Neil Crosby, Colin Lizieri and Pat McAllister Means, Motive and Opportunity? Disentangling Client Influence in Performance Measurement Appraisals Neil Crosby, Colin Lizieri and Pat McAllister 19 May 2018

The Research Problem Are Clients Able To Systematically Bias Real Estate Appraisal Outcomes? “In the most unstable and unprecedented market circumstances for very many years, UK valuers have demonstrated their ability to respond speedily to exceptional changes in sentiment despite the thinness of the evidence available to them.”

Client Influence on Appraisals: Prior Research A Sensitive Topic – Raising Ethical, Reputational, Negligence and Criminal Issues Questionnaire Based Research Smolen and Hambleton, (1997), Worzala, Lenk and Kinnard (1998), Gallimore and Wolverton (2000) Interview Based Research Levy and Schuck (2005), Baum et al. (2000) Experimental Research Hansz (2000), Diaz and Hansz (2001)

Study Context Sharp Market Correction in Second Half of 2007. Decrease In Transaction Levels → So Limited Evidence Of Changes In Pricing Levels. Different Types of Incentives for Different Categories of Client? Open-ended funds required rapid marking to market due to NAV-based redemptions Closed-ended funds, insurance companies, listed funds more concerned with LTV ratios, bonuses, share prices?

Research Questions and Method Did Different Client Categories Exhibit Different Patterns of Changes To Capital Values in this Period? Did OEFs Experience the Largest Falls? Was This Consistent Across All Sectors? Were Appraisers Pushed By OEFs or Were They Pulled By Other Client Groups? Anecdotal Evidence – But Not Proof Opportunity to Test for Evidence in the Appraisals Themselves …

Data and Analysis IPD UK Data on Quarterly Capital Returns Disaggregated by PAS Segment and Type of Client. Confidentiality Issues: No Asset-Level Data Some PAS Segments Masked Implication for Statistical Testing Calculate Capital Growth Series for Client Types Examine Data at Sector Level To Control for Different PAS Weightings, Hypothetical Benchmark Indices for Client Categories Created Statistical Analysis Conducted at PAS Segment Level

Fund Type and Capital Change H2 2007: OEF values fell 222bp more than average

Why Could OEF Values Fall Faster? Information Effect? More frequent valuations overcomes anchoring effects? “Catch up” might provide some evidence of this Valuer Firm Effect? But same firms valuing OEFs and other fund types No evidence from IPD of systematic firm level biases Compositional Effect? Is it just about portfolio mix? Some sectors fall faster? Asset Quality Effect? Do the different client categories own different qualities of asset? Client Influence Effect? Can clients push/influence appraisers to mark prices down, or hold them up?

Timing: Falls and “Catch-Up”? Fund Type Fall H2 2007 Fall H1 2008 Fall from Peak Closed Ended Funds -9.03% -10.50% -23.07% Open Ended Funds -13.25% -7.95% -25.43% Pension Funds -9.95% -8.70% -22.50% Insurance Companies -10.66% -8.82% -23.54% Property Companies -10.31% -7.40% -21.68% All Funds -11.03% -8.86% -23.84% All: 100(.8997)(.9114) = 82.00 OEF: 100(.8675)(.9205) = 79.85

Segment Variations

Segment Variations

Different Sector Allocations?

Different Sector Allocations?

Preliminary Statistical Analysis Limitations Caused By Nature of Data Can Analyse By PAS Segment & Fund Type Observations = PAS*Type – e.g. City Offices held by OEF 45 observations in total Preliminary Non-Parametric Analysis Strong statistical association between fund type & performance – 2 significant at 0.001 level. Basic Regression Analyses Analyse falls in value by time period, e.g. H2 2007 RHS: PAS and fund type attributes Strong common movement across segments

Preliminary Statistical Analysis Dependent Variable: Capital Value Change, H2 2007 Variable Coeff T-Stat Probability Constant -0.099 -29.88 0.000 Open Ended Fund -0.027 -3.25 0.002 Closed Ended Fund 0.021 +1.69 0.098 Retail Warehouse -0.047 -7.59 “Other” Property Type 0.045 +1.81 0.078 Adjusted R-Squared 50.4% Standard Error 0.025 F Statistic 12.18 Prob F: N Obs = 45, White robust standard errors and covariance

Conclusions Anecdotally, Clients Can Exert Influence on Appraisers to Adjust Their Valuations; To Date, Evidence Only from Anecdote or Second Hand, Survey, Interview or Experimental Methods; Sharp Market Adjustment From 2007 and Differing Client Needs Brings Opportunity for Direct Test; Evidence Consistent with Model of Client Influence; Individual, Disaggregated, Data Would Permit More Robustness in Statistical Testing.