© Copyright 2006 Rogers Publishing Ltd. The following article first appeared in the January 2005 edition of BENEFITS CANADA magazine.
In the search for alpha, pension plan sponsors need to give much thought to the construction of their portfolios.

Everyone is hunting for alpha. With return expectations generally much lower than historical returns over the last 20 years or so, the ability of pension plan sponsors to meet these expectations without taking some form of active risk has been greatly reduced. If passive asset-class returns are not enough to meet expectations, then either expectations must be reduced or returns to active management over and above the asset class benchmark(“alpha” for short)must be pursued.

The decline in nominal-return expectations has been matched almost one-for-one by a decline in real-return expectations, as inflation has been on a moderate uptrend in the last few years. Since real returns are what ultimately count, this makes the need to generate additional returns all the more necessary.

Arithmetic tells us that not everyone is going to succeed in finding enough alpha to see them through. Although alpha may be a zero-sum game and is an elusive quarry, it is possible to generate alpha. To do so consistently, a pension fund needs to be precise and thoughtful in how the overall portfolio is constructed.

One key realization is that some asset classes are far harder to add value to than others. Success in active management is proportional to the number of independent opportunities in an asset class and the number of people assessing the same opportunity set.

Lots of opportunities with few people looking at them simultaneously give the manager the best chance to add value. This means that adding value in the U.S. large cap space is generally going to be much harder than adding value in a small cap international portfolio where there are a lot more stocks and a lot fewer individuals analyzing all those stocks simultaneously.

However, most funds have a strategic policy that sets out the asset-class allocation. For Canadian plans, such a policy will likely include a hefty allocation to North American equities. This means that either the pursuit of alpha is restricted to other parts of the portfolio or, as many plans are now putting into practice, a portable alpha solution is used.

Alpha can be moved from one asset class to another in several different ways using derivatives. But the underlying principle is the same in each case: take an asset class and an alpha strategy based on that asset class, and through the use of derivatives change the asset class exposure while leaving the alpha portion unchanged. In this way, the links between the beta(or market return)part of the portfolio and the alpha part can be loosened, making it potentially easier to generate alpha.

This ties into another important aspect of the pursuit of alpha. If a manager has skill, then the more constraints imposed on the portfolio, the more of that skill will be lost as the portfolio is shaped to meet constraints. If the beta exposures are being constructed by other means, then there is less reason to impose constraints on manager positions and this will increase the opportunity set of the alpha-seeking manager.

Of course, this does not mean that plan sponsors should just hand over the money and let managers buy and sell whatever they wish. There is an art to finding the right blend of sensible constraints on how the assets may be invested and minimizing the impact on the ability of the manager to add value. Constraints may limit a strategy’s ability to make country bets or sector bets. Country-bet constraints may be sensible in an emerging markets strategy, but likely less so in a European- only portfolio where country effects are generally less important.

Allowing managers to take short positions in stocks is well worth consideration. In a paper by Clarke, de Silva and Thorley in the Financial Analysts Journal(September/ October 2002)titled “Portfolio Constraints and the Fundamental Law of Active Management,” the authors show that removing the long-only constraint on a conventional S&P 500 portfolio increases the potential valueadded markedly.

At a minimum, allowing managers to take short positions is worth discussion, since removing the long-only constraint on a portfolio increases the potential for higher returns, as can be confirmed by portfolio simulation and real-life experience.

This is to be expected. Most benchmarks—the S&P 500 included—are highly concentrated, meaning that a few assets make up a significant portion of the benchmark, and then there are multiple assets with much smaller benchmark weights.

For the S&P 500, the top 20 stocks account for 32.4% of market cap; the remaining stocks have an average weight of just 0.14% each. In Canada, in the S&P/TSX composite index, the top 10 stocks account for 34.5% of market cap and the remaining 213 stocks have an average weight of 0.31%. So a long-only manager investing in the S&P 500 can take a maximum short 14 basis-point bet against 480 of the names in the index, on average. That, and the equivalent constraint for the TSX, are fairly hefty constraints on the ability to add value.

A portfolio built along these lines would separate the alpha and beta decisions and concentrate the alpha exposure in those areas where the opportunities are relatively easier to exploit. This portfolio would also have fewer constraints on the alpha managers and would make much more limited use of constraints on managers than in the typical long-only portfolio. This is half of the task.

The other half concerns that enemy of returns: fees. If asset class returns are subdued, then the expense base of the fund needs to be kept under tight control. Some aspects of this are clear enough: don’t pay more than you have to for any part of the portfolio.

This is particularly relevant when thinking about the beta exposure. Active-management fees are much larger than passive-management fees. Thus, don’t pay an active manager for something you can get from a passive manager for a tenth of the cost. An active manager should be paid for generating alpha, not beta, which any self respecting large investment manager can provide economically. This can be done most easily by a set-up where passive managers provide the strategic exposure and the active managers are given long-short mandates where payment is performance-driven.

Another important fee reducer is to align the managers’ rewards as closely with those of the plan as possible. For example, a plan that hires 10 alpha providers all on a performance fee will end up fuming in some years; the managers that outperform get a performance fee, those that underperform get a flat fee. Thus, even if taking all of the managers together there is no net alpha for the fund, performance fees will still be paid out, raising the fee base of the fund.

If the same fund hired just one active manager and charged them with generating alpha from several different sources, then the performance fee would only be paid when value-added was generated at the aggregate level. Of course, having just one active manager may expose the fund to too much manager risk but the principle is clear: limit manager relationships and work with people who can offer diversified alpha pools as the bedrock of the alpha program.

Programs like this are attracting an increasing amount of interest as key components in the search for alpha. In contrast, sponsors should be careful of funds of funds as these managers are typically paying the performance fees to the individual managers and then adding a further fee layer on top. This is far from helpful in the search for alpha—unless the fund-of-funds manager is highly skilled.

Pursuing alpha means more risk. Conventional approaches to strategic asset allocation measure this risk by calculating tracking error relative to a strategic benchmark that itself is estimated by a separate process. This is not typically a good measure of the real risk to the fund, which is much better captured by using an integrative framework that allows risk relative to the liabilities to be calculated directly.

The conventional strategic benchmark-plus-tracking-error risk budgeting approach is often completely inaccurate as a measure of real risk to the fund. Using a better approach to risk measurement is not really needed if asset returns are going to be so high as to blunt all pain, but in the current environment, getting this trade-off right is crucial.

It’s important to note that even if a pension fund does all these things, there is no guarantee that alpha will be added. But if a fund follows none of these approaches and still manages to add a lot of value, then it must be very smart or very lucky. Happy hunting!

Anthony Foley is managing director with the Advanced Research Center of State Street Global Advisors in Boston. Anthony_foley@ssga.com