The author argues that the need for increased foreign investment exposure means increased exposure to potential currency losses, but also potential increases in total return of up to 190 basis points per year from more sophisticated management of currency risk. To Lindahl, that means pension funds and other large investment managers can no longer afford to leave their currency exposure unmanaged or passively hedged.
The percentage of pension assets invested overseas grew from near zero to over ten percent in the past fifteen years and many pension plans now have international allocations of up to thirty percent. This overseas migration was driven by studies that demonstrated adding foreign assets to a portfolio could boost returns and lower risk. These studies were usually based on an optimization of S&P500 index returns and EAFE index returns measured in dollars. Little attention was paid to the currency exposure and whether it would increase or reduce returns over time. International equity managers would hedge when necessary, but most of them did little or no hedging and the currency exposure was often left unmanaged as if it had been indexed. The years 1995-98 were not kind to this approach.
As plan sponsors realized that currency movements could have a large negative impact on returns they have turned to optimization programs to determine how much of the currency exposure they should hedge passively. Since it was generally assumed that currency had no expected long-term return, a passive hedge would, in theory, minimize the currency risk with little negative impact on returns. However, when currencies rose, hedges produced losses, and even though hedges were profitable when currencies fell, the cash flows associated with passive hedging were disruptive. These hedging programs merely moved the currency risk from the asset returns to the cash supporting the hedging.
Passive hedging should have been viable if currency had no long-term return and if currency returns were normally distributed. Alook at the German mark (DM) in 1969 to 1998 (the almost thirty years in which currencies have been floating) reveals that its monthly price changes seemed close to normally distributed (Chart 1). As a result, it could be expected that a passive hedging program would suffer large cash losses only infrequently. However, since pension funds measure results over calendar years, the use of annualized data led to misleading expectations.
CURRENCY RISK IS LARGE
When measured over calendar years, the DM rose an average of +3.5% per year in 1969-98 with a standard deviation of 12.8. The figures for the yen are +5.0% and 14.4. For comparison, the Dow Jones Industrial Average rose an average of 9.95% with a standard deviation of 15.8 over the same period - these currency changes were equivalent to a third to a half of the change in the Dow over 30 years! Prior to 1999 many of continental Europe's currencies tracked the DM (most of them have now effectively been replaced with the euro).
In the past fifteen years, for another example, the average unhedged annual currency return of the EAFE index was +3.5% (Chart 2), which would have made passive hedging of a pension fund's currency exposure expensive. Passively hedging the currency exposure 100% would have resulted in an average return of +0.1% and an opportunity loss, or cost, of 340 basis points per year (Chart 3). And, the cash flows supporting the passive hedging would have been large (Chart 4).
No one knows if the euro, the yen and other currencies will rise or fall relative to the dollar or each other in the decade(s) ahead. However, a decision to be unhedged or to passively hedge, even if the correct strategy for the long term were known in advance, would not prevent a pension fund from being impacted meaningfully by currency movements in the one to three years over which pension fund executives and money managers are commonly evaluated for performance.
A graph of the annual changes of the DM's value over the past 29 calendar years does not resemble the bell curve's normal distribution (Chart 5). In 23 of the 29 years (80%) the DM rose or fell by more than 6%. In only 6 of the 29 years (20%) it rose or fell less than 6%. Most of the returns were not in the middle of the "bell curve" but in its tails!
The risk of annual currency losses when exposures are unhedged and currencies go down is considerable, while the risk of large annual cash losses when fully hedged against currencies that go up is also large. If past is prologue, an investor can expect the euro to rise or fall by 10% or more in two out of three years!
A CURRENCY OVERLAY CAN BOOST RETURNS
Plan sponsors began to appoint active currency overlay managers in the late 1980s, but it was only last year that a comprehensive study by Brian Strange of Currency Performance Analytics revealed how successful currency overlay managers have been (see Pensions & Investments, June 15, 1998). The study included 11 overlay managers with over $40 billion in exposures managed in 152 accounts. Results were measured for the ten, five, three and two years that ended in 1997. For the full period the overlay managers produced an average annual added value of 1.9% against client-selected benchmarks. On average, 80% of the managers outperformed their benchmarks and they reduced risk (as measured by the quarterly volatility of the currency returns).
How do currency overlay managers compare with other managers? A study by Robert Zink, Director of Portfolio Strategy at Bridgewater Associates, which was based on performance data developed by William M. Mercer, provides the answer (see Global Pensions, June 1999). On average, currency overlay managers added more value than other managers, except non-US equity managers, for the ten years through 1997 and with less risk since the standard deviation of their added value was lower (Table 1). As a result, currency overlay managers had the highest information ratio (annualized added value/standard deviation). Simply put, currency specialists often knew when and how to best hedge currencies.
|TABLE 1 AVERAGE PERFORMANCE OF ACTIVE MANAGERS|
|10 Years Cap Growth through 1997||US Large Equities||US Large Cap Value Equities||US Fixed Income||Non-US Fixed Income||Non-US Equities||Currency Overlay|
|Annualized Value Added||-1.10%||0.60%||0.20%||1.50%||3.10%||1.88%|
|Source: Global Pensions, June 1999|
WHY OVERLAY IS ATTRACTIVE
The implication of the two studies is profound. With no or little increase in active management risk, pension funds with international equity allocations historically have boosted their international returns up to 1.9% per year by implementing an overlay while also reducing the risk of adverse currency movements. In one sense such a currency overlay is close to a "free lunch." Since investors with existing international assets already have currency risk, they don't need to make any additional investments to capture the extra return an overlay can provide. Since overlay fees are lower than equity and fixed income fees, the potential reward from each additional dollar of cost is high.
THE TEAM APPROACH
Most pension funds have guidelines that prevent equity managers from holding only a few stocks in a portfolio in an effort to "hit a home run." Diversification across many stocks, industries and countries is preferred because it reduces the risk of an equity manager underperforming badly relative to the benchmark.
However, little attention has been paid to the fact that the birth of the euro in January 1999 has concentrated the curren-blocks. In the Dow Jones World ex USA Index, for example, the currency risk resides in the euro and other minor European currencies (42.5%), sterling (19.9%), the yen (21.8%) and other currencies such as the Canadian, Australian, New Zealand, Hong Kong and Singapore dollars and emerging market currencies (15.8%) (Chart 6). In fact, 85% of the currency risk is concentrated in three currency blocks that each have a high correlation in their movements against the dollar. That strikes quite a serious blow at diversification efforts. In effect, as far as currency risk is concerned, the concentration is like having invested in only three companies. Naturally, the currency risk is even more concentrated in regional Europe-only and Japan-only mandates.
Overlay managers are normally not permitted to cross-hedge into other currencies, to prevent diversification of currency risk into other currencies. Although many overlay managers may have strong performance records overall, any such manager may fail to be hedged or unhedged at a critical moment. This means that appointing only one currency manager is less optimal than working with a team of two to three managers with different decision styles. The team approach increases the probability that an investor will realize results similar to those published in the study by Currency Performance Analytics.
Overlay managers fall into three basic decision styles although there is some overlap: model- driven managers, dynamic hedgers and discretionary or fundamentally driven managers. Currency Performance Analytics didn't measure which style is "best." That's a difficult task since currency managers work with different exposures, benchmarks and client-imposed constraints. However, Frank Del Veccio, who oversees General Motors Investment Management Corporation's $10 billion currency overlay program and who has experience mixing overlay styles, shared some insights at a currency risk management conference that the Association for Investment Management and Research (AIMR) held in November 1998.
Mr. Del Veccio said, "I did a study of about 10 currency overlay managers at GMIMCo, and every single manager added value for their composite returns. Correlations between managers tended to be pretty high, but it still makes sense to have a multiple manager structure. In particular, in my study, the correlation between the trend-following managers and the fundamental value managers was very high at times, but over a very long period, the correlation was close to zero. So, it makes sense to have a mix among styles. Another important reason to have multiple managers is to minimize your firm-specific risk." (See "Currency Overlay: Strategies and Implementation Issues" in Currency Risk in Investment Portfolios, AIMR Conference Proceedings, 1999.) We believe most pension fund executives in the US, Europe and Australia with overlay teams in place would agree with Del Veccio's comments.
WHY CURRENCY MANAGERS HAVE SUCCESS
The lack of comparable performance data on overlay managers and the widespread perception that currency management is synonymous with speculation have combined to discourage pension plans from implementing overlay programs. Since forecasting currencies has proved to be difficult, many argued, why should overlay managers be successful?
It's true that it seems impossible to forecast the price at which a currency will trade in the future. That has been demonstrated, for example, by The Wall Street Journal's annual polls of 50-60 economists in the past decade and their outlook for the Japanese yen. Each year their price forecasts have been approximately normally distributed. Each year roughly half of the economists thought the yen would rise and the others that it would fall. Even the average of their forecasts was usually off by a wide margin, even as to the yen's direction. You might think of these forecasts as akin to playing roulette betting on individual numbers, where the odds of winning are only 1 in 36 with each spin of the wheel.
Overlay managers do not try to forecast the actual future price of a currency. They focus on determining whether a currency is likely to fall or rise over the horizon over which they manage exposures actively. The probabilities of a fall or rise are constantly assessed to determine the best time at which to place or lift a hedge. The statistical profile of prices and their volatilities is the primary determinant that drives much or all of the decision processes that the current crop of overlay managers have developed.
Identifying the direction of a currency correctly can be likened to playing roulette by wagering that a number will be red or black. The odds of this strategy in roulette are 1 in 2, or 50%, with each spin of the wheel (ignoring the green zero). However, since overlay managers have developed techniques that allow them to be more right than wrong and since they cut losses on losing trades, they have improved on the odds in their favor and are able to produce profits over time.
Richard M. Levich, Professor of Finance and International Business, New York University, Stern School of Business, presented a summary of several studies of trend-following models at AIMR's currency conference. (See "Can Currency Movements Be Forecasted?" in Currency Risk in Investment Portfolios, AIMR Conference Proceedings, 1999.) Professor Levich found that if a currency manager attained a correctness of 60% in making directional calls, the probability of making money was high. He concluded, "A number of empirical studies... have demonstrated that particular models perform well at gauging the direction and sometimes the magnitude of currency movements over particular horizons."
Professor Levich's findings are supported by Brian Strange's study of actual currency overlay results and by the returns plan sponsors like GMIMCo have achieved over the years. Although currency overlay is a relatively new specialty, it seems a logical step in the asset management process.