rates and/or relative prices), it has a more convenient and direct causative interpretation. Second, Jordà and Taylor (2009) focus on this variable as being critical in predicting carry trade profits over the volatility index of Brunnermeier et al. (2009). The statistical design in this section provides a more convenient basis for comparing the relative merits of these two variables in our data-set and hence elucidates on the claims made by these papers.
The results in figure 1 are revealing. It is well understood that exchange rates are persistent but here it is clear that naïve carry trade investment strategies have successful runs that last many periods. This is presented in the top-left panel of figure 1, where even a year after impact, the currency continues to depreciate in a substantial manner. This likely explains the first part of the euphemism “going up the stairs and coming down the elevator” so often associated with the carry trade. The second component of the carry trade is the interest rate differential between the investment and funding currencies displayed in the top-right panel of figure 1. Although a disturbance to this differential is difficult to pinpoint in a statistical sense, it is clear that its effects are quantitatively substantial: on impact a 1% depreciation in the currency has a 0.26% effect on carry trade profits whereas a 1% impact on the interest rate differential has a 0.58% effect instead.
The bottom-left panel contains the response of the carry trade that is most revealing: that to disturbances in the fundamental equilibrium exchange rate. This response has the correct sign and is accurately estimated suggesting a return to long-run equilibrium with a half-life of one year and almost 80% after six quarters. As Chong, Jordà, and Taylor (2009) argue, the manner in which the response is estimated, isolates short-run frictions from the true speed-of-adjustment to PPP and captures more accurately responses that are often estimated to have far longer half-lives in the literature. These results are also consistent with those in Jordà and Taylor (2009). Instead, the bottom- right panel in figure 1 shows that for conventional values of volatility, the effect on carry trade profits is rather small and imprecisely estimated. Moreover, economically, the importance of this variable seems low. Even a 40 unit jump would be associated with less than a 1% drop in carry trade profits. No doubt that the quarterly frequency of the data represents too much time aggregation to permit measuring meaningful