The nature and workings of the direct real estate investment market differ from those of the other main asset classes. Unlike other major categories of investment where there are well-developed markets in which homogeneous investments are regularly traded, real estate is a heterogeneous asset. No two properties are the same, whether in terms of physical size, accommodation, structure, condition, differences in tenure or simply location. There is a paucity of information within the direct real estate market, exacerbated by there being no central market place for its transactions. Even where prices are published e.g. auction results, the full details of a deal are rarely made known.

The supply of real estate, certainly in the short-term, is relatively fixed. Thus in periods of strong demand, returns from real estate will contain a high element of economic rent. Conversely, low demand should reduce the economic rent. However the normal UK institutional lease provides for upwards-only rent reviews, which often act as a floor to the income generated. As a consequence, returns from real estate tend to be more volatile upon an upturn in an economy than the performance of that economy as a whole. In addition, the cost of acquiring or disposing of real estate is relativelyhigh and hindered by the complex legal structure of real estate rights, both in the UK and overseas. Properties also require maintenance, suffer from depreciation, and are often physically indivisible. The increasing effect of technological advancements on their usage and thus value, is also an issue.

Some evidence shows that real estate offers a long-term hedge against inflation. How- ever, the changing structure of debt financing -to short-term fixed or variable rate - combined with forecasts of a future low inflation environment, may have weakened this ability. In addition, the liberalisation of constraints on investment overseas, coupled with the enormous growth of derivative securities, has provided alternatives with which institutions can diversify UK equities.

However, almost all studies that employ conventional investment theory advocate the inclusion of real estate within a multi-asset portfolio. Those that do not, often depend on an excessive degree of unsmoothing. Although as they almost all do, by usingreal estate indices they implicitly assume that investors hold a `diversified' real estate portfolio.

The last 20 years has seen smaller funds gain a larger share of the total pension market. These funds are less likely to invest directly in real estate because of the difficulties in creating a sufficiently diversified real estate portfolio. However, the strength of this point is weakened when considered within the context of a multi-asset portfolio. Investors do not need to hold a fully diversified real estate portfolio in order to obtain exposure to the benefits of diversification. The degree of such benefits will however be linked to the level of diversification.

The liability structure of life and pension funds defines their natural preference for long-term financial assets. In contrast, however, institutions have been placing less weight on the long-term stability of returns, and relatively more weight on short-term portfolio performance. As fund managers face shorter time horizons over which their performances is judged. They are increasingly forced to optimise their investment strategy in order to best meet these performance targets, which real estate investment may not assist in the attainment of. Although, within this framework the concept of pricing only systemic risk relies on a reasonably long investment horizon. Therefore, shortening investment horizons also affect the quality of expected returns and asset correlation estimates, and reduce the validity of assuming the multivariate normal dis- tribution of returns. The comparatively poor availability and quality of information on the real estate market - compounded by the reduction in institutional involvement which further reduces information - has increased the relative risk of real estate investment. Short-term investment horizons, however, should not be of concern when considering the structure of life insurance and pension fund liabilities.

Taking cognisance of the above discussion, direct real estate investment encompasses a collection of characteristics not found in competing asset classes. As outlined, many researchers have considered the benefits of real estate investment. Almost all concluded that the variety of risks attached to real estate, within the framework of conventional investment theory, are not necessarily a sufficient explanation of its low weighting. While the case for investment in real estate by small or mature funds may be weak, there seems little support for its current (low) weighting in medium to large funds.

However, within the context of the optimal long-term portfolio mix, two issues remain unconsidered. Firstly, asset pricing models are volatile in their allocation of assets from period to period. As discussed, real estate portfolios suffer from considerable inertia, and it is thus impractical to reallocate funds on a monthly or even quarterly basis. When revision does occur, reallocation takes time. Secondly, few of the studies on the optimal amount of real estate to be held in institutional portfolios have taken into account investor attitudes towards risk. It may be argued from the discussion in section 2.3 that institutions are risk-averse, weighing this `natural' tendency against the benefits of real estate.

Therefore, in an attempt to supplement those areas inadequately dealt with in the literature discussed above, the following chapter will briefly review optimum portfolio selection. This will lead to consideration of constraints on investment in real estate, differing investor attitudes towards risk, and their effects on portfolio composition and fund performance.