Comparison of production risks in the state-contingent framework: application to balanced panel data

Abstract

In a balanced panel data setting, this article proposes an empirical application of the state-contingent (SC) framework for production uncertainty. The SC approach (e.g., Chambers and Quiggin 2000) casts production decisions under uncertainty as the decision to select a portfolio of Arrow-Debreu SC outputs, scheduled to be delivered in the contingent states of nature. Under some stationarity assumptions on the SC decisions (i.e., no technical change, time-invariant states of nature, time-invariant SC portfolio decisions) and regularity assumptions on the data generating process (i.e., cross-sectionally homogeneous state realizations), SC technology can be estimated from balanced panel data that are framed as cross-sectional data of partially-revealed SC portfolio decisions. This allows one to simulate an optimal SC portfolio, determined by the interaction between the estimated SC technology and given risk preferences. In the application to Maryland dairy production data, the stochastic technologies of confinement and intensive-grazing dairy systems are compared. Of the two time intervals (years 2000–2004 and ye0ars 2006–2009) separately analyzed, the optimal production decision has generally become riskier for the confinement system and less risky for the grazing system. These contrasting trends appear directly related to the volatile milk prices, feed cost hikes, and increasing organic milk production during 2006–2009. The risk associated with the optimal portfolio is substantially lower under the SC analysis compared to a typical residual-as-uncertainty approach, suggesting that the typical approach may overstate the risk due to uncertainty.