Hedging currencies with hindsight and regret by Meir Statman Glenn Klimek Professor of Finance Santa Clara University Leavey School of Business Santa Clara, CA 95053 firstname.lastname@example.org August 2004 I thank Eric Busay, Sridhar Chilukuri, Jennifer Chou, Ramie Fernandez, Kenneth Fisher, Bruno Solnik, David Tien, Rosalie Wolf and, especially, Roger Clarke and Mark Kritzman. I acknowledge financial support from The Dean Witter Foundation. Hedging currencies with hindsight and regret Abstract We examine hedged and unhedged portfolios during 1988-2003 and find that their realized returns and risk were virtually identical. Portfolio managers who care about the risk and expected returns of policy portfolios could have chosen to hedge or not to hedge by the toss of a coin. The mean monthly returns of unhedged global portfolios were higher than those of hedged ones in eight of the 16 years from 1988 through 2003 and lower in the other eight. The standard deviations of monthly returns of unhedged global portfolios were higher than those of hedged ones in nine of the 16 years and lower in seven. Money managers with reliable insights into the future movements of currencies should employ these insights to earn positive alphas, but the set of skills necessary for the construction of a good policy portfolio is different from the set of skills necessary for a good search for alphas. Portfolio managers who are pressed to earn positive alphas while constructing good policy portfolios are likely to do neither right. Such portfolio managers often forecast future movements of currencies by extrapolating past movements and hindsight and regret propel them from hedged portfolios to unhedged ones and back again, forever looking for positive alphas and finding negative ones. Hedging currencies with hindsight and regret. “Currency hedges may prevent losses on foreign stock portfolios if the dollar is rising,” wrote Pulliam (1992). “But these hedges can also generate big losses if the value of the dollar falls…” Well, of course. So what were the managers of the Alaska Permanent Fund thinking? The investment managers of the Alaska Permanent Fund, the state’s oil fund, wanted to play it safe two years ago when they begun investing in foreign stocks. So they decided to buy a currency hedge to protect their profits on the portfolio from losses if the value of the dollar rose and they sold the foreign stocks. But instead of protecting the fund from losses, the hedge has caused $38 million in losses of its own. Disillusioned with the strategy, the fund’s managers decided last week to abandon currency hedging altogether. ‘I can’t find any benefit to it’ said William Scott, executive director of the $14 billion Alaska fund, which each year distributes a portion of the states’ oil revenue to Alaska’s 550,000 residents in the form of a dividend. ‘I just want to stop the bleeding,’ he said. Currency hedges are advocated for the reduction of risk but they are often used for the enhancement of returns. The experience of the Alaska Permanent Fund is common and it illustrates the dangers of commingling the two. Arnott (2002) noted that the search for the right asset allocation is different from the search for the highest alpha. Asset allocation, including the allocation to currencies, determines the risk and expected returns of portfolios while well timed switches between currencies bring high alphas. Portfolio managers, such as the Alaska Permanent Fund, can study the liabilities of their funds and construct policy portfolios that match these liabilities with the right asset allocation. But the search for alpha need not be related to asset allocation. Alphas can be extracted wherever they can be found, in currencies, commodities, stocks or bonds, and these alphas can be ‘ported’ into portfolios to enhance their returns. 1 “Attempting to turn the treasury function into a profit center” is one way to lose money in derivatives, wrote Figlewski (1994, p. 77). He noted that such attempts press treasury managers into speculation that brings spectacular losses. Figlewski recounted the losses to Procter & Gamble from bets on the difference between US and German interest rates and the losses to Kashima Oil and Showa Oil from bets on the movement of the Japanese yen relative to the U.S. dollar. Attempting to turn policy portfolios into alpha centers is another way to lose money. We examine hedged and unhedged portfolios during 1988-2003 and find that their realized returns and risk were virtually identical. Portfolio managers who care about the risk and expected returns of policy portfolios could have chosen to hedge or not to hedge by the toss of a coin. The mean monthly returns of unhedged global portfolios were higher than those of hedged ones in eight of the 16 years from 1988 through 2003 and lower in the other eight. The standard deviations of monthly returns of unhedged global portfolios were higher than those of hedged ones in nine of the 16 years and lower in seven. Money managers with reliable insights into the future movements of currencies should employ these insights to earn positive alphas, but the set of skills necessary for the construction of a good policy portfolio is different from the set of skills necessary for a good search for alphas. Portfolio managers who are pressed to earn positive alphas while constructing good policy portfolios are likely to do neither right. Such portfolio managers often forecast future movements of currencies by extrapolating past movements and hindsight and regret propel them from hedged portfolios to unhedged ones and back again, forever looking for positive alphas and finding negative ones. The risk of currencies 2 The down and up of the Euro is one of many demonstrations of the volatility of currencies. The Euro was introduced in January 1999 at 1.18 to the U.S. dollar, decreased to 0.84 to the dollar by the end of May 2001, was back to 1.18 by the end of May 2003, and increased to 1.26 by the end of December 2003. So what proportion of the currency exposure should portfolio managers hedge? Answers to this question are rooted in mean-variance theory and range from zero to more than one. The volatility of currencies does not necessarily make them risky. Currencies can reduce the risk of portfolios if their correlations with other securities, both international and domestic, are negative. As Solnik (1998) wrote: “Foreign currencies provide an element of diversification against domestic budgetary, fiscal and monetary risks. For example, domestic inflationary pressures are usually bad for domestic interest rates and often lead to a depreciation of the currency. In this scenario, an inflationary rise in interest rates is bad for domestic bonds and stocks but good for foreign currencies. Although the value of foreign currencies is volatile, they bring some risk diversification to a domestic portfolio.” (p. 47) Not all agree that currencies reduce the risk of portfolios, and those who agree that they do often disagree on the optimal proportion of currencies in portfolios. Perold and Schulman (1988) argued for excluding currencies entirely from portfolios with full hedging. “When one buys foreign currency,” they wrote “the seller is buying U.S. dollars… there is little reason to assume that, in the long run, the rewards to bearing foreign currency risk will be one-sided.” (p. 47) Hedging, they wrote, reduces the risk of portfolios without reducing their expected returns. Froot (1993) argued for including currencies in portfolios with no hedging. Currencies go up and down, he noted, but they are not risky in the long run since they revert to their fundamentals. Campbell, Viceria and White (2002) argued for exposure to foreign currency 3 beyond the proportion in foreign stocks, to hedge the risk that domestic real interest rates might decline. Investors who hold domestic bonds but not international bonds should increase their currency exposure beyond the proportion of international stocks in the portfolio, such that the hedge ratio relative to international stocks exceeds one. Black (1990) offered a universal hedge ratio of 0.75 but Solnik (1998) prefers Gastineau’s (1995) hedge ratio of 0.50. “In a sense,” wrote Solnik, “Gastineau’s ‘why bother?’ approach is cleaner. He assumes that 0.50 is the best. Why 0.50? Because it is halfway between 0 and 1, neither of which is appropriate.” (p. 49). Global portfolios How do the data describe the risk and returns of hedged and unhedged portfolios? Consider U.S. investors with ‘balanced’ global portfolios composed of 60% in stocks and 40% in fixed income securities. The stock portion is divided equally between U.S. stocks and international stocks while the fixed income portion is divided equally between U.S. Treasury bills and Long Term U.S. Treasury bonds. The Russell 3000 Index represents U.S. stocks and the MSCI EAFE Index represents international stocks. MSCI provides currency-hedged and currency-unhedged returns of international stocks since 1988 and this is where we begin. Hedges come in two forms, perfect and probabilistic. Perfect hedges can be constructed when the correlation between two assets is a perfect 1 or a perfect –1. A perfect hedge can be constructed, for example, with a long position in the S&P 500 Index and a short position in a futures contract on the S&P 500 Index. The standard deviation of the returns of such a hedge is zero. Probabilistic hedges can be constructed when the correlation between two assets is high but not perfect. Probabilistic hedges reduce standard deviations, but not to zero. A hedged international stock position is composed of a long position in international stocks and a short position in the currency. The correlation between the two positions during 4 1988-2003 was -0.39. (See Table 1). The standard deviation of the returns of hedged international stocks was 15.92%, lower than the 17.06% standard deviation of the returns of unhedged international stocks, but the hedge was far from a perfect. (See Table 2). Mean-variance portfolio theory prescribes that risk be considered within the context of overall portfolios, beyond international stocks and currencies. Hedging currencies increased the correlation between the returns of U.S. stocks and international stocks, contributing to an increase in portfolio risk. The correlation between the returns of U.S. stocks and the returns of hedged international stocks was 0.71, higher than the 0.61 correlation between the returns of U.S. stocks and those of unhedged international stocks. But hedging currencies also contributed to a reduction of portfolio risk since the –0.04 correlation between the returns of Long Term U.S. Treasury bonds and the returns of hedged international stocks was lower than the 0.01 correlation between the returns of Long Term U.S. Treasury bonds and the returns of unhedged international stocks. We find that the annualized standard deviation of the returns of a global portfolio with unhedged currencies during 1988-2003 was 8.84%, slightly higher than the 8.73% annualized standard deviation of the returns of a global portfolio with fully hedged currencies, but the difference between the two is not statistically significant.1 The standard deviations of monthly returns of unhedged global portfolios were higher than those of hedged ones in nine of the 16 years and lower in seven. (See table 1) To check for statistical significance of the difference between the standard deviation of the hedged and unhedged global portfolios consider the annualized standard deviation of the monthly returns of unhedged and hedged global portfolios in each of the 16 years from 1988 through 2003. The mean difference between the two is 0.756%, consistent with the observation that the standard deviation of the unhedged global portfolios was higher than that of the hedged portfolio during the overall 1988-2003 period. But the standard deviation around that mean in the 16 years is 2.036%, and the t-statistic is 0.37, indicating that the difference between the standard deviations is not statistically significant. 1 5 Perold and Schulman studied global portfolios during 1978-1987 and found a larger difference between the standard deviations of unhedged global portfolios and hedged ones than we find for the 1988-2003 period. For example, Perold and Schulman found that an unhedged portfolio composed of 65% in stocks and 35% in bonds where 25% of stocks are international and 75% are domestic had an annualized standard deviation of 13.86% while the hedged one had a standard deviation of 12.48%. But they did not report whether the difference between the two standard deviations is statistically significant. The 8.53% mean annualized return of the unhedged global portfolio was slightly lower than the 8.60% mean annualized return of the hedged global portfolio during the overall 19882003 period, but as in the case of the standard deviations of global portfolios, the difference between the returns is far from statistically significant.2 Still, the two returns were very different in many years. For example, hedging boosted the returns the global portfolio in 1989. The annualized return of the hedged global portfolio in 1989 was 19.78%, higher than the 16.50% return of the unhedged global portfolio. But hedging detracted from returns 2003. The unhedged global portfolio gained an annualized return of 19.41%% that year while the hedged one gained only 14.76%.3 Portfolio managers who did not hedged were unnerved in 1989 while those who hedged were unnerved in 2003, but the number of years where the returns of hedged global portfolios exceeded those of unhedged global portfolios is equal to the number of years where the returns of unhedged global portfolios exceeded those of hedged global portfolios. Portfolio managers who care about the returns of components of portfolios, such as the international stock component, in isolation from the overall portfolio, were even more unnerved 2 We followed the procedure described in footnote 1. The mean annual difference between the returns of hedged and unhedged global portfolio was 0.050% and the standard deviation was 2.786%. The t-statistic is 0.02 indicating that the difference between the means is not statistically significant. 3 Annualized returns are calculated as twelve times mean monthly returns. 6 because differences between the returns of hedged and unhedges international stocks were huge in many years. For example, the 11.67% return of unhedged international stocks in 1989 was approximately half the 22.59% return of hedged international stocks and the 34.46% return of unhedged international stocks in 2003 was almost double the 18.99% return of hedged international stocks. It is little wonder that unnerved portfolio managers switch between hedged and unhedged portfolios in attempts to pick the winning one. The managers of the Alaska Permanent Fund know, with hindsight, that they were wise to drop their currency hedge in 1992. Unhedged international stocks did better than hedged ones in 1993 and 1994. But fates reversed in 1996 and 1997 when unhedged international stocks did worse than hedged ones. The cognitive bias of hindsight is followed by the emotion of regret. Some portfolio managers hedge 50% of the currency exposure of their portfolio to ward off the pain of regret, since a 50% hedge is sure to make them 50% right. A Mercer (2000) survey of the staff of large pension funds worldwide revealed that 34% of respondents with partially hedged benchmarks believe that currency exposure should be set at 50% to minimize regret.4 Conclusion We examine hedged and unhedged global portfolios during the 16 years from 1988 through 2003 and find that hedged global portfolios were as close to the mean-variance efficient frontier as unhedged ones. The two had virtually the same risk and return. Mean-variance portfolio theory teaches investors to search for portfolios on the efficient frontier with combinations of risk and expected returns that suit them best. But investors are A 50% hedged ratio also has risk-reduction benefits. Clarke and Kritzman (1996) offered a method for calculating hedge ratios that minimize the risk of overall portfolios. They found that minimum-risk hedge ratios are greater than zero in all but the unlikely cases where correlations between the returns of domestic assets and international currencies are negative and high. The minimum-risk hedge ratio for the 1988-2003 period was 67% but that figure was not known until the end of 2003. A 50% hedge ratio is a good rule of thumb for low risk. However, the difference in risk between hedged, unhedged and 50% hedged global portfolios is not statistically significant. 4 7 forever searching for sure winners. Securities that were sure to be winners in foresight often turn out to be losers in hindsight but the pain of regret does not stop investors for long. Investors are soon tempted to search again for sure winners. While the Alaska Permanent Fund hedged the currencies in its portfolio before 1992 and lost, Xerox did not hedge and won. “We seriously considered it a while ago,” said Robert Evans, an assistant treasurer at Xerox Corp. who oversees its pension investments. “But our treasury department thought the dollar would weaken,” he said. “Indeed, it has, and they saved us a bundle.” The decision to refrain from hedging brought a sigh of relief to Xerox in 1992 and perhaps some pride, but hindsight is always present and regret is never far behind. “In retrospect, one should have hedged all of the Asian currencies,” said Anthony Cragg, international equities manager at Strong Funds to Delaney (1997). “That’s because, besides watching stock prices plummet, U.S. investors have seen depreciating currencies further erode the value of their investments in Asia. The average fund investing in the region excluding Japan lost 33.7% this year, dragged lower in part by declines of more than 40% in a number of currencies.” Currencies present special forecasting temptations to investors since, as Solnik wrote, “Everyone, even a simple tourist, has an opinion on currencies,” (p. 49) and many are happy to magnify these temptations. “How you can play the falling dollar,” was the heading of Opdyke and Sesit (2002) article. Investors who have reliable insights into future movements of currencies should bet on them but words, such as hedge, should not confuse them. Those who hedge foreign currencies in global portfolios place bets and so do those who do not hedge. Fuerbriner (2003) wrote that “some portfolio managers have recently dropped their hedges as the dollar’s decline stretched into its 18th month.” We do not know if Fuerbringer’s mangers have reliable insights into future 8 movements of currencies and do not predict if they might the win the bets. But we do predict that many investors, driven by hindsight and regret, will continue to jump from bets on hedged portfolios to bets on unhedged ones, forever searching for sure winners. 9 Reference: Arnott, Robert (2002). “Risk budgeting and portable alpha,” Journal of Investing, v. 11, no. 2, Summer: 15-22. Black, Fischer (1990). "Equilibrium exchange rate hedging," Journal of Finance, v45(3), 899908. Campbell, John, Luis Viceira and Joshua White (2002). “Foreign currency for long-term investors,” NBER, working paper. Clarke, Roger and Mark Kritzman (1996). Currency Management: Concepts & Practices. Association of Investment Research Management. Delaney, Kevin (1997). “Funds generally don’t hedge Asian bets,” Wall Street Journal, December 22: C27. Figlewski, Stephen (1994). “How to Lose Money in Derivatives,” Journal of Derivatives, Winter: 75-82. Fisher, Kenneth and Meir Statman (1997). “Investment advice from mutual fund companies,” Journal of Portfolio Management, Fall: 9-25 Froot, Kenneth (1993). “Currency hedging over long horizons,” National Bureau of Economic Research, working paper 4355. Fuerbringer, Jonathan (2003). “Currency hedges: Thick with obscurity,” New York Times, July 6: BU12. Gastineau, Gary (1995). “The currency hedging decision: A search for synthesis in asset allocation,” Financial Analysts Journal, May/June: 8-17. Opdyke, Jeff and Michael Sesit (2002). “How you can play the falling dollar,” Wall Street Journal, June 11: D1. Perold, Andre and Evan Schulman (1988). "The free lunch in currency hedging: Implications for investment policy and peformance standards," Financial Analyst Journal, v44(3), 45-52. Pulliam, Susan (1992). “Pension fund managers find currency hedge is risky business,” Wall Street Journal, August 18: C1. Shefrin, Hersh, and Meir Statman (2000). “Behavioral Portfolio Theory.” Journal of Financial and Quantitative Analysis, vol. 35, no. 2 (June): 127–151. Solnik, Bruno (1998). “Global asset management,” Journal of Portfolio Management, (Summer): 43-51. 10 William M. Mercer Investment Consulting (2000). Results of Survey on Pension Plans on Currency Issues, September. 11 Table 1: Correlations between components of hedged and unhedged global portfolios, 1988-2003 Unhedged Int'l Stocks Hedged Int'l Stocks International Currency U.S. Stocks U.S. T-Bills U.S. T-Bonds Unhedged International Stocks 1.00 Fully hedged International Stocks 0.87 1.00 International Currency 0.39 -0.13 1.00 U.S. Stocks 0.61 0.71 -0.09 1.00 U.S. T-Bills U.S. T-Bonds -0.06 0.01 0.01 -0.04 -0.14 0.08 0.05 0.10 1.00 0.08 1.00 U.S. stocks are represented by the Russell 3000 Index. International stocks are represented by the EAFE Index. Table 2: A comparison of annualized returns and standard deviations of global portfolios, with various currency hedge ratios, 1988-2003. Returns of unhedged international stocks 26.54% 11.67% -21.99% 13.39% -11.58% 30.27% 8.55% 11.72% 6.42% 3.25% 20.47% 25.10% -14.09% -22.23% -15.21% 34.46% 0.51 6.11 4.92 17.06 Returns of unhedged global portfolio 16.23% 16.50% -4.86% 17.83% 1.68% 16.24% 1.99% 19.79% 8.97% 13.73% 16.88% 12.58% -1.04% -8.17% -7.47% 19.41% 0.71 8.53 2.55 8.84 Returns of hedged global portfolio 17.94% 19.78% -7.91% 15.84% 2.37% 14.79% -0.79% 19.77% 11.09% 17.55% 15.34% 14.74% 2.04% -6.14% -11.71% 14.76% 0.72 8.60 2.52 8.73 Standard deviation of unhedged Portfolio 23.60% 26.28% 47.82% 34.48% 20.08% 23.45% 26.40% 14.98% 20.41% 32.91% 38.34% 26.56% 32.40% 35.67% 34.03% 28.85% Standard deviation of hedged portfolios 24.52% 22.75% 46.16% 30.91% 17.40% 21.76% 27.07% 13.72% 23.08% 34.36% 38.53% 27.06% 29.00% 35.58% 35.85% 26.38% Return of minimum-risk Return of 50% hedged global hedged global portfolio portfolio 17.37% 17.09% 18.68% 18.14% -6.89% -6.39% 16.51% 16.84% 2.14% 2.02% 15.28% 15.52% 0.14% 0.60% 19.78% 19.78% 10.38% 10.03% 16.27% 15.64% 15.86% 16.11% 14.02% 13.66% 1.01% 0.50% -6.81% -7.15% -10.30% -9.59% 16.32% 17.09% 0.71 8.57 2.51 8.68 0.73 8.74 2.50 8.67 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Monthly Avg Return Annualized Avg Return Monthly Standard Deviation Annualized Standard Deviation Returns of hedged international stocks 32.23% 22.59% -32.16% 6.76% -9.25% 25.46% -0.71% 11.68% 13.50% 16.00% 15.32% 32.30% -3.83% -15.47% -29.34% 18.99% 0.53 6.32 4.60 15.92 Annualized mean returns are calculated as 12 times the mean monthly returns. Annualized standard deviation of returns are calculated as 121/2 times the standard deviation of monthly returns.