Intertemporal risk management decisions of farmers under preference, market, and policy dynamics (China)

Three separate, but related chapters of this dissertation examine the risk management issues related to dynamic stochastic agricultural production, following the introduction of the whole dissertation in Chapter 1.; Chapter 2 adapts a generalized expected utility (GEU) maximization model (Epstein and Zin, 1989 and 1991) to examine the intertemporal risk management of wheat producers in the Pacific Northwest. Optimization results based on simulated data indicate the feasibility of the GEU optimization as a modeling framework. A comparison between the GEU and other expected utility models further implies GEU has the advantage of specifying farmers' intertemporal preferences separately and completely.; Based on the GEU framework, Chapter 3 examines the impacts of farmers' risk aversion, time preference, and intertemporal substitutability on their optimal risk management decisions. It further extends the GEU model by incorporating a welfare measure, the certainty equivalent, to investigate the impacts of U.S. government programs and market institutions on farmers' risk management decisions and welfare.; Results imply that farmers' optimal hedging is sensitive to changes in the preferences and the effects of these preference changes are intertwined. Target price and loan rate levels, offered by certain government payment programs, can lead to the substitution of government programs for hedging. The evaluation of current risk management tools shows both crop insurance and government payments can improve farmers' welfare significantly. Government payment programs have a greater effect on farmers' welfare than crop insurance and crop insurance outperforms hedging.; Chapter 4 explores the market integration of Chinese wheat futures to the world. It compares the price behavior of China Zhengzhou Commodity Exchange (CZCE) with that of the Chicago Board of Trade (CBOT) in the US using ARCH/GARCH-based univariate and multivariate time series models and cointegration analysis. Results show both markets can be modeled by ARCH(1)/GARCH(1,1) and the models have a better fit when the conditional error variance is t distributed. The price series in CZCE and CBOT are interrelated but not cointegrated. The existing interrelations between the two markets are significant and asymmetric. CBOT holds a dominant position in the interactions while CZCE behaves like a follower.