Rational Mean-Variance Decisions for Subsistence Farmers

This paper explores the issue of approximating expected utility in applying portfolio theory. It has been demonstrated that expected utility is very closely approximated by an appropriate quadratic function. Recent studies have again questioned the empirical validity of mean-variance analysis. These efforts disregard the importance of the differences among various approximations, and historically, have focused on domestic financial securities. We re-examine the problem in the context of determining risk-efficient portfolios of production activities for subsistence farmers. The risks faced by these farmers are large and probably actuarially nonneutral. Thus, recommendations based on an inappropriate quadratic approximation could result in substantial losses in welfare.