In application of ideas from modern portfolio theory (MPT) to commercial property markets, two broad strands can be discerned. The first concerns the appropriate weighting for real estate in the mixed asset portfolio. Here, debate has extended well beyond the "Case for Property" arguments which na?vely applied MPT to generate unrealistically large allocations to property.

More considered treatment of property performance data, introduction of a broader range of assets, setting constraints on class weightings and use of factor models to identify unique property models has increased understanding of the asset class in aggregate. A second theme has concerned the correct strategy for structuring portfolio allocations within the property portfolio.

Here, the importance of sectoral and geographical dimensions have been explored as a means of structuring investment and reducing the impact of large lot size and indivisibility. Within the UK, such studies have tended to suggest that sectoral factors are of prime importance and that any geographical classification needs to reflect different economic drivers of performance. 

 These two scales of analysis - aggregate asset class research and consideration of differential performance within property - have tended to be kept separate. While some authors have noted that use of a Markowitz optimiser approach to asset allocation may result in sub-optimal allocations when linked to a staged top-down allocation process, this has, to the best of our knowledge, not been tested empirically in relation to property markets.

Let us suppose that allocations are carried out in a staged process beginning with the aggregate asset classes: say domestic and overseas equities, bonds, real estate and cash. At this level, target allocations to the asset classes are set based on the averaged risk, return and covariance characteristics of each class. At the next stage, allocation is made within each asset class - to sectors in the share market, to some regional/sectoral division within property again based on averaged performance at sub-class level.

The allocation proceeds in this manner up to stock selection. This may mean that individual stocks and individual property types which are driven by the same underlying factors receive allocations that ignore their individual covariances. An obvious example might be City of London offices and the stock market performance of major financial services firms. 

 Such interactions will be analysed using property performance data at sector-region level, stock market data at sector/industry level and bond/gilt data. Real estate information is available from the IPD monthly index, while financial performance data have been extracted from Datastream. A first stage will be to examine the time series qualities of the data, remove the impact of serial correlation where appropriate and produce both nominal and real series. Thereafter, a series of statistical analyses will be conducted to shed light on the research question. This will probably include: 

   (a) application of optimisation methods to simultaneous and top-down allocation strategies to investigate differences in the final allocations - optimal portfolios being determined by, for example, their Sharpe ratio or similar; 

   (b) use of exploratory data analysis methods to identify performance groupings amongst the asset sub-classes (that is: does property cluster together or do mixed property and stock groupings emerge);

   (c) use of multi-variate procedures to explore whether there are common "industry" factors across stocks and real estate or whether asset class factors dominate (a methodology used by Drummen and Zimmerman in examining European stock performance: here it amounts to a test of the existence of a distinct property factor).

 Analysis will be completed by the summer of 1999, allowing adequate time for preparation of the final report and presentation. While the analytic techniques are exploratory, it should be stressed that the research will be grounded in investment theory. The results will contribute to our understanding of the nature and drivers of commercial property market performance and to the nature of the portfolio allocation process.