Essays in Currency Markets

This dissertation is comprised of three chapters, all with the aim to better understand currency behavior. The first chapter brings well-known, profitable currency trading signals to bear on the Meese-Rogoff puzzle (Meese and Rogoff (1983)), a robust finding regarding the difficulty of outperforming the random walk in forecasting nominal exchange rates in an out-of-sample fashion, particularly at short horizons, using macro aggregates. Using carry, momentum, and value trading signals, which resemble macro variables that have traditionally been considered in forecasting exchange rates, we find much more evidence of out-of-sample forecastability at short horizons when the signals are used in a cross-sectional, multilateral manner than in a time-series, bilateral manner. This suggests macro fundamentals may be more informative of exchange rates than previously thought, if considered jointly rather than pairwise as has conventionally been done. A caveat is that the forecast noise stemming from limited data availability renders both time-series and cross-sectional signals' outperformance against the random walk unreliable at short horizons in practice. The second chapter revisits the question of whether cross-sectional differences in currency excess returns can be accounted for by downside market risk. Estimating the downside risk model proposed by Burnside and Graveline (2014), a piecewise linear factor model that subsumes previous downside risk models of currencies, we find that downside market risk does a poor job in explaining the cross-section of currency portfolio returns, regardless of whether we use an equity-based or currency-based market factor. Even if the model were to price the cross-sectional returns well, we would still face counterintuitive factor model estimates that go against the basic assumption of risk aversion. That is, we find a downside market risk puzzle for currencies. The third chapter examines the relationship between fiscal sustainability and nominal exchange rate. Using the annual change in debt to GDP ratio as our measure of fiscal sustainability, we find that fiscal sustainability contains predictive information on exchange rate in both the time-series and cross-sectional dimensions, controlling for potentially confounding macroeconomic factors, and for up to three years. Impulse response estimates show that a shock to fiscal sustainability, i.e., an unanticipated increase in the debt ratio, induces a contemporaneous depreciation in exchange rate lasting for three to five years. All in all, we find strong evidence of a linkage between fiscal sustainability and nominal exchange rate.