uncovered interest rate parity (UIP) hypothesis suggests that this ex-ante arbitrage opportunity should not exist, meaning that the carry gain should be offset by a corresponding depreciation of the investment currency, the reverse is often found to be the case in what is referred to as the “forward premium puzzle,” see, e.g. Bansal (1997), Bansal and Dahlquist (2000), Verdelhan (2006), Burnside, Eichenbaum and Rebello (2007).
Deviations from UIP, even if persistent, could be reconciled with the observation that dramatic exchange rate realingments (crashes) are often observed when the carry trade unwinds in the absence of obvious shifts in macroeconomic fundamentals. Therefore, persistent carry trade opportunities may be seen as commensurate compensation for risk (see, e.g. Burnside, Eichenbaum, Kleshchelski and Rebello, 2008; Brunnermeier, Nagel and Pedersen, 2009; and Jordà and Taylor, 2009).
This paper investigates the relationship between the carry trade and exchange rate volatility with this backdrop. Our analysis is primarily empirical but with an eye toward some of the theoretical issues we have raised. Specifically, Brunnermeier and Pedersen (2009) suggest that persistent carry trade returns reflect a liquidity premium. Moreover, they argue that negative skewness arises when speculators hit funding constraints that amplify their losses by forcing them to unwind their carry trade positions. Thus, to the extent that volatility reflects fluctuations in market liquidity, we investigate the relationship between carry trade and exchange rate volatility. However, Jordà and Taylor (2009) find that while purchasing power parity (PPP) is a poor predictor of short-run exchange rate fluctuations (see the voluminous literature that followed Meese and Rogoff's 1983 seminal paper such as Rogoff, 1996; and Taylor 2002), it seems to be a good signal of an impending realignment of the exchange rate. As a consequence, we are interested in separating this effect from our investigation of exchange rate volatility on carry trade fluctuations.
We focus our investigation on recent data for the period December 2000 to March 2009 and therefore include the dramatic episodes in the fall of 2007 and thereafter. The short sample reflects data availability because, unlike other studies (e.g. Burnside et al. 2007, Burnside et al. 2008, Brunnermeier et al. 2009, and Jordà and Taylor, 2009), our focus is purposefully tilted toward Asian economies. Specifically, the