Skip to main content
Think and Do The Extraordinary
Support the Unit

Zhaochen Sun

Graduate Student

4305 Nelson Hall

Dissertation Chair: Barry Goodwin

We utilize county level crop insurance data from 1981 to 2019 and consider various measures of loss risk, including the Loss Cost Ratio and Loss Ratio. This is matched to extensive weather data obtained from the National Climate Data Center of the National Oceanic and Atmospheric Administration (NOAA). This includes monthly averages of temperature (averages and extremes), a range of drought indexes, heating and cooling degree days, soil moisture indicators, historical yields, and other weather outcomes. LASSO estimation allows a data-driven approach to determining those variables that provide the best out of sample predictions of outcomes of interest. The approach allows one to consider cases where the number of potential covariates exceeds the number of observations. Once appropriate weather variables are identified, the crop insurance data are conditioned on weather and the extent of unexplained loss variability is evaluated. Of principal concern is the extent to which market and policy variables appear to be related to loss outcomes once exogenous weather variability is accounted for. If higher market prices, price changes in the early growing season, and higher input costs appear to be related to loss outcomes, one may infer that moral hazard may be a relevant causal factor in determining losses.

Our empirical application is to crop insurance experience in major growing regions of the US Corn Belt for corn and soybeans. We utilize county-level weather and crop insurance data and focus on yield coverage, which was prominent through the 1990s, and revenue coverage, which is the predominant form of crop insurance coverage in more recent years. We also utilize cause of loss data to focus on weather-related losses. Preliminary results find that losses are indeed significantly impacted by changes in important economic variables, such as output and input prices and price changes following planting but before crop emergence. We quantify these relationships and discuss implications for the fiscal stability of the US crop insurance program.