Abstract
This paper seeks to map out a practical way for managing risk in property portfolios. The approach adopted is to identify the factors that cause volatility in property returns. The factors are categorised in two groups – ‘fundamental’ causes of property return volatility and ‘modulators’ that dampen or exacerbate the variance emanating from the fundamentals. This approach enables a risk profile for individual property portfolios to be presented showing their relative exposures on the different risk dimensions. Analysis of the IPD UK universe of funds shows both that a number of the identified risk variables are correlated with higher tracking error and that the main types of UK investment fund have different risk profiles. Preliminary analysis is presented on the development of a model for predicting the tracking error for UK portfolios. The initial results show some promise but the model as specified here is not capturing a significant portion of the variance.
Acknowledgements
The authors wish to acknowledge the help they have received from many colleagues in formulating this paper but particularly the contributions of Gregory Golovkin of LaSalle, Lee Marshall at IPD and James Crutcher (formerly IPD now at ING REIM).
Notes
1. Assuming other things are reasonably equal. Portfolio concentration is measured throughout this analysis by calculating the proportion of portfolio value in the five largest assets. This is because large portfolios can be very ‘lumpy’ in nature and so portfolio returns are more sensitive to the performance of the large assets, rather than the number of assets per se.
2. The difference between Tables and is that the average risk scores over the period 1993–2002 are used in the former and the 2002 scores in the latter. Reassuringly, the significant relationships in table are confirmed when the risk scores for a single year are analysed.
3. Reflecting the intellectual inspiration behind the concept.
4. Three‐year periods were selected. Shorter periods would have contravened IPD’s commitment to data confidentiality. However, the possibility that longer periods are more relevant has not been discounted.