1、AN EMPIRICAL ANALYSIS OF ACREAGE EFFECTSOF PARTICIPATION IN THE FEDERAL CROPINSURANCE PROGRAMBARRY K. GOODWIN,MONTE L. VANDEVEER, AND JOHN L. DEALThe extent to which crop insurance programs have resulted in additional land being brought intoproduction has been a topic of considerable debate. We cons
2、ider multiequation structural models ofacreage response, insurance participation, CRP enrollment, and input usage. Our analysis focuses oncorn and soybean production in the Corn Belt and wheat and barley production in the Upper GreatPlains. Our results confirm that increased participation in insuran
3、ce programs provokes statisticallysignificant acreage responses in some cases, though the response is very modest in every case. In themost extreme cases, 30% decreases in premiums as a result of increased subsidies provoke acreageincreases ranging from 0.2% to 1.1%. A number of policy simulations i
4、nvolving increases in premiumsubsidies are considered.Key words: acreage response, crop insurance.U.S. crop insurance programs have undergonesignificant changes in recent years. The 1980Federal Crop Insurance Act significantly ex-panded the program and provided large subsi-dies to encourage particip
5、ation. The 1994 CropInsurance Reform Act brought about a briefperiod of mandatory participation in the pro-gram, expanded premium subsidies, and insti-tuted a “catastrophic“ (CAT) level of protec-tion that was intended to replace disaster reliefpayments at very low cost to producers. The1994 Act als
6、o mandated development of “costof production” insurance programs that, alongwith innovations by private insurers, led to thedevelopment of several different revenue in-surance programs. Legislative actions in thelate 1990s and in 2000 further expanded pre-mium subsidies. Participation has grown sub-
7、stantially in response to these new programsand policy changes that have made crop insur-ance more attractive to producers.1Barry K. Goodwin and John L. Deal are with North CarolinaState University. Monte Vandeveer is with the Economic ResearchService.Helpful comments and suggestions were received f
8、rom DanSumner, Hyunok Lee, Vince Smith, Jeff Lafrance, Bruce Babcock,Luther Tweeten, Stephanie Mercier, Bob Myers, and two anony-mous reviewers. This research was supported in part by the Eco-nomic Research Service and the North Carolina AgriculturalResearch Service.1For example, in 1995, total liab
9、ility for corn insurance was$6.7 billion. In 2001, total liability for corn had grown to nearly$10.7 billion, with $6.7 billion being accounted for by the CRCrevenue insurance program (USDA-RMA 2003).The U.S. crop insurance programs clearlyprovide positive net benefits to participatingproducers. Any
10、 provision that enables an eco-nomic agent to better withstand risk will pro-vide incentives for agents to assume greaterrisk. Likewise, if participation in any programconveys positive net economic benefits to pro-ducers, the program may alter incentives andthus affect production patterns. Concern a
11、sto whether government risk management pro-grams have affected planting decisions goesback at least to the 1970s, when observers spec-ulated that disaster payments were encourag-ing production in high risk areas (Gardner andKramer). To the extent that the benefits ofrisk management programs are not
12、homoge-neous across crops and across regions, regionalcrop production patterns may be influenced.Acreage effects brought about by participa-tion in crop insurance programs could affectprices and input markets and thus have impor-tant policy implications.Existing research on the production ef-fects o
13、f insurance programs is limited. Re-cent results based on a single-equation, ag-gregate analysis by Keeton, Skees, and Longimplied that expansions in risk managementprograms in the 1980s led to the introduc-tion of about 50 million new acres of U.S.crop land into use (where use includes plantedacres
14、, idled acres, and land in conservation re-serves). A large proportion of this increase,Amer. J. Agr. Econ. 86(4) (November 2004): 10581077Copyright 2004 American Agricultural Economics AssociationGoodwin, Vandeveer, and Deal Acreage Effects of Insurance 1059approximately 35 million acres, was accou
15、ntedfor by land put into the Conservation ReserveProgram (CRP). The large increases impliedby this study stimulated considerable atten-tion on the topic and concern on the part ofpolicymakers. An alternative analysis using anational policy simulation model by Young,Vandeveer, and Schnepf suggested v
16、ery mod-est aggregate U.S. acreage responses to theprovision of insurance subsidies. Wu examinedthe effects of crop insurance on crop mix andinput usage for a cross section of 235 farms inthe Central Nebraska Basin in 1991 and foundthat farms that insured were more likely to pro-duce soybeans, thoug
17、h no effect on corn wasrevealed because all farms in the sample werecorn producers.Though the existing research has providedimportant insights into the effects of the expan-sion of risk management programs on produc-tion patterns, a wide gap exists in the implica-tions of existing studies. Some stud
18、ies (Keeton,Skees, and Long) point to large effects whileothers (Young, Vandeveer, and Schnepf) showmodest effects. An additional limitation asso-ciated with much of the existing research is itsfocus on national acreage and production re-sponse. In reality, production conditions andthe parameters of
19、 risk management programsare heterogeneous across regions and crops,suggesting that a focus on aggregate effectsmay conceal crop- and region-specific effects.The focus of our analysis is on a detailed em-pirical assessment of the effects of crop insur-ance program participation on acreage alloca-tio
20、n decisions among competing crops in tworelatively homogeneous growing regionstheCorn Belt and the Northern Great Plains. Thecentral goal of our analysis is to test the hy-pothesis that crop insurance programs havehad no discernable effect on agricultural landuse. We utilize estimates of structural
21、modelsreflecting the endogenous decisions of agricul-tural producers to simulate the possible effectsof large premium changes. Our results gener-ally imply very modest though statistically sig-nificant acreage effects of expanded insurancesubsidies.Modeling Framework and DataOur empirical analysis c
22、onsists of a six-equation structural model representing acre-age, insurance, conservation program partic-ipation, and input usage decisions for twoprimary crops (corn and soybeans in the CornBelt and wheat and barley in the NorthernGreat Plains). The specific equations used inour analysis, variable
23、definitions, and summarystatistics for our data are presented in table 1.Theory of the demand for insurance gener-ally considers the actions of a risk-averse agentfacing a single source of risk for which a risk-neutral insurer offers some level of protection.Stylized models that attempt to capture t
24、he es-sential elements of crop insurance plans aretypically rather simplified and thus often fail tocapture actual characteristics of the operationof the crop insurance program. For example,most crop farms are multiproduct operations.Multiple crops face an array of risks from var-ious sources. In ad
25、dition, risks are often cor-related across crops, so that yield outcomes(and thus loss events) for individual crops arecorrelated.The insurance choice (i.e., the participationdecision) is made jointly with other produc-tion decisions that must be made by produc-ers. In our analysis, we focus on thre
26、e decisions(choice variables) that are relevant to the in-surance participation decision. At the time ofplanting, a producer must decide what to pro-duce, how to produce it, and whether to par-ticipate in a myriad of government programsthat may be available. We focus on two specificpolicies that wer
27、e relevant to production deci-sions in the 1980s and 1990sthe federal cropinsurance program and the CRP. Thus, for aproducer facing the option of growing multi-ple crops, the decision involves the choice ofa level of production (acreage for each cropand input usage), the level of insurance to pur-ch
28、ase for each crop (which potentially couldbe zero), and whether or not to enroll land inthe CRP program.The focus of our analysis is empirical andthus we make no pretense as to the devel-opment of a detailed theoretical frameworkfor jointly evaluating acreage response, insur-ance participation, inpu
29、t usage, and conserva-tion program participation for multiple crops.Innes and Ardila provide a detailed evaluationof insurance participation and soil depletion ina single crop model. Our primary motivationinvolves a consideration of the extent to whichfederal crop insurance programs may have hadeffe
30、cts on the acreage decisions of producers.As we have noted above, a range of resultsexist in the literature. Though the theory ofthe demand for insurance does not providea prediction as to the sign and magnitude ofthis effect, we expect that the provision of eco-nomic benefits through insurance will
31、 result in1060 November 2004 Amer. J. Agr. Econ.expanded acreage and thus a positive relation-ship between acreage and participation.Smith and Goodwin demonstrated that fer-tilizer and chemical usage for Kansas wheatproducers tended to be negatively correlatedwith insurance purchases. That is, growe
32、rs whopurchased insurance tended to use less inputsTable 1. Model Specification, Variable Definitions, and Summary StatisticsVariable Definition Mean Std. Dev.Model SpecificationInsurance Participation for Cropi= f (Premium Ratei, Loss-Ratioi, Premium RateiLoss-Ratioi, Livestock Sales, Yield CVi, Fe
33、rtilizer, County Acres,Acresi, Annual Dummies)Acres for Cropi= f (Pricei, Acresj, Insurance Participationi, Acresi,t1, CRP Enroll-ment, Soil K-Factor, Soil T-Factor, Diverted Base, County Acres,Land Capability, Annual Dummies)CRP Enrollment = f (Rental Rate, Cost Share, Erosion Index, CRPt1, County
34、Acres,Annual Dummies)Input Usage = f (Insurance Participationi, Insurance Participationj, Acresi, Acresj,Land Capability, Annual Dummies)Heartland Corn and Soybeans (19861993)Insurance Participation (corn) Liability/maximum possible liability (corn) 0.2303 0.1758Input Usage Fertilizer and chemical e
35、xpenditures (real$thousand)/planted acre0.0612 0.0180Land Capability Proportion of land in capability classes 1 and 2 0.3091 0.1422Livestock Sales Livestock revenues/total farm sales 0.4652 0.1901Yield CV (corn) CV of historical corn yields 22.3801 9.2657Loss-Ratio (corn) Historical mean loss ratio
36、(corn) 1.7908 1.2449Acres Planted (corn) Acres planted of corn (10,000) 8.6205 5.7226Insurance Participation(soybeans)Liability/maximum possible liability (soybeans) 0.1943 0.1472Yield CV (soybeans) CV of historical soybean yields 16.4246 6.8982Loss-Ratio (soybeans) Historical mean loss ratio (soybe
37、ans) 1.6304 1.0001Acres Planted (soybeans) Acres planted of soybeans (10,000) 7.2133 4.5591Price (corn) Real corn price ($/bushel, basis adjusted priceplus USDA projected deficiency payment)3.9330 0.3546CRP Enrollment New enrollment in CRP program (10,000 acres) 0.1577 0.3619Soil K-Factor Universal
38、K-factor 0.3245 0.0460Soil T-Factor T-factor representing tolerance to soil loss 4.4354 0.4788Diverted Base (corn) Adjusted base for corn (10,000) 6.9077 4.6865Price (soybeans) Real soybean price (cents/bushel), basisadjusted price7.2636 0.8480Rental Rate Real CRP rental rate/acre 90.6399 16.9327Cos
39、t Share Real cost share payments for CRP/acre 59.1881 57.3752Erosion Index Soil erosion index 5.0777 3.3679County Acres County size (total land area in hundredthousand acres)3.3678 1.1481Premium Rate (corn) Insurance premium rate (corn) 0.0456 0.0204Premium Rate (soybeans) Insurance premium rate (so
40、ybeans) 0.0448 0.0213Revenue Liability (corn) Revenue insurance liability/total liability(corn, 19971998)0.0500 0.0210Revenue Liability (soybeans) Revenue insurance liability/total liability(soybeans, 19971998)0.0579 0.0204Revenue Premium Rate (corn) Revenue insurance premium rate (corn) 0.2735 0.17
41、73Revenue Premium Rate(soybeans)Revenue insurance premium rate (soybeans) 0.2688 0.1743than those growers that did not buy insurance.It should be acknowledged, however, thatother research by Horowitz and Lichtenbergsuggested that fertilizer and chemical inputsraise risk and thus were correlated with
42、 in-surance purchases. In more recent work, Wufound that changes in the crop mix may makeGoodwin, Vandeveer, and Deal Acreage Effects of Insurance 1061Table 1. ContinuedVariable Definition Mean Std. Dev.Great Plains Wheat and Barley (19861993)Insurance Participation(wheat)Liability/maximum possible
43、liability (wheat) 0.4699 0.2464Input Usage Fertilizer and chemical expenditures (real$thousand)/planted acre0.0298 0.0171Land Capability Proportion of land in capability classes 1 and 2 0.2585 0.1271Livestock Sales Livestock revenues/total farm sales 0.4904 0.2597Yield CV (wheat) CV of historical wh
44、eat yields 32.8363 12.6502Loss-Ratio (wheat) Historical mean loss ratio (wheat) 2.1226 1.2725Acres Planted (wheat) Acres planted of wheat (10,000) 15.7722 10.9336Insurance Participation(barley)Liability/maximum possible liability (barley) 0.3598 0.2653Yield CV (barley) CV of historical soybean yield
45、s 37.5481 16.1103Loss-Ratio (barley) Historical mean loss ratio (barley) 2.0498 1.2965Acres Planted (barley) Acres planted of barley (10,000) 4.5503 4.3995Price (wheat) Real wheat price ($/bushel, basis adjusted priceplus USDA projected deficiency payment)5.9179 0.6494CRP Enrollment New enrollment i
46、n CRP program (10,000 acres) 0.6269 1.1245Soil K-Factor Universal K-factor 0.2895 0.0395Soil T-Factor T-factor representing tolerance to soil loss 4.3733 0.5441Diverted Base (wheat) Adjusted base for wheat (10,000) 10.7269 8.2175Diverted Base (barley) Adjusted base for barley (10,000) 3.6720 3.7166P
47、rice (barley) Real barley price ($/bushel, basis adjusted priceplus USDA projected deficiency payment)3.1707 0.4611Rental Rate Real CRP rental rate/acre 46.3033 8.9150Cost Share Real cost share payments for CRP/acre 36.5884 24.9507Erosion Index Soil erosion index 3.7853 1.7765County Acres County siz
48、e (total land area in hundred thousandacres)10.1749 6.0275Premium Rate (wheat) Insurance premium rate (wheat) 0.0671 0.0212Premium Rate (barley) Insurance premium rate (barley) 0.0883 0.0299the overall relationship between insuranceparticipation and fertilizer and chemical us-age less clear. If insu
49、rance encourages shift-ing toward crops with more demanding inputrequirements, insurance participation may ac-tually increase fertilizer usage. Thus, the ex-pected relationship between insurance partic-ipation and input usage is unclear.Our empirical analysis makes use of pooledcross-sectional, time-series data collected atthe county level. We utilize a wide variety ofsources to obtain county-level data. We focusour analysis on the years 19851993, thoughwe also consider an extension to 1997 and1998, a period characterized by a different mixof programs. It is important to recognize t
