Content area
Full Text
WHAT CAUSES CROP FAILURE?
ROBERT MENDELSOHN
Yale FES, 230 Prospect Street, New Haven CT 06511 E-mail: [email protected]
Abstract. One of the banes of farming is the frequent complete loss of crops due to adverse weather conditions or pests. In this paper, we explore what causes catastrophic crop failures. The study demonstrates that 39% of the variation in expected crop failure rates across the United States can be explained by soils and climate. The analysis shows that precipitation, soils, and especially temperature all explain average crop failure rates. Surprisingly, the analysis did not suggest that annual warming would increase crop failure rates. However, decreases in annual precipitation or increases in interannual or diurnal variation would all be harmful to crops.
1. Introduction
Although there is an extensive literature on crop insurance (for example, Knight and Cobble, 1997; Wu, 1999; Young et al., 2001), there are surprisingly few studies of the actuarial risks of crop failure. The crop insurance literature has focused primarily on measuring the size and inuence of government subsidies for crop insurance. Much less attention has been given to why crops fail and how these failure rates vary consistently across counties. The popular press reports crop failures due to lack of precipitation (drought). However, empirical studies that systematically link long term precipitation and average crop failure rates are difcult to nd. This paper explores how climate affects long-term crop failure rates in counties across the United States.
This study performs an empirical analysis to explain why some counties have higher and some have lower average crop failure rates. Crop failure rates for American counties in 1978, 1982, 1987, 1992, and 1997 are gathered from the US Census of Agriculture (USDA). Crop failure is dened as the complete loss of crops on a farm. In 1997, 3.6 million acres in 57,000 farms experienced crop loss failure. The expected failure rate for a county is computed from the average of the failure rate (failed cropland/all cropland) across these ve Census years (spanning 20 years). We wish to test what explains these long-term failure rates. Combining failure rates with soils and climate data (Mendelsohn et al., 1994) yields a complete cross-sectional data set. The expected crop failure rate is then regressed on soil and climate variables. The...