Читать книгу Property Investment Appraisal - Andrew E. Baum - Страница 13

1.3.1 Accuracy, Bias, Smoothing, and Lagging of Valuations

Оглавление

Since valuations are estimates of what the exchange price or investment worth might be, it follows that such estimates are subject to a degree of uncertainty. The principle of uncertainty around valuations is now firmly embedded in valuation standards and both the IVSC and the RICS have addressed this with guidance on reporting uncertainty around the valuation and its causes (IVSC 2019; RICS 2014). These discussions around the reporting of uncertainty have been given added momentum by the GFC and COVID-19 globally and Brexit within the UK. For example the “RICS Material Valuation Uncertainty Leaders Forum” meets weekly during any “unusual” events that may cause valuations to be subject to abnormal uncertainty and both RICS and IVSC have produced additional advice in 2020 in response to the COVID-19 pandemic (IVSC, 2020; RICS 2020b). Nonetheless, there is an expectation that appraisers will produce a solution that lies within certain parameters. In some countries, these parameters have been discussed by courts during valuation negligence cases.

Meanwhile, the degree of variation from prices or from other valuations has been examined in a number of studies. These have taken place for the large property investment markets of the UK and US, as well as for parts of mainland Europe, Australasia, Asia, and Africa. Crosby (2000) reviews studies for the UK, US, and Australia conducted during the 1990s. Since then, Cannon and Cole (2011) analysed 7,214 sales of apartment, retail, office, and industrial properties in the US over the period 1984–2010 for which performance measurement appraisals had been previously undertaken. Cannon and Cole found an average difference of +3.9%, with prices higher than valuations. This hid larger positive differences for times when market prices had appreciated, while during a declining market in 2008 and 2009 average differences were negative, with valuations higher than prices. The findings suggest that market valuations could be biased estimates of sale prices, but that the direction of bias changes between rising and falling markets. This is consistent with the hypothesis that appraisals are lagged indicators of value.

Measurements of the average difference allow positive and negative differences to at least partly cancel out. The absolute average is, therefore, another common metric in such studies, capturing the typical difference between valuations and prices regardless of which was higher or lower in any given instance. In Cannon and Cole (2011), the average absolute difference across the 25 years studied was 12.5%.

Another source of evidence is a series of regular studies carried out by MSCI. The last 30 years of the MSCI UK database has been examined by these studies to make similar comparisons of prices and valuations. Some of the results from these exercises are illustrated in Figure 1.1.


Figure 1.1 UK valuation accuracy and bias (value weighted figures): 1983 to 2017.

Source: RICS/IPD (2005), RICS (2019), and Reid (2016).

Over the whole period, the average mean difference between valuations and subsequent sale prices has been around 5%, with prices higher than valuations. This suggests a bias to under-valuation. The average mean absolute difference has been around 13%. Both of these figures mask substantial changes during the period. The absolute difference fell during the 1980s and 1990s before stabilising at around 8–10%.

If appraisals keep pace with how prices are changing, there should be no pattern to how the mean difference changes over time. However, the same pattern emerges in the UK results as in the US results discussed above. In the boom years of the late 1980s, valuations appear to have lagged further and further behind prices, suggesting that they rose at a lower rate than that at which prices were rising. Then, in the recession of the early 1990s, they seem to have followed prices down at a slower rate than prices were falling. After five years of falling and then static prices, valuations caught up in 1995, only to be left behind as property markets started to rise again post-1995. The pattern repeated itself in the years 2001–2009, but the catching up process in the downturn was much faster in this case, with valuations higher than prices in 2008. Since then, a positive mean difference, with prices higher than valuations on average, has re-emerged.

This type of analysis has been extended by MSCI to 12 nations where a long time series of valuation-based performance figures exists: Australia, Canada, France, Germany, Italy, Japan, Netherlands, South Africa, Sweden, Switzerland, the UK, and the US. Tables 1.1 and 1.2 set out the mean absolute difference and the mean difference (or bias) for those countries over the period from 2000 to 2018. The results shown in Table 1.2 are particularly worthy of comment. Many markets recorded their largest positive mean difference (with prices higher than valuations) in 2006 or 2007, during which the peak of the last global property cycle was reached. The UK, the US, and Sweden then had valuations higher than prices in 2008, while Germany, Japan, the Netherlands, Canada, and Australia had reversed by the end of 2009, with valuations higher than prices.

Table 1.1 Weighted average absolute differences between valuations and prices 2000 to 2018, by country.

Source: MSCI (2019).

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
Australia 3.8 3.9 4.4 4.8 4.8 5.5 9.5 13.7 10.1 8.2 5.6 6.6 3.6 5.8 9.2 13.1 8.8 11.9 10.5
Canada 7.4 6.4 9.7 8.1 9.3 11.6 15.7 13.4 8.8 9.9 8.2 10.7 16.7 7.4 9.5 15.6 12.0 9.2 13.1
France 6.6 7.9 6.6 5.4 11.5 10.2 14.5 12.9 9.9 7.8 11.1 12.6 9.7 9.2 9.3 12.7 12.8 9.8 11.1
Germany 12.2 9.2 7.1 11.6 5.0 6.0 14.3 14.3 14.8 6.1 11.2 9.6 10.0 7.9 8.4 9.9 13.4 18.1 9.7
Italy 20.5 12.7 7.9 19.1 4.7 9.3 14.0 17.5 12.5 12.2 7.6 10.0 6.9 9.2 8.6 8.6 10.8 17.7
Japan 4.2 14.3 22.3 8.1 12.9 7.3 10.2 8.6 8.2 12.0 11.6 10.9 14.0 13.0 9.5 11.7
Netherlands 10.2 8.4 9.0 8.1 8.2 8.6 11.6 12.4 5.6 9.3 4.8 5.1 7.5 9.5 5.8 7.1 10.2 10.1 10.2
South Africa 10.9 9.2 9.2 8.1 6.9 11.4 9.2 21.7 9.5 9.4 6.9 10.6 10.1 9.3 9.2 3.7 5.2 3.6 4.8
Sweden 18.5 9.6 10.1 8.0 10.0 10.4 21.6 15.8 13.7 16.7 9.0 16.1 7.9 10.8 10.8 13.0 12.4 8.6 9.2
Switzerland 23.9 10.3 9.1 8.5 8.2 9.4 9.1 13.5 11.3 10.7 12.3 7.5 7.2 19.2 23.9 19.5 10.3
UK 7.8 7.1 7.6 7.5 7.9 7.8 8.9 9.9 9.4 11.2 9.7 9.7 8.7 10.3 10.9 10.1 8.6 9.0 8.5
USA 5.1 8.6 7.7 6.3 9.6 11.6 10.8 10.3 8.6 14.9 10.1 9.9 9.0 10.4 8.5 8.1 6.2 7.0 6.7
Other 10.7 16.3 8.1 9.2 14.4 15.5 17.6 13.6 10.4 12.6 10.5 9.0 8.3 10.1 9.5 10.9 13.6 12.6 12.4
Global 8.6 7.3 7.6 7.3 8.4 9.2 11.5 11.9 10.1 11.1 9.4 9.9 9.2 9.4 9.4 10.3 9.2 9.3 9.0

Table 1.2 Weighted average differences between valuations and prices 2000 to 2018, by country.

Source: MSCI (2019).

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
Australia 0.4 −0.1 0.0 1.8 1.3 0.4 6.0 11.8 3.5 −4.6 1.4 3.6 −0.3 3.2 7.4 12.2 4.2 8.0 5.9
Canada −1.3 0.7 2.0 2.5 6.5 5.5 15.4 10.4 5.8 −0.4 1.5 6.8 14.4 3.6 −0.4 11.8 7.7 6.1 8.1
France 2.5 −0.2 1.2 2.5 8.8 8.6 12.9 10.4 3.5 0.9 5.6 10.6 5.9 5.1 3.7 9.2 5.3 7.5 9.5
Germany 3.6 1.2 −3.3 −4.8 −1.1 0.8 0.0 11.5 9.7 −3.0 5.2 7.3 4.2 2.8 4.0 6.5 6.1 13.8 6.5
Italy −5.5 12.7 7.9 19.1 1.0 6.1 13.9 16.2 3.9 0.9 3.1 1.9 −4.4 0.1 2.4 0.3 7.1 1.0
Japan 4.2 11.8 21.0 7.1 11.5 4.8 −8.5 −3.6 −2.1 −3.4 2.3 9.1 10.1 4.9 8.0 9.5
Netherlands 7.6 2.2 6.2 2.6 4.7 5.2 4.6 10.5 2.8 −5.3 2.1 1.9 −0.2 −6.0 −1.4 −1.5 3.5 3.8 4.4
South Africa 0.5 −3.0 0.5 1.4 −0.4 6.7 0.4 18.1 1.4 6.6 0.8 5.4 6.4 7.7 −0.3 1.3 1.1 0.7 −0.6
Sweden −8.7 4.8 5.9 4.5 7.3 7.5 21.3 10.8 −7.7 12.9 0.9 15.0 4.7 5.9 1.2 7.3 4.0 2.5 1.1
Switzerland 19.8 6.2 4.4 0.7 5.9 6.6 7.3 13.0 9.7 9.0 9.9 5.8 5.3 15.4 20.6 18.5 8.2
UK 3.8 3.4 5.1 5.5 6.0 5.8 6.6 2.9 −2.9 3.8 5.2 6.2 2.7 7.3 8.6 6.2 3.0 6.2 4.5
USA −1.5 1.0 4.2 4.4 5.5 7.4 7.5 4.8 −4.3 −10.3 4.8 5.2 4.5 2.2 5.2 5.6 0.6 3.1 2.3
Other 3.8 6.2 5.8 6.2 11.2 12.5 13.3 9.4 5.6 −0.7 1.3 5.8 2.6 −0.1 0.4 6.8 11.1 5.8 10.7
Global 0.9 2.2 3.6 3.9 5.4 6.4 7.4 7.6 0.1 0.0 2.4 6.1 4.2 3.4 4.7 6.7 3.6 5.5 4.7

These analyses assume that sale prices are independent of valuations. However, Baum et al. (2000) found that valuations were not independent of prices and prior valuations played an important part in deciding which properties were bought and sold – and at what minimum price. On this basis, it would be expected that prices would exceed valuations generally, as funds are less likely to sell or buy at prices that do not meet the last valuation or the next prospective valuation. Some funds struggled to get trustee approval for selling at less than prior valuation and some buying funds checked formally that their portfolio valuer would at least confirm the purchase price at the next valuation before proceeding to purchase. Another example is that of German open-ended funds where the rules governing these funds prohibit sales of assets at amounts that are more than a few percentage points below the valuation. Prospective sale prices at less than valuation would not be completed, so the samples of sales would be biased towards cases where prices exceeded the prior or prospective market valuation. Therefore, a positive bias in any accuracy study should be expected.

The expectation that valuers will lag and smooth the peaks and troughs in prices has been discussed by Geltner et al. (2003) and in Geltner et al. (2007). These authors are not alone in suggesting that one of the reasons for valuations lagging market movements is anchoring by valuers on the information contained in past comparable transactions and valuations. Anchoring refers to a psychological tendency to rely on an initial known figure (such as a prior valuation) by more than would be justified from its relevance to the appraisal at hand. This is arguably quite rational behaviour given the nature of the valuation process and the need to minimise errors that arise from noisy contemporaneous market signals (Quan and Quigley 1991), as well as the possible requirement to justify the valuation in court. In less liquid markets (where liquidity is defined as the depth of transaction activity), the anchoring on past valuations is likely to be stronger.

However, the accuracy studies show that the gap between prices and valuations increases in booms, which is where transaction activity is at its greatest. This suggests that the speed or extent of value change has a major impact. New information may be available, but it might not be fed into valuations quickly enough to keep them abreast of rapidly rising prices. This raises questions for the organisation of valuation services. In some jurisdictions, valuers are separate from the marketplace and are housed within specialist firms. Yet, in many mature markets, valuations are undertaken by the same firms that carry out agency functions, thereby improving access to market intelligence or “soft” information that can feed into a valuation. This is in addition to past valuations and evidence from completed transactions.

Smoothing relates to the extent to which valuations reduce the volatility of actual prices by missing the cyclical peaks and troughs of price movements. The impact of smoothing has been studied using individual valuations and valuation-based property performance indices. This includes attempts to quantify its extent, as summarised in Geltner et al. (2003). More recent analysis of a transaction index based on sales from the MSCI UK dataset (Devaney and Martinez Diaz (2011)) indicates that the current level of smoothing might be less than that found in previous studies, with the index being 1.4 times more volatile than an appraisal-based counterpart. Yet perhaps the most interesting finding was that, unlike other studies, particularly for the US (e.g. Fisher et al. 2007), the turning points appeared to be the same in both appraisal and transaction-based indices, suggesting that valuers might have smoothed the peaks and troughs, but not lagged turning points in the most recent cycle.


Figure 1.2 Volatility of valuation and transaction-based series (SD % pq): Q1 2002 to Q3 2019.

Source: Constructed by the authors from MSCI Global Intel 2020.

MSCI has, for several years, compiled transaction-based indices for some national markets (MSCI 2013), and comparisons with valuation-based indices for the same markets provide some varied results. The last set of comparisons that are available for a wide range of markets covers the period 2002–2019 (MSCI 2013). The results are set out in Figure 1.2. Some countries over this period had volatile property markets, like Switzerland, Italy, and Sweden, with standard deviations of 7.1%, 5.2%, and 4.8% per quarter, respectively. Others, such as the Netherlands, France, and Germany, were much more stable with standard deviations of 2.6%, 3.2%, and 3.8% per quarter, respectively. The more interesting differences are between the valuation and transaction-based series for each country. In the UK, the standard deviation for changes in the transaction index is only a bit above that for the valuation index, at 4.6% and 3.2% per quarter, respectively. At the other extreme, the valuation-based indices in Switzerland, Italy, and Germany have very low standard deviations at 0.2%, 0.6%, and 0.8% despite, in the case of Switzerland and Italy, having the highest volatility of transaction-based returns.

These results raise issues and concerns. Despite the basis of market valuation across the globe within international, regional, and national valuation standards being broadly agreed, valuations in some European markets may not follow prices as closely as others.

Property Investment Appraisal

Подняться наверх