Compass’ investment process emphasizes the importance of asset allocation as a determinant of return. Compass research shows that diversification strategies are more effective over the longer term than are strategies such as market timing, country timing, and style timing for controlling risk and return. Diversification is the major weapon of risk management, and thus our emphasis on diversification of asset classes, country exposures, currencies, styles and sub-styles.

This emphasis on asset allocation raises an important question: What is an asset class? In our view, to qualify as an asset class with specific return, volatility and correlation assumptions, the return stream of an investment should result from an economic activity or process which is fundamentally different from the other investments in a multi-asset portfolio. Equities and fixed income securities satisfy this requirement by their very nature, and we make distinct allocations to these asset classes within our portfolio. Emerging market debt and emerging market equity are distinct asset classes because of the sovereign risk aspects of developing markets. Emerging market debt, for example, differs fundamentally from U.S. investment grade debt because the return component from sovereign risk (risk that a foreign government will default on its loan or fail to honor other business commitments because of a change in national policy) is largely independent of U.S. interest rates. On the other hand, U.S. small cap equities are not a distinct asset class from U.S. large cap equities because the economic activity which generates the return streams to a large company and a small company are fundamentally identical.

These are not unimportant distinctions for multi-asset portfolio construction. For example, as an asset class, we develop specific expected return, volatility and correlation assumptions for emerging market debt, and we include it in our portfolio within a mean-variance framework. As a result of its differentiated characteristics, the weighting of emerging markets debt in our portfolio is considerably different from its market capitalization. While the market capitalization of emerging market debt represents only 2.5% of the global bond market, we have a 10% weighting in our global bond portfolio. In contrast, we do not make allocations to U.S. small cap equities within a mean-variance framework, but we hold a market capitalization weighting which is approximately 20% of the U.S. equity market.

We are frequently asked why Compass does not have an allocation to real estate and venture capital in our multi-asset models. Are the return generating mechanisms of real estate and venture capital fundamentally different from that of other equities? Compass believes that they are not. As with equity investments in general, the returns to real estate, for example, are determined by cash flows, which are derived by the overall level of demand for goods and services, and the pricing of those cash flows, which is affected by the cost of capital and interest rates. While there are real estate industry-specific characteristics such as the prevalence of long-term leases which influence cash flows, these characteristics are no more unique to real estate than are industry-specific characteristics which exist in the energy industry, airline industry, semi-conductor industry or any other industry. Thus, real estate is no more an asset class than are oil companies.

One of the reasons why real estate has been frequently considered as a separate asset class is the perceived low correlation between indices of real estate such as the Russell/NCREIF Index and equity market indices. As a result, portfolio optimizations using mean-variance models frequently result in 5% to 20% allocations to real estate. However, the appraisal-based nature of these indices makes estimates of correlation extremely unreliable. The appraisal process results in unrealistic valuation changes which, by construction, lag the business cycle and equity market returns, and therefore, give the illusion of lower correlation to the equity market. To show this, if we value the stocks in the S&P 500 using the same appraisal-based methodology used to value the properties in the Russell/NCREIF Index, the "appraisal-based" S&P 500 has the same low correlation with the market priced S&P 500 as the Russell/NCREIF Index has to the S&P 500.

For this reason we believe that a better measure of the correlation of real estate to the equity market is from the returns of equity REITs. Over rolling four year periods from 1982 to 1997, the correlation of quarterly returns between the Wilshire REIT Index and the Russell Small Cap Value Index has ranged between 0.6 and 0.9. These correlations are not indicative of a separate asset class, but are comparable to the correlations observed between other industry returns and market indices. These data further confirm our view that real estate is a sector of the equity market, and is not a separate asset class.

If real estate is an industry sector, then investments in real estate securities will be a part of the overall equity portfolio. This is the approach which we have taken to real estate in our investment portfolio. Investment managers who specialize in small cap value stocks will generally provide exposure to real estate securities. Approximately 4% of the small cap value portion of our portfolio is currently invested in real estate securities. The allocation to real estate will vary based on our small cap managers’ views on the relative valuation of real estate compared to other investment opportunities within the small cap value sector of the market. The real estate sector represents approximately 13% of the small cap value benchmark, and the current allocation reflects a view that other areas of the small cap value sector provide more attractive opportunities.

The analysis of venture capital follows a similar reasoning. First, the return generating mechanism of venture capital is most similar to that of any smaller capitalization equity security. Again, however, the appraisal-based nature of the return to venture capital (venture capital investments are generally carried at book value until they are sold) has underestimated its correlation. This has led mean-variance models to dramatically overweight venture capital relative to its market capitalization. To provide perspective on the actual size of the venture capital industry, the annual investment by venture capital firms in the late 1980s was less than the annual spending by IBM for research and development! As a result, mean-variance allocations to this asset class are unrealistic. Like real estate, we believe that venture capital is a sector within the small cap market. To capture the higher expected returns of very small capitalization securities, one of the managers in the SEI Small Cap Growth Fund, Wall Street Associates, specializes in micro-cap securities, and currently, approximately 4% of the fund is invested in securities with market capitalization of less than $150 million, and 14% of the fund is invested in securities with market capitalization of less than $250 million.

Traditional direct real estate and venture capital investments are not, in our view, an investment in distinct asset classes, but represent an investment in the private equity market. Real estate securities and micro-cap securities, on the other hand, represent investments in the public equity market. Again, the return stream of a private equity investment results from the identical economic activity as a public equity investment. The primary difference between a private equity investment and a public equity investment is the illiquid nature of private equity. As a result, investors in private equity expect a return premium to compensate for this illiquidity.

Long-term investors may benefit from capturing the illiquidity premium of private equity. However, the large increase in funding for private equity from institutional investors has inevitably reduced returns to the point where the excess returns may no longer compensate for the lack of liquidity. Further, it is our observation that very few investors have an investment horizon that enables them to realize this return premium. Reflecting their lack of liquidity, private equity investments are often sold at distressed prices, further increasing investor dissatisfaction. Even when the return premium is realized, the impact on the overall investment portfolio tends to be low at typical levels of private equity commitment. For these reasons, we believe that public equity investments are generally preferred to private equity investments, particularly for real estate.

In the case of venture capital, however, it may appear that the process of going from public to private is a transaction-based return generating mechanism which is independent of public market returns. On the other hand, this transaction-based return may also be viewed as a mechanism for realizing the illiquidity premium. Further, this portion of the return to venture capital is not independent of activity in the small cap market. For example, if returns to small cap equity are low and there is a lack of liquidity in this market, then transaction-based returns to venture capital are likely to be low as well. We continue to research the return generating mechanism and characteristics of venture capital as a possible addition to our global portfolio.

In summary, asset classes are distinguished by the characteristics of their economic exposures. They are not defined simply by their historical statistical correlation with each other. To qualify as an asset class with specific expected return, volatility and correlation assumptions, the return stream of an investment should result from an economic activity or process which is fundamentally different from the other investments in a multi-asset portfolio. Diversification among highly differentiated asset classes can improve the risk and return characteristics of a portfolio.

June 5, 1998

Timothy F. Shanahan, CFP

President