A rejoinder to Silberman

Abstract

Jonathan Silberman's comment makes some contributions to the empirical analysis of campaign funds. In particular he uses probit analysis to estimate the combined electoral production function, in which money is an input and probability of election is the output. However, several of his points involving functional form, specification of variables, and a policy conclusion should not go unchallenged. Silberman first estimates the production function in which money is an input and expected vote percentage is the output. For a functional form he chooses a Cobb-Douglas production function to replace my choice of linear and double-log forms. He argues that because a campaign is a zero-sum game, the value of the output is unaffected by proportional changes in inputs. While palusible this conclusion is a non sequitur; symmetry in the outputs of two candidates' production functions does not logically imply symmetry in the input-output relationship. Nor do campaigns always have such symmetry. For instance, proportional increases in the amount spent to get out the vote usually aid Democratic candidates over their Republican opponents. Thus, a functional form which assumes symmetry is not "more theoretically appropriate" (italics mine). From the estimates of the Cobb-Douglas function, the elasticities of the ratios of inputs can be read directly. Unfortunately, the marginal products (MP) of money