A Risk-Based Liquidity Theory of International Currency (2025, draft, slides)
This paper develops a risk-based liquidity theory of international currency, grounded in the endogenous ranking of currencies as media of exchange. These rankings emerge from their asymmetric hedging properties against counterparties' future income shocks within an open-economy search model featuring multiple currencies. The model rationalizes various currency payment patterns as equilibrium outcomes, including single-currency dominance and the coexistence of multiple currencies. Empirical evidence supports the model’s prediction that the share of the dollar in international trade settlements and currency holdings is more prevalent in countries with more procyclical valuations of local currencies against the US dollar. The calibrated three-region model successfully replicates the observed dominance of the US dollar in the U.S. and Latin American regions due to its superior insurance properties against local economic conditions, alongside the local adoption of the euro in the Eurozone. Furthermore, the model underscores the significant welfare costs of de-dollarization policies, which result from the deterioration of terms of trade caused by the exclusion of optimal foreign currency payment arrangements, particularly as risk aversion increases.
A Macro-Finance Model with Ambiguity as Partial-Identification Problems (2024, draft, slides)
Previously circulated as "Model Uncertainty as Partial-Identification Problems: Application to Policy Promises during Crises"
I present a general equilibrium model of intermediary asset pricing where investors are struggling with uncertainty represented by a set of models about future asset return distribution. Each model is consistent with observable information and the understanding of the economic structure but has different implications for return distribution. Confronting such uncertainty, investors fear models predicting low future returns, reducing asset demand due to uncertainty aversion and amplifying the risk premium, especially during crises when capital within the intermediation sector is scarce. Following validation of subjective beliefs by various survey expectations, I evaluate credible policy promises that eliminate some adverse models from investors' set as inconsistent with announcements. I demonstrate the efficacy of announcements that eliminate pessimistic prospects for dividend growth and restore risk appetite. Methodologically, I develop agents' inference framework from endogenous variables, where subjective beliefs and other equilibrium dynamics are jointly determined.
Dynamic Portfolio Choice under Nonlinear Dynamics, (2022, draft)
This paper investigates a nonlinear dynamic portfolio choice of an investor endowed with recursive preferences. I estimate a quadratic autoregressive (QAR) process for the evolution of the investment opportunity set, including the aggregate vacancy rate as a return predictor. Jointly incorporating multivariate nonlinear dynamics in conditional means and stochastic volatility substantially improves standard measures of portfolio performance relative to using a linear time series model: the cumulative wealth path between 1972 and 2014 by 93%, Sharpe ratio by 20%, and utility-based economic welfare by 48%. Methodologically, I derive an analytical approximation of the optimal dynamic portfolio in a multivariate nonlinear environment embedded in the QAR model.
The First Arrow Hitting the Currency Target: A Long-Run Risk Perspective, with Takashi Kano (2017, Link), Journal of International Money and Finance, 74: 337-352.
This paper reconsiders the successful currency outcome of the first arrow of Abenomics. The Japanese yen depreciation against the U.S. dollar after the introduction of the first arrow co-moves tightly with long-term yield differentials between Japan and the United States. The estimated term structure of the sensitivity of the currency return of the Japanese yen to the two-country interest rate differential indeed shifts up and becomes steeper after the onset of Abenomics. To explain this structural change in the term structure of the Fama regression coefficient, we employ a long-run risk model endowed with real and nominal conditional volatilities as in Bansal and Shaliastovich (2013). Under a plausible calibration, the model replicates the structural change when nominal uncertainty dominates real uncertainty in the U.S. bond market. We conjecture that the arrow was shot off from the U.S. side, not the Japan side.