The Accuracy of Appraisal-Based Returns

In the UK, there is no single national real estate index. Instead there are many different indices, each based on a comparatively small number of properties and portfolio mixes.

The absence of a widely accepted benchmark reduces investor confidence that any index genuinely reflects the behaviour of the market as a whole. 

'Investors cannot have the same degree of confidence in property performance measures as is the case in other markets, so that the perceived riskiness of property is (perversely) higher than the indices suggest, as a result of uncertainty.' 13

Although the Investment Property Databank ("IPD") indices have begun to assume this role, they are based on institutional-grade properties which may not be representative of the real estate market as a whole. In addition, it remains a relatively short time-series, being quarterly from 1980, and monthly only from 1986.

Valuation also depend inevitably upon the subjective judgement of a valuer at a point in time, rather than on actual observed market transactions.14 Valuations of real estate are carried out periodically and short-term fluctuations in value, due perhaps to a pending planning decision, are therefore 'smoothed'. This results in real estate indices tracing a spuriously stable path compared to other asset classes.

The majority of studies based on commercial real estate returns use appraisal-based returns.15 In general these studies agree that such data is less accurate than that on financial assets, since the latter are based on transactions16 or quotations while the former are, at best, based on a mixture of transactions and appraisals.17 By definition, appraisals depend upon the subjective judgement of an appraiser at a particular point in time; they are not market transactions. Also, as the estimation of value is based on periodic appraisals, there is at least some potential for the elimination of short-term fluctuations in value. Thus the volatility of real estate returns series is underestimated and, consequently, the risk-adjusted return is overestimated.18

Secondly, valuations included within an index are unlikely to be carried out at exactly the same time. Any index, depending on quality and frequency, is likely to include a number of valuations which were undertaken both before and after its reported date of completion. The resulting index is thus a moving average.

An additional problem is the possibility of returns being overestimated. Giliberto [1988] showed that even where returns are i.i.d.19 over time, the derivation of returns from valuation based indices may result in an overestimate of returns.

'Unsmoothing' real estate returns series has therefore been the subject of much research. Assuming that the underlying true returns are i. i. d. across time (consistent with weak-form informational efficiency), Blundell & Ward [1987] proposed a simple inverted AR(1) model to unsmooth appraisal-based returns. This work was extended by, inter alios, Geltner [1989; 1991], and Ross & Zisler [1991]. The assumption that real estate markets are informationally efficient was relaxed in Geltner [1993] and Barkham & Geltner [1995]. However, in order that these models may be invertible, so as to reveal the `true' returns series, most corrections for smoothing have focused on the second moment.

___________________________

13Baum [1989, p. 5].

14It is interesting to note that, in 1989, the annual IPD Index only recorded approximately 60 sales per month (Godson [1991]). In the US, over the period 1978-1992 an average of only 38 properties p.a. or less than 5% -were sold from the Russell-NCREIF Index (Miles, Guilkey, Webb & Hunter [1993]). If this sample is representative, then the low frequency of transactions, leave valuers little information from which to determine market value at specific points in time. 

15Examples, include Brueggerman, Chen & Thibodeau [1984], Ibbotson & Siegel [1984], Hartzell, Hekman & Miles [1987] and Liu, Hartzell, Grisson & Greig [1990].

16Miles, Hartzell, Guilkey & Shears [1991] analysed the prospects for a transaction-based index and considered one unlikely for the foreseeable future. 

17For a discussion of the traditional valuation process and valuation accuracy, see Sykes [1983]. Fraser [1985a], Hager & Lord [1985], Brown [1985b; 1985a; 1985c] and Baum & MacGregor [1992].

18For a discussion of the performance characteristics of real estate indices, see Fisher, Geltner &  Webb [1994]. Morrell [1991] reviewed real estate performance indices in the UK over the period 1980-1990.

19All abbreviations and acronyms are explained in the glossary of this thesis.