Production risk and optimal fertilizer rates: A random coefficient model

This paper demonstrates that low levels of fertilizer use on rainfed rice in the Philippines cannot be attributed to production risk. A random coefficient model estimated the objective probability distribution of yield. This was incorporated into a utility-maximization framework to predict hat moderately risk-averse farmers would apply only seven to ten kilograms less than the profit-maximizing N-rate. Previous studies have established that risk was nota major impediment tofertilizer use in irrigated areas. This paper extends this conclusion to rainfed rice production in the Philippines. Key words: fertilizer, Philippines, production risk, rainfed rice, random coefficients. Risk as an impediment to progress is a recur-rent theme among development practitioners (Wharton, Galbraith, Lipton). In the context of the new rice technology it has been argued that the full potential of modero varieties (MVs) has not been realized because year-to-year variability in crop response has caused farmers to use less than the economically op-timal level of inputs (Barker et al.). Over the years a growing body of literature has emerged, quantifying the effect of risk aversion on input use, particularly fertilizer. The evidence is conflicting and the contro-versy continues. While Moscardi and de Jan-vry, and Ryan and Perrin conclude that risk aversion substantially lowers input levels, Roumasset, and Rosegrant and Herdt show that the low observed levels of input use can-not be att ¡ to risk. Further empirical evidence on this issue is provided in this paper. Objective probability distributions of rice yields at different nitrogen (N) rates were estimated using a random coefficient model (RCM). These were incorpo-rated into risk-neutral and risk-averse deci-sion-making models to estimate the effect of production risk on nitrogen application. Joyotee Smith and Gloria Umali are a visiting associate economist and a research assistant, respectively, at the International Ric