakotas’ ethanol plants are corn-based. Hence, we ask how the advent of an ethanol plant affects corn plantings in its proximity. There are 19 ethanol plants in Dakotas (four in ND and fifteen in SD) with a combined capacity of 1,386 million gallons per year (mgy, 363 mgy in ND and 1,023 mgy in SD). Together, the Dakotas provide for about 9% of the total U.S. ethanol production capacity, currently at 15,198 mgy. Fourteen (out of the nineteen plants in all) started operations in 2006- 2008 period, i.e. after the first RFS program was launched under the Energy Policy Act of 2005 and when rapid land-use conversion rates are found by the pertinent literature, discussed above. 7 To motivate the economic incentives from ethanol plants, we compare trends in county- level corn basis, before 2006 and after 2008, for counties that house these 14 ethanol plants (see figure 1). An increase in corn basis implies an increase in local corn prices relative to the corn futures price. Such an increase in corn basis could be tied to the incentives from the ethanol plants to land owners with farms in the plants’ proximity. It is possible for the ethanol plants to provide such incentives to the farmers who supply them corn from near-by areas, since it saves transportation costs for both supplier and the plant. Figure 1 shows a steeper basis trend for corn in post-2008 periods compared to the pre-2006 period. Therefore, we conjecture a positive and statistically significant impact of ethanol plants on local corn acreage. We also extend our models to analyze the effect of ethanol plants on corn-soybean rotations. We do this by separately analyzing evolution combined acreages of corn and soybeans in relation to the advent of an ethanol plant, and then compare these with that of corn acreage. If the effect of an ethanol plant on corn acreage is higher than on the combined acreage of corn and soybeans, then the implication is intensified corn cropping has occurred through reduced corn-soy rotations due to the ethanol plant. This paper is subdivided into the various sections. First, a literature review section discusses the relevant findings of the impacts of ethanol plants from studies in the past. Second is a data section that discusses how we constructed a spatially delineated dataset for this analysis and provides a detailed explanation of the relevant variables. Third, the methodology section presents our research design and the Differences-in-Difference model in conjugation with Propensity Score Matching. Fourth is a section for estimation results for each ethanol plant. Lastly, we include discussions and conclusions in another section. Literature Review 8 Earlier attempts in this direction involved evaluating indirect impact of ethanol plants on land use change by way of analyzing impacts on local corn prices and farmland values. In the more recent years studies have considered direct impact of ethanol plants on corn acres as measure of land use change. We provide a detailed review of the analyses of impacts on land acreage because these are of direct relevance to this article. We also provide a brief review of analyses involving grain prices and farmland values followed by direct impacts literature. Direct Impacts: Corn Acreage Miao (2013) has evaluated the proportion of corn acreage for the Iowa counties in response to location, capacity and ownership capacity of ethanol plants. He utilized a county-level panel data set from 1997 through 2009 and the Arellano-Bond generalized method-of-moments estimator to estimate the effect of ethanol plants on land use shares in the region. The specialized estimator attempts to controls for the endogeneity of ethanol plants and allows controlling for corn- soybean rotations by including lagged dependent variable (that is, proportion of corn acreage). He found a positive and significant impact of ethanol plants on intensity of corn production in Iowa. He further found that, all else equal, locally owned ethanol plants have twice as strong an effect on local corn acreage as their non-locally owned counterparts. Motamed and McPhail (2011) used remotely sensed data to estimate a non-linear response of proximity to ethanol plants on corn acreage for 12 U.S. Midwestern states: ND, SD, NE, MN, WI, IA, KS, OK, MI, IL, IN, OH. They utilized a panel regression model with corn acreage on each of 10 km X 10 km land parcels from 2006 to 2010 as dependent variable. Their explanatory variables include capacity of the nearest ethanol plant, distance to the nearest ethanol plant and grain elevators, cash bids at the nearest grain elevator and a soil productivity index for these parcels. To incorporate non-linearity of response, their regression model includes 9