Appraisal-Based Indices and Information

Quan & Quigley [1989; 1991] develop the following model of appraiser behaviour, at the individual property level (Quan & Quigley [1991, p. 138]):

They justify their approach by suggesting that

'...if the dispersion of valuation uncertainty were to increase as a result of wide fluctuations in transaction price arising from low volumes of market deals, then the value of K will fall, with the result that appraisers will place more emphasis on their previous valuation figure. It is our contention that there can be considerable variation in the size of the two information sources and, consequently, in their relative value.'

Working under the same rationale, by allowing the unobserved `true' value of real estate to be a function of information, real estate appraisals may offer an improved proxy of value in times of high market activity, as measured by the number of observable market transactions.10 To paraphrase, in times of high market activity a valuer places less import on a past valuation. The variance of observable transactions about the true underlying market clearing price will fall as the amount of comparable evidence that is available increases.

If this is the case, then the degree of 'smoothing' within the return series will be a function of market activity. Value indices of commercial real estate will be a better proxy of underlying (unobservable) value in times of strong market activity, but as market activity thins and there are a lack of comparables to estimate values upon, so the accuracy of the series falls. Value indices of commercial real estate provide a reasonable proxy of value within the bounds of a period of high market activity. However, fail to report the true volatility of real estate at other stages of the cycle, as discussed by Firstenberg, Ross & Zisler [1988].

This argument is indirectly supported by evidence from the securities markets. French [1980] and Gibbons & Hess [1981] found a significant difference in the mean returns of Mondays compared to other days of the week. These authors and Fama [1965] also found that the standard deviation of Monday returns is higher than those of the other days of the week. This latter result is supported by the intuition that more information relevant to price formation will accumulate over weekends than merely overnight for the remainder of the trading days in a week.

Information signals have been shown to lead to parameter shifts. Beaver [1968] and Patell & Wolfson [1981] demonstrate that seasonal announcements result in rate-of-return observations with higher variance during the disclosure period than during nonannouncement periods. Ball & Torous [1983] support this argument. They provide evidence consistent with a mixture of two normal distributions model for daily returns, resulting from a Bernoulli jump process, to describe the arrival of information.

Since this information signal of parameter shifts may be generalised (Kon [1984]), the principal argument of the above studies - that volatility is a function of information may be applied to real estate returns series.


10It could also be argued that the amount of noise inherent in the series increases as a market encounters a recession, and the 'quality' of transaction information - for example, upon tenant break clauses and/or rent free periods - decreases.