Abstract: We extend the theory of Gabaix and Maggiori (2015a, 2015b) to study currency risk premiain a multi-country world with imperfect financial markets. Currency returns are connectedto financiers’ limited commitment, captured by the complexity of their balance sheets in thetrade imbalance network. Guided by the theory, we construct a Centrality Based Characteristic(CBC), based on the centrality of the imbalance network and variance-covariance of currencyreturns. Sorting currencies on CBC generates a high Sharpe ratio, and the resulting excess returnscannot be explained by standard currency factors and intermediary asset pricing factors,suggesting a novel source of currency predictability.
Thomas Mertens, Federal Reserve Bank of San Francisco
Jingye Wang, Renmin University of China
Abstract: Canonical long-run risk and habit models reconcile high equity premia with smooth risk-free rates by inducing an inverse functional relationship between the variance and the mean of the stochastic discount factor. We show this highly successful resolution to closed-economy asset pricing puzzles is fundamentally problematic when applied to open economies with complete markets: It requires that differences in currency returns arise almost exclusively from predictable appreciations, not from interest rate differentials. In the data, by contrast, exchange rates are largely unpredictable and currency returns differ because interest rates differ widely across currencies. We show that no complete-markets model with canonical long-run risk and habit preferences can match this fact. We argue this tension between canonical asset pricing and international macroeconomic models is a key reason why researchers have struggled to reconcile the observed behavior of exchange rates, interest rates, and capital flows across countries. The lack of such a unifying model is a major impediment to understanding the effect of risk premia on international markets.
Discussant: Riccardo Colacito, University of North Carolina-Chapel Hill
Abstract: We introduce the nonlinear arbitrage correction (NAC) as the residual thatrenders a linear benchmark model for basic assets arbitrage-free. Return datafor several economies reveal that NAC is countercyclical, related to financialuncertainty, and foreign exchange option returns, both in- and out-of-sample.We find that NAC predicts future market dislocations, including covered interestrate parity deviations, particularly out-of-sample. We show that conditionallinear asset pricing models perform well on average and during normal times,while they imply larger NAC during crises.
Discussant: Zhengyang Jiang, Northwestern University
