Re-discovering important, inconvenient and (therefore often seemingly) ignored research is always a pleasure. One of my favourites, that I keep on re-discovering is “An assessment of the impact of valuation error on property investment performance measurement” by Bowles, Tarbert and McAllister; earlier versions are available from 1997 but the published paper is from 2001*. The idea that is laid out in the paper is straightforward.

Valuations are not prices. Valuations are estimates of a likely sales price and, if valuations are unbiased estimates of a likely sales price (which they may not be, but that is another matter), there will be a range of estimated values around the unobserved sales price. This range is not the result of valuers making a mistake (an “error”) but of imprecision in the estimate. Indeed, studies (by the industry valuation body, RICS**) have suggested that for the UK, on average, only around 84% of valuations lie within plus or minus 20% of sales price (consistent with a standard deviation of around 15%). This always surprises people. This level of uncertainty is normal, not an error, and it may even underestimate the true level of uncertainty. This is because the only differences that can be measured are those where an asset is actually sold, most assets are not sold each year, and, property owners and their investment managers tend to use valuations as minimum sales prices, which they are not, or at least try to sell as close as possible to the valuation.

It follows then, that, if valuations are uncertain then property investment performance measurement, which uses such valuations, is also subject to uncertainty. Bowles, Tarbert and McAllister provide a useful table in the appendix of their paper which summarises the range of portfolio valuation uncertainty, based upon varying degrees of uncertainty in individual valuations for properties in that portfolio and the number of properties in the portfolio. For a portfolio of 20 properties and a level of individual valuation uncertainty of 15%, consistent with the RICS study, the portfolio-level valuation uncertainty is shown to be plus or minus 4.65% (with 95% confidence). So, a portfolio valued at £100m could reasonably sell for between £95.35m and £104.65m, simply on the basis of imprecise valuations. For a portfolio of 40 properties the portfolio valuation error falls to plus or minus 3.29%. It takes around 5,000 properties in the portfolio for the level of uncertainty to fall close to zero.

It would be easy enough to dismiss this work and just carry on regardless; that looks to be exactly what has happened. However, since performance reporting, funds’ unit pricing, funds’ performance rankings, managers’ performance fees, individuals’ compensation, inferences around manager skill, etc. are typically based around portfolio valuations it would seem more sensible to take this work seriously. Bowles, Tarbert and McAllister concluded and recommended that, “Property performance analysis … is likely to be constrained by uncertainty surrounding the inputs” and “corroborating evidence should be obtained to support apparent superior or inferior performance. Differences from benchmarks should be tested for significance. Similarly ranking of funds by performance may be problematic given that in many cases performance differentials may be statistically insignificant.”

In the context of the FCA’s recent consultation regarding open-end funds, this information seems important as there appears to be a lack of understanding about the inherent uncertainty in the individual property valuation process, the extent of reasonable uncertainty and the impact of that on portfolio valuations. It also begs the question of what “material uncertainty” in valuations is, given the extent of normal uncertainty, as evidenced by the RICS study which was first carried out on data from the last century and has been there for all to see.

* Bowles, Tarbert and McAllister, (2001), “An assessment of the impact of valuation error on property investment performance measurement”, Journal of Property Investment and Finance, Volume 19, Issue 2, pp 139-157

** RICS, (2019), “Valuation and sales price”