A study of micro‐level variation in appraisal‐based capitalisation rates

Abstract

This paper explores how appraisal‐based going‐in and exit capitalisation (cap) rates vary on the micro‐level, i.e. how they differ from property to property. The studied database consists of 3022 discounted cash flow market valuations of office properties in Stockholm, Gothenburg and Malmö during 1998–2004. The purpose of the paper is to test the ‘rationality’ of Swedish office property valuations. By rationality is meant the extent to which appraisals, in particular appraisal cap rates, follow from economic theory. This is an important issue since commercial property markets rely heavily on valuations. Cap rates are regressed on characteristics of the property, other valuer assumptions regarding the property (i.e. the property’s market rent) and variables that capture broad time series variation in cap rates. For the most part the studied appraisals follow the expected pattern. They do not exhibit major evidence of irrationality in the above mentioned sense though some of the findings point to the need for further research.

Keywords:

• appraiser behaviour
• capitalisation rate
• appraisal
• office
• micro‐level variation

Acknowledgements

The author would like to thank SFI/IPD for the provision of the data used in this paper and Lars Erik Lundberg Scholarship Foundation for funding the research. I would also like to thank the anonymous referees for their helpful comments and suggestions.

Notes

1. The actual variable used in subsequent regressions is defined slightly differently (see the data section). First, the variable actually used takes into account that market rent is measured as rent per square metre of rented space in the raw data used, whereas current rent is measured as rent per square metre of total space. Second, the variable used has the actual to market rent subtracted by one. This means that the variable can be interpreted as the percentage below or above market rent that the actual rent is.

2. This is the regression for Stockholm. Different location dummies are used in the Malmö and Gothenburg regressions.

3. The Wald test statistic is F _{2, 339} = 3.27 for Malmö and F _{2, 2111} = 8.10 for Stockholm (F _{2, 339, 0.05} = 3.02, F _{2, 2111, 0.001} = 6.93).