Schwartz and Tomz (1997) have correctly pointed out an error in our 1987 article: We had neglected to analyze how changes in group composition improve the performance of centralized institutions over time. The change affects only the case of perfect monitoring, however. We argue, moreover, that even in this special context, our main qualitative conclusion—that the centralized structure has no positive advantage over the decentralized one—continues to hold. We reach different conclusions about the relative roles of the two institutions, partly because we had in mind a positive interpretation, whereas Schwartz and Tomz select a normative interpretation of the issue of institutional choice. Finally, we believe that imperfect monitoring is essential to the theory, in order to derive conclusions that are not driven by artifacts of the model or by arbitrary equilibrium selections.