Testing for limited asset market participation, consumption, and real exchange rates: Evidence from household data

Standard models of open-economy macroeconomics predict a strong correlation between real exchange rates and relative consumption across countries. This prediction fails overwhelmingly when confronted with data. A promising way to break the link between real exchange rates and aggregate consumption is to account for heterogeneity across households due to differences in asset market participation status. This dissertation studies the equilibrium condition that arises when only a fraction of households participate in asset markets in each period. In such an environment, real exchange rates are related to the consumption of the households that participate in asset markets instead of the aggregate consumption.; The empirical work starts with an analysis of the asset market participation decision. Estimation of a probit model using information on Italian households reveals that factors such as age, education, occupation, and gender of the household head as well as geographical area, of residence, and time contribute significantly to this decision.; Next, the equilibrium condition that relates real exchange rates to relative consumption of the asset market participants is tested for twenty administrative regions of Italy. Empirical work utilizes micro data on household consumption and financial assets from the Survey of Household Wealth and Income collected by the Bank of Italy. Real exchange rates are constructed using regional price indices. The nonlinear GMM estimation of moment conditions provides evidence in favor of the model's implications. In particular, overidentification restrictions of the model are not rejected at conventional levels of significance. The evidence on the estimates of the risk aversion parameter a is more mixed yielding positive and statistically significant parameters for three regions when asset holding status is based on the predictions from the probit analysis. Consumption figures for actual asset holders slightly improve the results to yield positive and statistically significant parameter estimates for four regions. There is stronger evidence from panel data with positive and statistically significant parameter estimates for nine regions. The empirical evidence presented in this dissertation underlies the role of household heterogeneity resulting from limited asset market participation toward a thorough understanding of the consumption-real exchange rate puzzle.