Improving the risk return trade off from carry trades
March 10, 2016 | 3 min read
If we could somehow forecast to some extent how currencies would move in the future, we might be able to mitigate some of the currency risk for investors. In a recent paper, forthcoming in the Journal of International Money and Finance, Helen Lu and Ben Jacobsen show that this seems possible.
More and more evidence appears in academic journals that financial market returns are to some extent predictable. Many of these results are used in investment strategies of practitioners. One strand of the academic literature that has received relatively little attention from practitioners to date focuses on ‘cross asset return’ predictability, which means that changes in one financial market may forecast movements in another market.
The intuition is straightforward. Prices in financial markets aggregate information of traders and as a result are good indications of the future. A general rise in stock market prices indicates that investors expect that our economic future looks brighter.
Bank and oil stocks
Now, if investors cannot observe all the information at the same time and information gradually diffuses into prices this may lead to cross asset return predictability. An analyst of bank stocks may closely follow the interest rate but pay less attention to oil prices. An analyst of the oil industry may focus on oil prices but less on the interest rate. But both changes in the interest rates and oil prices may affect bank and oil stocks. Hence, it might be possible to forecast prices of bank stocks using changes in the oil price and prices of stocks in the oil industry using interest rates.
This is of course a stylized example but there are many academic studies that suggest that using prices changes in one market do forecast prices in other markets. The returns of small firms are predictable using past price movements of big firms. Returns of the US stock market predict other markets world wide and oil price changes forecast stock markets.
A natural question to ask is whether currencies may to some extent be predictable using information contained in prices of other markets. Roughly speaking there are two sorts of currencies. Some tend to be more risky and pay a high interest rate (like the Australian dollar) and others are considered safer bets and pay a lower interest rates (like the Japanese Yen). Many investors use this difference in interest rates to construct ‘carry trades’. They borrow money in the low interest rate currency and invest that money in high interest rate currencies hoping to benefit from the difference.
This strategy will work well if the exchange rates movements do not more than offset the interest rate differentials. But currency risk tends to be large in these trades. If we could somehow forecast to some extent how currencies would move in the future we might be able to mitigate some of the currency risk. In a recent paper, forthcoming in the Journal of International Money and Finance we show that this seems possible and also how this can be done.
We find that equity returns predict profits from shorting low-yielding currencies (the short leg profit). Monthly carry trade profits (after crossing bid–ask spreads) from the short leg tend to be lower if equity indices drop and higher if equity indices increase over the preceding three months. The equity effect we document appears to be faster than the commodity effect. The predictive effect from stocks on the short leg carry trade is most pronounced at a two month lag. By contrast, the delay averages three months for the commodity effect on profits from buying the high-yielding currencies (the long leg profit).
These delayed predictive effects are not only statistically significant but also economically significant. For example, a movement in monthly equity returns of one standard deviation (about 4.39%) predicts same-direction changes in short leg profits after two months by 0.17 of one standard deviation (about 0.48%). Similarly, a change in commodity returns of one standard deviation (about 2.78%) three months ago positively predicts a change of 0.24 of one standard deviation (about 0.83%) in long leg profits in the present month. Results from testing this forecasting ability in actual out of sample trading strategies suggests that investors can improve the risk return profile of their carry trades.
About the author
Ben Jacobsen is professor in the FinanceLAB at TIAS School for Business and Society. His research focuses on forecasting financial markets and behavioural finance.
Helen Lu, Ph.D., M.B.A., M.Econ, BE, Ben Jacobsen, Ph.D. (2016) 'Cross-asset return predictability: carry trades, stocks and commodities', Journal of International Money and Finance, Elsevier.