Attribution of Long–Short Funds

Paul Copplestone
Gartmore

In current practice, long–short attribution typically lies in the absolute return contribution space, with most analysis simply calculating how much the fund made from long and short investments.

Although the results of such analyses look fairly simple, several complications must be addressed — the main one being how to classify long and short positions. For example, what does one do with a position that doesn’t exist at the start of the day but is shorted in the morning and bought back to go long in the afternoon before being closed at the end of the day? If one assumes the performance system contains ‘long equities’ and  ‘short equities’ as distinct asset classes, the problem is how to apportion the trade so that it takes the position from short to long between these asset classes in an automated way. As we shall see, the proper apportionment is not easily decided; it may be necessary to settle on an arbitrary rule or, by default, to ascribe the impact of the transaction to a residual effect.

It is nonetheless possible to achieve realistic results. The attribution is relatively easy to calculate: It is simply a case of multiplying the weight of the asset class in the portfolio by the actual asset class return to categorize the contribution in terms of country, sector, industry, strategy, or stock. At Gartmore, these calculations are done on a daily basis.

Another hurdle is deciding what to do with the residual when compounding daily contributions across time and comparing them with the actual fund returns at the end of a reporting period. Is it preferable to smooth the residual away or report it as it stands?

Once these issues have been resolved, the resulting analysis provides reasonable insight into the sources of return. The main drawback, however, of absolute contribution analysis as just described is that it does not take into account the market in which the fund is investing.

To scrutinize where the returns come from, we have developed a relative attribution approach that compares the fund’s returns with those of the market in which it invests. Of course, defining the relevant market may be appreciably more complex for long–short funds than for long-only funds because the former typically have fewer restrictions as to which markets they are able to invest in. If compromises are needed to achieve a measurable market proxy, this must be borne in mind when interpreting the results.

The first step is to examine the firm’s investment process and describe the potential sources of performance. In the illustrative case of a fundamental stock-picking fund manager, we initially defined the portfolio’s return in excess of the benchmark return as coming from three different areas:

• the fund’s net exposure to the market;
• the stock selection of the long equities relative to the market; and
• the stock selection of short equities relative to the market.

In this framework, we developed an attribution methodology that breaks the portfolio’s return into six effects, the first one of which — the net exposure contribution — captures the impact of being in the market. As will be seen, in an active long–short strategy, the net exposure contribution may not necessarily be identified with the return due to systematic risk (beta). The remaining effects are those resulting from long stock selection, short stock selection, hedging, the cash allocation, and inevitably, a residual or unexplained factor.

Net exposure contribution is calculated by taking the fund’s net exposure to the market at the beginning of the day and multiplying it by the market return of the day. The results of this calculation are presumably a proxy for the fund’s beta; for example, a fund that is continually positioned net long will enjoy a strong positive net exposure contribution during a period when the market rises. But a fund manager who regularly transitions between net long and net short positions can legitimately claim that a positive number here is a form of alpha or value-added return due to successful market timing.

Long stock selection is calculated by multiplying the weight of the long equities by their return relative to the market. Just as in long-only funds, this calculation shows the skill of the manager in picking stocks that outperform the market.

Short equity selection is calculated by multiplying the weight of the shorted equities (a negative value) by their relative return. This measure indicates the skill of the fund manager in selecting overvalued stocks, because a short asset that underperforms the market will produce a positive effect.

Calculating the results relative to the market gives much more clarity than using the standard absolute attribution because it breaks out two sources of return — market direction and fund manager skill.

Hedging impact is calculated by taking the weight of the asset multiplied by the relative return if the fund manager uses futures contracts to hedge the market (on a long or short basis). The calculation measures whether the fund manager’s chosen futures positions are effective hedges to the market proxy.

We calculate the impact of cash by taking the cash weight multiplied by the cash return. Although seemingly straightforward, measuring the impact of cash may prove to be the most difficult part of the attribution if the fund is leveraged. The complexity of determining the true cash exposure of leveraged funds combined with the fact that you are performing daily calculations on this data means you may end up with distorted monthly returns that are laborious and time-consuming to correct.

The residual, of course, is the unattributed or unexplained remainder. Under the approach described here, the size of the residual is driven by one major factor: Using start of day weights may not reflect the fund’s true exposure to the market because it is entirely possible that a fund that is 10 percent long to the market can be 10 percent short within the first few minutes of trading, depending on the style of the manager and any overnight news. Presently, we have no solution to this issue. We do find substantial residuals in funds that trade their net exposure aggressively or are intensively day-traded funds. In effect, we know why the residual is there and can thus reasonably attribute the residual to those activities.

Each daily contribution is then compounded over time.

Figure 1 illustrates this methodology by presenting the monthly attribution results for a long–short equity fund over a period extending from November 2005 through January 2009. The fund has achieved compelling results from all the attribution factors except hedging, which calls for further analysis. Clearly, this is useful information for the firm and the fund manager.

Figure 1. Long Short Attribution

All opinions and estimates constitute our judgment as of 20 February 2009 and are subject to change without notice. This material is for information purposes only and is not intended as an offer or solicitation with respect to the purchase or sale of any security.