(Extract from page 37 of BIS Quarterly Review, December 2011)
Research on the economic drivers of carry trade returns has seen significant advances over the past several years. It has been established that it is difficult to explain carry trade returns purely as compensation for risk exposure with standard risk factors - that is, conventional asset pricing models based on covariance risk with, for instance, the broader market or business cycle factors (Burnside et al (2011a)). This has led researchers to emphasise aspects such as funding market constraints and crash risk (Brunnermeier et al (2009)), and to argue that currencies share a common risk factor (Lustig et al (2011) and that carry trade premia are compensation for systematic volatility and liquidity risks (Menkhoff et al (2011a)).1
In addition to work based on observable risks, an alternative explanation is that carry trade returns might be a compensation for the risk of rare disasters with significant losses which do not occur in-sample (Burnside et al (2011a)).2
Whereas the literature on carry trades is meanwhile quite extensive, much less is known about the potential drivers of currency momentum. 3 This is especially so for FX momentum strategies relying on a broad cross section of currencies that have been introduced more recently. Recent empirical studies suggest that currency momentum returns cannot be successfully explained by the risk types that seem plausible for carry trades (Burnside et al (2011b), Menkhoff et al (2011b)). This research also documents that the anatomy of carry trade returns is very different from that of currency momentum returns.
There is evidence that momentum returns in part reflect the gradual incorporation of news into prices and a resulting return drift, as shown in Menkhoff et al (2011b).4 In addition, this research also points to country-specific risks, transaction costs and other forms of limits to arbitrage as likely explanations for the continued presence of momentum returns. Our finding of substantial downside risks for the currency momentum strategies presented in the main text squares well with this explanation. Following momentum strategies can expose investors to potentially painful short-term losses, as illustrated in the main text. This may discourage market participants from taking aggressive positions to trade momentum profits away (Shleifer and Vishny (1997)). Arbitrage capital might therefore move slowly, which could possibly explain why an apparent market anomaly like FX momentum continues to exist (Duffie (2010)).
1 There is also empirical evidence that carry returns co-vary more strongly with the equity market in volatile periods (Christiansen et al (2011)).
2 This explanation is often referred to as the peso problem (Krasker (1980)).
3 Okunev and White (2002) were to our knowledge the first academic researchers to document the profitability of momentum strategies relying on a broad cross section of currencies. Most other earlier research on trading strategies which exploit short-term trends focused on individual exchange rates. Menkhoff and Taylor (2007) and Neely and Weller (2011) comprehensively review this literature on so-called technical trading rules. A major aim in much of this work has been to determine which rules work best and how stable they are over time (Neely et al (2009)).
4 This finding - which was first established for equities by Jegadeesh and Titman (2001) - suggests that momentum profits across asset classes may share a common root.