A pension fund's decision to in-crease its allocation to global in-vesting is generally motivated by the desire to move to a point on the efficient frontier beyond what may be offered in the indigenous capital markets.

Specifically, a pension fund is seeking to build a portfolio with ex-pected return and risk parameters which are preferable to those of the portfolio prior to the increase in the global allocation. The increase in the global allocation is not made in isolation but is part of a larger portfolio decision.

Implicit in an increase in the global allocation is additional currency risk and return in the portfolio. The analysis of the impact of the increase in the global allocation on the risk and return of a fund's portfolio requires certain assumptions as to currency risk and return and ultimately explicit decisions with regard to: setting benchmarks, the decision on how to manage the currency risk, and structuring an active overlay programme.

Increasingly, mean variance optimisation, long used for determining the composition of portfolios on the efficient frontier, has been adapted by leading pension funds and their consultants to help in the currency decision process. This allows the currency management decisions to be made in a portfolio context and not in isolation.

The financial literature is replete with papers which discuss how currency benchmarks should be set. It is not our intention here to debate the merits of various benchmarks but to propose an intellectually justifiable framework for how a pension fund can select a benchmark which fits its base currency, its specific portfolio and its appetite for risk.

This framework takes as a given the pension fund's base currency and the current or planned asset allocations. Holding the asset allocations constant but varying the returns on the non-base currency denominated as-sets among passively hedged, un-hedged or options coverage, one can determine the mix of portfolios on the efficient frontier, including the one with the highest Sharpe ratio.

The result, while time period dependent, is a range of benchmarks (or in some cases a single benchmark) which would have been optimal given the fund's base currency and the specifics of its portfolio. In practice, these analyses result in three broad conclusions, two of which are intuitively obvious and one of which is not.

First, the larger the allocation to foreign equities, the higher the benchmark hedge ratio. Currency returns and local equity returns tend to have low or even negative correlation. As a consequence, small allocations to foreign equities tend to have an insignificant impact on total portfolio volatility. It is only as the allocations grow that the currency contribution to volatility becomes meaningful in a total portfolio sense.

Second, the greater the diversity of currency in the underlying portfolio, the lower the benchmark hedge ratio. The portfolio effect of spreading the currency risk over a broad spectrum of currencies with low correlation to each other dampens volatility and tends to result in a lower benchmark hedge ratio than a global allocation concentrated in a limited number of currency exposures.

Third, the benchmark hedge ratio is also a function of the base currency. Pension funds in countries with chronically high interest rates should generally have higher benchmark hedge ratios than funds in countries with chronically low interest rates. This is a manifestation of the forward rate bias which is well documented in the academic literature. Stated simply, the forward rate is related to the current spot exchange rate by the interest differential between two countries. Forward rates, however, are a biased predictor of future spot rates in that they tend to overpredict future movements in the spot rate. As a consequence, consistent passive hedging by a fund in a relatively low interest rate country will over time result in costs which exceed the currency losses. Conversely, consistent passive hedging by a fund in a relatively high interest rate country (a good example is New Zealand) will, over time, generate earnings which exceed the currency gains from being unhedged.

Once the pension fund has established a benchmark hedge ratio which represents the optimal passive currency management approach, it faces the decision of how it will actually manage its currency exposure. The fund has three choices to consider:

The first is to hedge passively the currency exposure at the level of the benchmark hedge ratio. This ap-proach is analogous to indexing. Funds taking this approach often use an independent third party manager.

The second is to allow the equity or fixed income managers to manage the currency risk in conjunction with their active management of the un-derlying assets. Having established a benchmark hedge ratio, each manager's performance would be compared to the appropriate benchmark.

The third approach is to manage actively the currency risk using an independent overlay manager. Currency returns do not follow a random walk. Various academic studies have demonstrated that they are serially correlated (they trend), that daily returns are quite leptokurtotic (a fat tailed" distribution), that currency volatility is heteroskedastic (it clusters) and that there is the forward rate bias described above. These inefficiencies allow skillful overlay managers to add risk adjusted alpha by actively managing the foreign ex-change risk in the portfolio around the benchmark hedge ratio. The same mean variance optimisation which proved useful in evaluating the passive hedging strategies can be used again by adding in a manager's historical active management results, to determine how much of the currency risk should be managed actively.

Finally, it is important to note that many funds use parts of each alternative. For example, a fund may allow its global bond managers to manage the currency risk inherent in their underlying investments and measure those managers against a fully hedged benchmark. On the other hand, the same fund may measure its equity managers against an unhedged benchmark. Furthermore, the fund might combine a 50% passive hedging overlay for some part of the currency exposure on its equity portfolio with an active overlay programme measured against a 50% benchmark on the rest.

Structuring overlay programmes offers two other ways in which currency management and global asset allocation strategies are converging analytically: multiple managers programmes and independently optimising currency decisions.

Pension funds often combine non-correlating management styles in traditional asset classes to reduce the volatility of returns, particularly in asset classes where they believe that active management can generate risk adjusted alpha. Pension funds have begun to take the same approach in currency overlay management. Careful analysis of manager styles and historical returns can result in a 'portfolio' of two or three overlay managers with better risk adjusted alpha than any of the managers individually.

Similarly, a number of pension plans recently have begun to adopt a more flexible - non traditional or enhanced - approach to currency overlay programmes. An appropriate country mix for local equity returns does not necessarily provide optimal currency returns in an international portfolio. For this reason, the selection of currencies used in the enhanced currency overlay portfolio is independent of -although similar in makeup to - the currency mix of the underlying equity portfolio. The goal of this selection is to produce a portfolio which has the best chance of producing consistent returns from currency management over the long term. The diversity of market environments in which this style of currency overlay can add value enhances the performance of the currency overlay programme in absolute terms and on a risk adjusted basis.

Roderick Porter is president of FX Concepts in the US, Pike Talbert is director of marketing and Scarlett Mendoza is assistant vice president."