Frictions, Heterogeneous Preferences and International Consumption Risk Sharing. 2010.
In contrast to goods market frictions, the role of financial market frictions is not well understood in the international macroeconomics literature. This paper provides evidence on the significance of financial frictions in explaining the lack of international consumption risk sharing, in addition to goods market frictions. Using the gravity estimation approach developed by fitzgerald (2007) to measure frictions, the paper augments the model to allow for heterogeneous preferences in a multi-country setting. Consequently, estimating the gravity equation requires a new set of asymmetric importer-exporter dummies. In contrast to earlier studies, results from the likelihood ratio tests reject the null hypothesis of frictionless financial markets within a group of 22 developed countries after controlling for the heterogeneity across country pairs. However, the tests associated with high dimensional gravity equations have a large size distortion (the type I error). Hierarchical Bayesian estimation can control for the large dimensionality of the parameter space and reduces the size distortion. The Bayesian results select the model with both financial and goods market frictions over the model with goods trade costs only. Furthermore, I find a large consumption home bias for each country, decreasing goods trade costs, and a negative relationship between the risk sharing level and the volatility of output growth across countries.
Hierarchical Bayesian Method and Gravity Equations. 2010
Using panel data with multilevel structure, there are many ways in which fixed effects may be included to control for specific features; estimates of key parameters can vary considerably with the choice of those dummy variables. This model selection issue cannot be easily answered with conventional likelihood ratio tests and Wald tests because the high dimensionality of parameter spaces leads to a large size distortion (the type I error). Here the hierarchical Bayesian method is applied to reduce the over-parametrization problem and to help choose models more credibly. Particularly, this study estimates gravity equations for the Euro Zone effect on trade to illustrate how the currency union effects on imports can be estimated distinctly, why the size distortions arise, and how the Bayesian method helps alleviate the pitfall.
Asset Holdings, Financial Frictions and Heterogeneous Firms. with Huiran Pan, 2010.
Motivated by the positive relationship between bilateral equity holdings and goods trade, the paper studies the effect of financial frictions on U.S. imports and foreign equity holdings from 2001 to 2007 based on micro data. We evaluate how credit constraints on heterogeneous firms determine the relationship between U.S. sectoral imports and equity investments for a group of 43 countries, along with traditional trade costs. Using an IV-Tobit, IV-PPML, and IV-Probit models to control for endogeniety problem and many zero values in the data, we find significant negative effects of financial frictions on U.S. equity holdings and imports. The effects are more prominent for sectors with high financial dependence and high short-term debt interest repayment constraint. We also find a sector level positive but not universally robust relationship between trade and equity holdings, compared with the pervasively positive relationship in the early literature using aggregate country level data. Finally, we support that productivity is the main driving force for firms to export and cross-list in the US stock market.