In a previous post, we reported on the evidence that the carry trade strategy – investing in currencies with high interest rates and borrowing in currencies with low interest rates – earns high returns with low risk. In this post we describe in more detail some of what is known about the risk and return tradeoff on a carry trade investment. Our focus is on the risk and the cost of risk.
The returns to the carry trade, like the returns on a number of other trading strategies, are sometimes characterized by the phrase “They take the stairs up, but the elevator down.” That is, the profits accumulate gradually, but once in a while there arises a very, very large loss. The figure below, taken from a recently published paper by Burnside, Eichenbaum, Kleschelski and Rebelo (here is the free working paper version), shows the distribution of returns to a carry trade strategy between 1987 and 2009. The shaded bars are the observed sample. The black line is a normal distribution with the same mean and standard deviation as the sample.
What you can see in the figure is that the sample has a fatter left tail than the normal distribution, as highlighted by the red circle. Those few bad returns are the events at issue. They are very bad, and although few, they occur much more often than is predicted by the normal distribution. And these few extreme events matter to the total return of the strategy: Harvard’s Jeffrey Frankel noted that “In one week of 1998 (October 4-10), the yen rose 16% against the dollar, thereby suddenly reversing years of profitable carry trade from the low-interest-rate yen into the higher-paying dollar.”
The carry trade, of course, depends upon the peculiar dynamics of exchange rates. And historically, exchange rates have sometimes exhibited sudden, dramatic changes. Often times this reflects a government policy under stress, where the authorities resist a devaluation until ultimately they are forced to succumb and do so in a sudden break. Take the example of the many Latin American countries in the 1980s and the 1990s when growth slowed down, public debt exploded and currencies sharply devalued.
Other times this reflects the sudden arrival of a financial crisis. Think back to 1992 and the UK’s exit from the European exchange rate mechanism or to 1997-1998 and the East Asian financial crisis, or, more recently to the impact of the 2008 financial crisis on that darling of the carry traders, the Icelandic Krone.
Do these few extreme events explain the usual profitability of the carry trade?
No, they don’t. At least that’s the answer if we stand by any one of the widely used risk models in finance that focus on the correlation between an investment’s return and the return on some benchmark stock portfolio. This includes the CAPM, the Fama-French and the Carhart factor models, among others. Using any of these models, and looking at the historical distribution of returns on the carry trade, the trade looks to be a good bet even accounting for these few bad returns.
So, where do we go from here?
One route being explored focuses on what economists have long called the “peso” problem and which is somewhat similar to the now popular black swan concept. The reference to the peso is to events in the 1970s when the Mexican government attempted to keep the exchange rate between the peso and the US dollar fixed as it had been since 1954. But at this time Mexican interest rates were higher than those in the United States, which meant that investors could borrow in US dollars and invest in Mexican pesos, profiting by the difference in interest rates. Noting that Mexico was then experiencing high inflation and a heavy national debt, Nobel laureate Milton Friedman is said to have pointed out that investors were concerned that the peso would be devalued, an event that had not yet been observed in the data, but which could certainly explain why markets were setting the two interest rates differently. In August 1976 the peso was allowed to float against the dollar and its value quickly fell 46%.
What if the historical profits on the carry trade reflect the existence of other “peso” events—bad outcomes that have not yet been observed? Just because the carry trade strategy has been profitable doesn’t mean that it will always be. We may just not yet have seen how bad the results can be.
That is the argument made in the work by Burnside et al. cited earlier in this post. In order to evidence the peso problem they craft an alternative carry trade strategy that is hedged against extreme outcomes. The difference between the two strategies should reflect the market’s assessment of the extreme events, including those not yet observed in the historical sample. These authors find strong evidence that the market places a high value on these extreme events, and that is the danger of these extreme events that justifies the historic profits earned on the carry trade in the past.
Investment strategies that consistently earn a small positive return, but that occasionally experience dramatic losses are sometimes described as picking up nickels in front of a steamroller. So, corporate treasurers, if you are going to pursue the carry trade strategy, picking up those high interest rate “pesos” lying on the road, be sure to watch out for that steamroller bearing down on the company.