Growth in the number of currency managers and funds has been rapid over the past five years and is creating a headache for investment consultants and their clients. "Trustees and pension boards are under an obligation to display due diligence," notes Colin Robertson, senior consultant at Hewitt Associates, "but what if there is a scarcity of reliable performance data for them to base their judgement on?" Differentiating active managers by investment process is a key element in due diligence. With traditional asset classes manager differentiation by style is not a problem. By contrast, differentiating and selecting an active currency manager is more complex and problematic.

Proving the case for currency at a macro level is easy enough. The data most often quoted comes from the Bank of International Settlements (BIS). Daily turnover in the foreign exchange markets was $1.8trn in 2005, the last year surveyed, of which only an estimated $80bn was for profit, indicating that these markets are not efficient. The resulting inefficiencies are there to be exploited by active managers. But analysing the actual performance of active currency managers and subjecting them to a performance attribution analysis requires much more granular detail. An analysis by Barclays Trading Group (BTG) completed in May 2006 is widely quoted, also serving as a basis for further research by investment consultants advising pension trustees on manager selection.

This study monitored 108 alpha currency programmes supplied by 82 currency managers, and comprises the best available snapshot of the industry. A little caution is needed, however. The data may show survivorship bias, as only open funds are included. And one fact immediately glares out from this data: "No more than 17% of these programmes have been in existence for ten years or more," notes Amy Middleton, vice-president of portfolio & risk strategy at Bank of America. Only six funds from the universe have a track record lasting over 15 years or more. By contrast, no fewer than 40% of the funds in the universe were launched within the last three years, and no less than 75% launched in the last five years.

The BTG data makes no analysis of survivorship bias by manager style. Elsewhere, Parker Global Strategies, an active currency manager and currency index provider, has analysed the rate of closure in its fund universe by style; 55% of systematic funds close, and 65% of discretionary ones. However the distinction between the categories is not necessarily clear cut. "Care is needed over the definition of terms like systematic and discretionary," warns Michael Victoros, foreign exchange and tactical asset allocation specialist at ABN AMRO.

In fact, the subject of active currency management is complicated by a lexicon of more or less loosely defined terms that seek to describe investment style and implementation. Style refers to the process by which an actual trading decision is generated, whereas implementation refers to the execution of these decisions. There are a number of styles, such as "trend-following". Once a trend has been identified, implementation may be systematic, generated by algorithm, or on a discretionary basis, coming from an individual or trading team.

Style and implementation are different sides of the same process; a currency programme could be trend-following in style but systematic in implementation, or fundamental and discretionary.

The variations are wide and data representing style needs to be treated with caution, not least because it is based on voluntary self-disclosure. In the BTG study, 82% of the surviving funds described themselves as trend-following, 12% as fundamental and 6% as blends of trend-following and fundamental. But the distribution of assets is heavily weighted to trend-following funds. These run $14.9bn, while fundamental funds attract only $0.2bn and mixed funds a mere $0.07bn (source BTG/BoA).

The past couple of years have seen a huge increase in the amount of new money going into these funds. According to data from Bank of America, capital under management has shot up from $3bn in 1998 to $11bn in May 2004 and $14.8bn today. A nuanced interpretation of this data shows that trading styles vary over the lifetime of surviving funds.

More than 80% of funds running for 15 years or more were trend following, the remainder fundamental in style. For 10 to 15 years all were trend following, with fundamental funds only re-appearing among those opened three to five years ago. Of funds started between one and five years ago, over 70% are trend-following, 20% are fundamental and the balance are mixed. Quite suddenly, fundamental style disappears among funds started within the past year; trend-following funds account for 75% of new launches and the remainder are mixed . "There are many hedge funds also managing currency," adds Robertson, "not to mention a growing number of products in which currency is included, such as global tactical asset allocation strategies."

Currency for alpha is a pure form of hedging activity, although generally providers shy away from being labelled as "hedge funds". Yet all the problems familiar from hedge fund indices and performance measurement recur with active currency indices. There are four widely recognised indices: Stark, which is money weighted; CISDM, which is median weighted; and Barclays and Parker, which are equally weighted. As with hedge fund indices, these vary by the number of funds assessed and have similar survivorship bias problems.

From December 1989 to June 2006, annualised index returns from these indices were 6.38%, 7.98%
7.05% and 8.95% respectively, according to Middleton. But within that period were years judged good and bad for currency. In one of the good years, 2002, the respective returns for these indices were 6.91%, -0.07%, 6.10% and 7.68%. In one of the difficult years, 2005, their respective returns were -0.57%, 3.30%, -1.20% and 0.04%. "These are widely dispersed results," notes Nigel Jenkins, of Payden & Rygel, which actively manages currency as well as bonds and equities. "Choice of index is important and the extent to which they can be used as a benchmark for a manager is conditional on understanding how each one has been constructed."

 

Differentiating currency managers also depends on understanding each fund's process. As we have seen, trend following or technical analysis is far the most popular style, measured by the volume of money under management. It is based on the assumption that currency markets are in some sense mean reverting; they study past trends and extrapolate these to the future. There is strong evidence that trend following can generate alpha, with one major proviso. Currency trends always reach an inflection point and turn. "Trend followers make money during a trend, but tend to lose money when the trend ends because they are bad at predicting when currencies inflect," warns Thanos Papasavvas, currency manager at Investec Currency Management. Managers using more fundamental styles claim an ability to predict inflection points.

The fundamentalists believe that they can forecast successfully, although there are at least five different sub-classes of fundamental analysis, each emphasisng a particular forecasting tool, or combining them to produce an aggregate forecast. One of the most widely used tools employs the AFX Currency index which is designed to replicate the most popular trading rules among futures currency traders. Each day moving averages (of 32, 61 and 117 days) are calculated for each currency pair involved in the benchmark. These averages are compared to the current price of the currency pairs. If the currency price exceeds the moving average then a 24 hour long position is taken, and vice versa. The attraction is that this can be used as a proxy for trend-following, but can also predict market inflection points.

Differentials provide a second tool. This approach puts a primary emphasis on these interest rate differentials (IRD) to generate buy and sell decisions. The manager goes long on the high yield currency and short on the low yield currency.

A third trading rule uses IRD's but compares the value on two parts of the yield curve, the short rate for a currency and the long rate for the same currency, usually taken as the one-week and the 10 year deposit rates. This tool is used to short the currency with the steepest yield curve, on the assumption that this will be depreciated by inflation relative to the other member of the pair.

Another tool also uses IRD's but with real rather than nominal interest rates, thereby trying to capture short-term inflation. The IRD's are calculated as before but adjusted by the year on year change in the inflation rate. A fifth and widely used tool is based on purchasing power parity (PPP). This tool says that if a one month forward rate is above that of the 12 month lagged rate, then the manager should short the next month and vice versa.

Fundamental analysis extends far further into long and short term drivers of currency valuation. Long term drivers include trade flows and structural efficiencies, and short term drivers include monetary and fiscal policy, real economic growth, market sentiment and asset returns. In some emerging countries, political factors and insider information may help to predict interest rate policy.

Similarly, the price of commodities, particularly oil, can have a direct effect on the value of a currency. A cottage industry of boutique consultants make part of their living gathering this information and selling it to currency managers. Furthermore managers scrutinise each other's process and trading patterns, searching for a consensus view of the market.

Understanding how these styles and tools are combined is crucial for performance attribution. There are signs that particular styles are yielding diminishing returns; some funds are closing to new money. Perhaps the future lies with mixed funds or portfolios like that offered by Goldman Sachs. "Our target IR on our fundamental fund and technical funds are 0.75," says Andy Bound, head of fundamental currency at Goldman Sachs, "combined they create a portfolio with a target IR of 1.25".