Representative agent models have been a useful paradigm for macroeconomics for at least twenty years. Despite their many uses, there are well-known deviations between their predictions and macroeconomic data. For example, representative agent models cannot account for the large difference between stock and bond returns observed in the United States and other countries (at least with standard formulations of preferences). This proposal asks the following question. Representative agent models essentially assume that the allocation of resources across people in the economy is Pareto optimal. Can models in which allocations are constrained Pareto optimal do a better job empirically?
In terms of intellectual merit, the proposal brings together two previously disparate fields of inquiry: dynamic contracting and empirical macroeconomics. There is a large theoretical literature on the properties of Pareto optima in dynamic settings with private information. There is a large empirical literature on the failings of the representative agent model especially in regards to asset pricing. This project uses a new result in the former literature to fix remaining problems in the latter. The novel ingredient in building this connection is the use of cross-sectional data on consumption. The data used come from the Consumer Expenditure Survey (CEX). The CEX has only a limited panel dimension. However, the evaluation of the empirical implications of the theory does not require high-quality longitudinal data on consumption, but only a time-series of cross-sections - which is exactly the content of the CEX.
In terms of broader impact, the results of this research will be disseminated broadly through scientific journals and (in simplified form) through general audience publications like the Federal Reserve Bank of Minneapolis Quarterly Review. The societal consequences are potentially large. It is natural to ask whether the empirical failings of the representative agent model imply a role for government interventions of some kind. The answer to this question is negative if the proposed empirical work finds that these failings can be rationalized through private information considerations.
