This chapter briefly reviewed the foundations of portfolio selection. It then applied and extended the MPI model of Grauer & Hakansson [1986] to construct and rebalance portfolios composed of UK equities, long-term government bonds, a risk-free asset, and both constrained and unconstrained investment in real estate over the period 1977- 1994. Returns were assumed to be normally distributed and transaction costs were ignored. Portfolios were then generated, using ten years of past realised returns to forecast future returns. The effect of constraining real estate investment on portfolio performance was also examined, together with the effect of including Far East equities on the proportion of real estate held. Real estate returns were unsmoothed.

Employing Barkham & Geltner's [1994] methodology in unsmoothing real estate returns resulted in a reduction in the percentage of real estate held. This point is increasingly relevant the more risk-averse an investor.

A constraint on the proportion of real estate held in a portfolio was shown to reduce investment performance, when compared to an unconstrained strategy that included real estate. However, on a risk-adjusted basis it still provided a substantial improvement in performance when compared to investment in financial assets alone. The constraint also enhanced the robustness of the model, as the adverse effect of assuming no transaction costs was reduced.

The results also suggested that, if a high proportion of assets are held in real estate, treasury managers should diversify company risk by investing cash balances designated for long-term capital expenditure in a diversified equity portfolio, or index fund, rather than in bonds. Immunising long-term cash commitments e. g. the future purchase of capital goods, by making an investment in fixed interest securities, may expose a company to increased risk.

Expanding the investment horizon, to include Far East equities reduced the overall proportion of real estate held in portfolios. However, this effect was limited for more risk-averse investors, and when considered in relative terms provided limited reason for the current low level of life and pension fund investment in real estate. With the  model still advocating that a substantial proportion of total portfolio value be held in real estate.

This argument is further strengthened, when considered against the model's assumptions. A fixed constraint may be an acceptable assumption when considering the investment behaviour of an individual or company. However, it is too restrictive to include institutional investors without an inherent investment in real estate. Thus relaxing the assumed fixed constraint, by allowing investors to alter their allocation to real estate by a certain percentage of portfolio value each year, would further enhance returns and strengthen the case for real estate investment.

As stated, placing constraints on the level and liquidity of real estate investment improved the robustness of the model. However, the results, in line with current theory, still suggest that life and pension funds should hold a higher proportion of real estate than current practice would suggest.

The next chapter will therefore review the principal methods of performance evaluation. It will discuss the definitions of risk and return, and then explore the methods developed to relate them. This will be followed by a consideration of the main issues in performance evaluation.