Linear programming versus positively estimated supply functions: An empirical and methodological critique

This paper examines whether supply relations derived from a static profit-maximizing linear programming model can be used as information to explain supply response with at least the same reliability as partial adjustment regression models. Programming estimates of direct supply elasticities of California commodities are compared with estimates from time-series regressions. Differences on individual crops are considerable, but on average the elasticity estimates are comparable. Programming parameters are combined with time-series data to test whether imposing such relative parameters on time-series models need not significantly reduce their predictive ability. This imposition does not reduce predictive ability or significantly improve it. Key words: linear programming, prediction, regression, supply functions. This paper evaluates the reliability of linear programming (LP) as a method of estimating supply elasticities and predicting future pro-duction levels in a situation where economet-ric tools are viable alternative methods. L