International Portfolio Investments with Trade Networks (Job market paper)

Paper  Appendix 

Abstract. What determines international portfolio investments remains an open question in international finance. In this paper, I solve in closed-forms for the optimal equity and bond portfolio investments in a multi-country model with arbitrary global input - output linkages as well as cross-country taste differences. I show that an important determinant of international equity portfolios is a measure of pairwise trade in value-added that is a function of the international trade networks. This measure extends the closed-economy “Domar weights” to the international setting and captures countries' interdependence through both direct and indirect trade linkages.

Using data from the World Input - Output Database (WIOD) and Coordinated Portfolio Investment Survey (CPIS), I apply the framework to a network of 43 major developed and emerging economies and find four main results. First, the theoretical network portfolio is a significant predictor and explains almost half of the variation in international bilateral portfolio investments. The significance of the network portfolio is robust to controlling for gravity factors (e.g. market capitalization, distance, and common language). Second, including the network-based portfolio in a gravity model for assets removes the effect of distance for asset holding, alleviating the puzzle of why distance matters to asset trade at all. Third, indirect trade linkages matter for portfolio determination, highlighting the need to explicitly account for intermediate inputs trades. Finally, I show that the changes in trade network structure explains the cross-section of the decline in equity home bias that has occurred after 2000.

Behavioral-Attention Phillips Curve: Theory and Evidences from Inflation Surveys

Paper  Appendix

Abstract. I derive a theory of endogenous uncertainty and attention choice that can jointly account for two recent phenomena: (i) the flattening Phillips Curve and (ii) “well-anchored” inflation. In particular, I derive a Behavioral Attention Phillips Curve (BAPC) whose slopes on the output gap and inflation expectations decline when inflation is less uncertain. When inflation uncertainty is low, firms find it less costly to misperceive aggregate demand and inflation expectations, thus pay little attention to monetary shocks and change prices less. The dampened price response flattens the Phillips Curve. Inflation becomes more anchored with low uncertainty because costly attention moti- vates firms to rely more on “rules-of-thumb” such as the 2% inflation target.

Using novel measures of inflation uncertainty constructed from surveys of infla- tion expectation, I show that the new Phillips Curve performs better both in-sample and out-of-sample than the traditional Phillips Curve with constant slope. Particularly, the BAPC does not generate the counterfactual prediction of large disinflation after the 2008-2009 Financial Crisis as does traditional Phillips Curves, resolving the Missing Disinflation Puzzle. I show that multiple equilibria arise with medium volatility due to the comple- mentarity between pricing and attention choices. This gives rise to a novel policy paradox for the Central Bank and makes it hard to raise inflation in a quiet, low volatility period.

Research in progress

A network model of currency growth (joint with Kenneth Rogoff, Ethan Ilzetzky) [draft coming soon]