Production risk and decision making: Testing alternative econometric models with evidence from Egyptian cotton production

Abstract

The major objective of the study is to develop and apply theoretical and empirical methods for evaluating plternative stochastic production models and assessing the degree to which incorporating production risk in decision models helps explain farmers' behavior. The study compares and contrasts risk-responsive and risk-neutral models derived from both the traditional specification and the stochastic production risk model proposed by Just and Pope (1978). The traditional model implies that the marginal effect of an input on output variabi Iity has the same sign as the expected marginal product. In contrast, the Just and Pope model allows the inputs to have different effects on the mean and the var.lance

of output.

Using a sample of cotton farm data from Egypt, Cobb-Douglas versions of the traditional and the Just-Pope pro~uction'models are estimated. The traditi~nal model estimates were judged satisfactory except for

a few variables. The mean production function estimates of the .lust-Pope model indicate that all inputs except two are mean production increasing. The output variance estimates show that five inputs increase variability of output and five inputs reduce variability.

Non-nested hypotheses tests ~ere performed on four decision Inodels -...... representing farmers' behavior under risk-response and risk-neutrality

I i

within the traditional and the Just-Pope frameworks. These tests reject all models on the basis of their ability to predict input use. In contrast, however, some of these models seem to perform adequately based on conventional R2, F and t statistics. Another notable result is that the models produce different risk coefficient estimates. The study concludes that for input-use prediction, not much dif­ ference exists between any of the models whether or not farmers are asswned to be risk averse. However, in measuring the degree of risk aversion and in measuring the effects inputs have on output variance, the Just-Pope model is to be preferred to the multiplicative error model. However, this research found that more work is needed to develop a satisfactory model for analyzing production risk.