Adrian Hussey – Vice-President, Capital Markets, Manulife Financial

The well-known studies of Brinson, Hood and Beebower (1986), and Brinson, Singer and Beebower (1991) showed that asset mix accounts for more than 90 per cent of institutional returns.

When thinking about absolute return strategies (ARS), pension plans should focus on the potential contribution of asset allocation to portfolio efficiency.

The effort devoted to asset class decisions can be viewed as an allocation problem. Sponsors should devote more resources to asset allocation decisions that offer greater marginal increases in portfolio efficiency. We can learn which decisions provide the greatest efficiencies by building up the efficient frontier in steps. We can also gauge the relevance of ARS.

The base case is an efficient frontier constructed with broad indices of bonds and equities, SMU Bond Universe, and the TSE 300. We then examine the effect of treating as separate asset classes short-term, medium-term, and long-term bonds, and value, growth and small cap equities. We can also add U.S. equities, international equities (MSCI EAFE), and finally ARS. This process reveals several things: allocating bond classes separately provides a small increase in efficiency at the low-risk end of the frontier; allocating large cap growth and value stocks separately adds a small amount of efficiency at the high-risk end of the frontier; adding small cap stocks to large cap value and growth adds very little to efficiency; the addition of U.S. equities increases efficiency markedly; and, finally, international equities add little more to return but are selected as a diversifier in the non-domestic component.

There are reasons to expect ARS to make a meaningful contribution to portfolio efficiency. Assets that offer relatively high risk-adjusted returns or low correlations with other asset classes are intrinsically attractive. Correlations at any level less than one will contribute some diversification benefits but correlations of less than 0.5 are required before strong effects are observed. The historical data show that the risk-adjusted returns and correlation characteristics of ARS make them very attractive from a portfolio perspective.

The correlation characteristics are particularly important. Variation in return between ARS funds is higher than with traditional funds, but the correlations between ARS funds are much lower. Correlations between ARS styles are generally low as well. Consequently, multi-style, multi-manager ARS structures are able to make use of very strong diversification effects and result in highly efficient portfolios. Also, due to the low correlations of ARS fund-of-fund portfolios to traditional asset classes, they can add significant efficiency to a portfolio of traditional asset classes.

When ARS indicies are combined with broad bond and equity indices to construct an efficient frontier we see the following:

* Improved efficiency across the full efficient frontier including the conservative (left-hand) end where other alternatives added very little efficiency.

* Where ARS weights are unconstrained, portfolios are dominated by ARS assets.

* Where ARS weights are constrained (to 30 per cent and to ten per cent), significant efficiency improvements remain.

* The results are preserved when I restrict my ARS strategies to market-neutral strategies.

There are caveats. Survivorship bias is almost certainly present and is difficult to estimate; liquidity may be problematic; reporting is inconsistent and often difficult to verify; new managers are entering the business at a furious rate; some successful styles are limited in volume; and total ARS assets are dwarfed by total institutional assets. Also, markets are very good at removing inefficiencies.

Over a period of 10 years, ARS Managers as a group have compiled a record of high risk-adjusted returns and low correlations with traditional asset classes.

They represent a significant opportunity to improve portfolio efficiency. *