Dynamic Heterogeneous Agent Models of Default on Farm Real Estate Loans

According to the USDA ARMS data, farm real estate values have increased almost threefold since 1987, but this trend is leveling off. The Federal Reserve Bank of Kansas City reported that both irrigated and non-irrigated farmland was trending negatively in the second and third quarters of 2015. This phenomenon not only happens in district 10, but reports from districts 11, 8, and 7 of the Federal Reserve are also indicate a downward trend in farmland value. It is likely that the value of farm real estate, especially in middle America, is just beginning a downward slide. This decrease in land value is correlated with low commodity prices and low expected returns from the agricultural sector. According to the USDA, net U.S. farm income tumbled 38% to $55.9 billion in 2015, the lowest in more than a decade (Newman, 2015). Agriculture is by nature a cyclical industry. In the 1980s, the bust of the agricultural economy resulted in an increase in farmer defaults and agricultural bank failures. In 1985 and 1986, agricultural banks charged off $2.5 billion in loan loss, and 50 agricultural banks failed each year from 1985 to 1987. Therefore, banks and shareholders are very interested in whether the decline in farmland prices and weak agricultural profitability will cause another agricultural credit crisis. The agricultural credit crisis in the 1980s and the current agricultural economy expectations highlight the importance of understanding the economic mechanisms triggering agricultural loan default and the rise in charge-off rates. Insights into these issues may then inform political debates on how to prevent future foreclosure crises or mitigate their impacts if they must happen. To date, a clear lack of structural theory on farm real estate loan default behavior exists. This paper contributes to this research agenda by developing a heterogeneous agent model to study the effects of a farmland price shock and commodity price shock on the default decisions of farmers. Findings from simulations of this structural model can help policy-makers understand the mechanisms of farmland loan default