The confusion in this paper appears to be caused by the author's failure to appreciate fully a lengthy footnote (footnote 5) in my article. In this note a distinction was made between announcing or failing to announce a mixed strategy. The results from either of these approaches will be Pareto optimal ex post but not ex ante. Thus, if a price system works (which it usually does not), the citizens ex ante should prefer it to the vote unless they are risk positive.

My paper had the narrow goal of pointing out that no attempt had been made to compare the price system and the voting mechanism in the case where the price system is efficient. It then showed that given risk aversion, it was possible that society would prefer the price system. The weaknesses in the article were in the model—particularly in specifying how to embed it in a dynamic context and how to give a mixed strategy (concealed or declared) a more plausible interpretation. My model was adequate for the type of counter-example I was constructing. I would have been happier had the author “bought” far less of my model and tackled the problem of improving its sophistication for a more constructive purpose.

First, Shubik's comment distinguishes between ex ante and ex post judgments about markets and elections. He claims that given his assumptions, both markets and elections are efficient ex post, though elections may not be efficient ex ante. I agree insofar as in my last paragraph I state, “If … we ask … ‘For some forthcoming sequence of one or more social decisions, do citizens prefer elections to markets?’ our answer is ‘Possibly no.’”

But Shubik's essay contains no distinction between ex ante and ex post judgments. Shubik begins, “it is demonstrated that if all the conditions for the existence of a competitive equilibrium are satisfied, then simple majority voting to determine the distribution of goods may be less efficient than a price system” (page 179). Shubik notes later, “if they [voters] are riskneutral or risk-averse, then, whenever the ‘political noncooperative game’ has no pure strategy solution, the result will not be Pareto optimal” (page 180). These two statements indicate that Shubik's judgment is ex post and not ex ante, and ex post he is wrong.

Second, to complicate matters further, Shubik fails to distinguish correctly between mixed and risky strategies by the dictates of these concepts' theoretical definitions. I show that if these concepts are correctly interpreted, then a Pareto efficient allocation occurs no matter what form the candidates' electoral strategies take. Manifestly, an ex ante comparison of markets and elections by the Pareto efficient criterion is meaningless, as ex post they are both Pareto efficient.