The scope for a manager to add (or subtract) value for a particular currency overlay mandate over a particular period will be heavily dependent on how the structure of the mandate under consideration interacts with exchange rate movements over that period.

For example, let's take a mandate with a totally unhedged benchmark that provides the manager with scope to hedge to the base currency, but not to short the base currency. Over periods when the base currency is strong, this mandate will provide the manager with plenty of scope to outperform, and very little scope to underperform. Any hedging to base currency over these periods will add value. Over periods when the base currency is weak, however, the same mandate will also provide the same manager with very little scope to outperform, and plenty of scope to underperform. Any hedging to base currency over these periods will subtract value.

Just recently it has been relatively easy for managers to add value, and difficult for them to underperform, for mandates with unhedged benchmarks and dollar or sterling base currencies. However the converse has been the case for mandates with unhedged benchmarks and Yen or A$ base currencies, so the base currency matters as well.

To translate this for mandates with fully hedged benchmarks that provide the manager with scope to go partially unhedged but not to go beyond fully hedged, we need to flip it around. The opportunity sets under fully hedged mandates have been quite favourable recently for yen or Australian dollar-based accounts, but not so favourable for dollar or sterling-based accounts.

By contrast with these two extremes, mandates with 50% hedged or other partially hedged benchmarks will often (but not always) provide a more balanced mix of opportunities to add value that will be less dependent on exchange rate movements over the period under review.

In general, it is very important to take the opportunity set provided to a manager under a particular mandate into account when evaluating the manager's performance under that mandate. Obviously it is also important to take this issue into account when you set the parameters of the mandate in the first place.

Investors seeking to evaluate potential candidates for a new currency overlay mandate will quite naturally seek to compare the track records of these candidates. However making such comparisons is not as simple as it is for equity and bond managers, for a number of reasons.

Firstly, any comparisons must be made in terms of performance relative to benchmark rather than in absolute terms. This point applies to equity and bond mandates too, but is sometimes ignored for equity and bond performance comparisons where the benchmark returns are often the same across all of the track records being compared.

Secondly, as mentioned above, the scope that a manager had to add (or subtract) value relative to benchmark over a particular period will have been heavily dependent on how the structure of the mandate interacted with exchange rate movements over that period.

In theory, the candidates could all be asked to provide track records for a particular type of mandate, so that comparisons can be made across managers on a like-for-like basis. In practice, however, there is so much variation between mandates that i t will often be impossible to identify a particular type of mandate that all candidates have a track record for. In any case, comparisons between managers based on the results for one mandate type only may be biased by the way that this mandate type constrained the scope to add or subtract value over the period under review.

Another option would be to ask each manager to provide historic data for all of their currency overlay accounts, including any that have been terminated. In practice, however, this may result in information overload, and make it difficult to see the woods for the trees.

In our view the most practical approach in most cases would be to ask each manager to provide track records for a number of accounts which between them cover a broad range of different opportunity sets, and to provide details of the benchmarks, base currencies and other mandate parameters for each of these mandates. Examining a broad range of track records for each manager should help you to gain a better understanding not just of how well each of the candidates have performed in the past, but also how the results have depended on the structure of the mandate. However you should be wary of selection bias - managers will often tend to put forward the selection of track records that shows their results in the best light.

The more aggressively the currency overlay programme is managed relative to its benchmark position, the greater the scope to add or subtract value. Hence the track records that look best in terms of value added relative to benchmark will often be those for the most aggressively managed mandates.

It makes sense to adjust the return relative to benchmark to allow for the level of aggressiveness, so that comparisons can be made in risk-adjusted terms. As almost all currency overlay managers are able to gear the level of aggression that they take up or down to suit client requirements, the desired level of aggressiveness should not be a factor that influences the currency overlay manager selection decision - it should instead be a factor that influences the decisions on the structure of the mandate.

A reasonable (but by no means perfect) approach is to compare track records in terms of information ratios rather than in terms of value added. The information ratio is the value added divided by the tracking error, which is annualised standard deviation of the value added. To put it more simply, the information ratio is a measure of the outperformance per unit of risk.

Some managers present composite data showing asset-weighted averages of the value added across all of their currency overlay accounts, sometimes subdivided between different broad categories of accounts. They do this in an effort to be as open and honest as possible, and to comply with the spirit as well as the letter of the regulatory requirements applicable within the US and some other countries.

Currency overlay composites can provide a useful broad-brush reasonable check on the extent to which selection bias may have crept into the track records quoted for individual accounts, and we would not want to discourage managers from preparing them. However we believe that composites are of only limited use unless all of the accounts included in a particular composite have identical or very similar benchmarks, base currencies and other mandate parameters.

Interpreting the impact of different opportunity sets and levels of aggressiveness in making comparisons between individual account track records is difficult enough. Interpreting the impact of these factors in comparisons between composites, which are each made up different mixtures of different types of mandates, is virtually impossible. Also, information ratios calculated from composite data can be very misleading in cases where the number of accounts included in the composite changed over time, giving rise to significant changes over time in the tracking error of the composite as a whole.

Many managers present figures on the volatility of the returns before and after the impact of currency overlay management, to show the impact that currency overlay management has on the risk level as the difference between these two measures. This is all well and good, and we would not want to discourage this practice either. However these presentations can sometimes be misleading, and there are some 'smoke and mirrors' tricks that you need to be aware of.

First, it is common to see presentations of track records for mandates with unhedged benchmarks, with scope to hedge back to base currency only, suggesting that the risk was reduced through active currency management. In many of these cases an even greater level of risk reduction could have been achieved by setting a partially hedged or option-hedged benchmark, and passively adhering to this benchmark position. The apparent risk reduction in these cases will often owe more to the structure of the mandate than to the manager's approach to active management.

Secondly, in some cases the risk reduction measures presented relate to the currency component of the returns only, and in some other cases they relate to the returns on the international component of the portfolio. Arguably, a third measure that is more important than either of these two measures is the impact that currency overlay management has on the risk level of the overall portfolio returns, including the domestic component of the returns. It is not all that uncommon for currency overlay management to reduce risk on the basis of one of the first two measures, but not on the basis of the third, and vice versa. Often the currency overlay manager won't even be able to track this third type of measure, because the client won't always provide this manager with information on its overall portfolio returns.

Thirdly, some currency overlay products feature controls over risk measures other than just the volatility of outperformance - eg a key risk control objective might be to avoid underperformance by more than 3% over any one year. For these types of products the risk analysis should cover more than just volatility. Clients making comparisons between products incorporating different risk control objectives should also give some consideration to which of these objectives are best suited to controlling the risks.

Is that all we need to look out for? Not quite. There are a number of additional issues relevant to the calculation of the returns themselves, but these get a bit too technical to be worth explaining here. Suffice to say that, in currency overlay, there are a few more issues to watch out for in analysing and interpreting performance data than there are for equity and bonds. However, given that the scope to add value through currency management is arguably at least as great as the scope to add value through active equity or bond management, we don't think that these issues should discourage plan sponsors from considering it.

Bill Muysken is head of manager research - global with Mercer Investment Consulting, and chairs Mercer's Global Manager Research Committee;

Michael Collins is a senior research consultant with Mercer, and is responsible for leading Mercer's research effort on UK-based currency overlay managers.