THE ACCURACY OF PROPERTY FORECASTING IN THE UK GRAEME NEWELL University of Western Sydney and PATRICK McALLISTER University of Reading June 2009
PROPERTY FORECASTING Importance Uncertainty Procedures quantitative - qualitative Role of judgement 2008 property environment - UK@ -22.1% - Ireland @ -34.2% - Norway@ -4.7% - Sweden@ -3.3% - Spain@ -2.9% - France @ -0.9%
PREVIOUS RESEARCH Forecasting rents, yields etc. Econometric/structural modelling Comparison of forecasting procedures Simple forecasts versus econometric models
ACCURACY OF PROPERTY EXPERT FORECASTS US: Ling (2005) UK: McAllister, Newell and Matysiak(2008); Tsolacos (2006) Australia: Newell and Karantonis (2003); Newell and MacFarlane (2006) Consensus and individual forecasts
ACCURACY ISSUES Uncertainty Disagreement Conservative forecasts; bias Inertia Group differences “Numbers” versus “turning points” Individual forecasters - consistency - banding - persistence
PURPOSE Assess accuracy of UK property forecasts re: 2008 IPD Overall @ -22.1% IPD Office @ -22.4% IPD Retail @ - 22.6% IPD Industrial @ -21.2% Accuracy - uncertainty - disagreement Behavioural issues
METHODOLOGY Investment Property Forum “Survey of Independent Forecasts” 1998 – 2009; quarterly; UK Expert opinions : #= 18-37 - property advisors - fund managers - equity brokers Capital returns, rental growth, total returns Property sub-sectors Forecasts generated to end of year - up to 3 years ahead
METHODOLOGY Focus = 2008 total return forecasts Up to 36 months ahead 36M, 33M, …, 9M, 6M, 3M # 2006-08 participants: 24 – 37 # property advisors: 10 – 18 # fund managers: 9 – 16 # equity brokers: 3 – 5 Statistical analysis - MAE - MAPE - range - Theil U1 statistic
Target = -22.1%
MEAN ABSOLUTE ERROR 36M 24M 12M 6M 3M All : 22.4 19.7 13.3 8.4 5.3 PAs 36M 24M 12M 6M 3M All : 22.4 19.7 13.3 8.4 5.3 PAs 22.8 20.1 13.5 8.7 5.5 FMs 21.7 19.1 12.8 7.5 4.3 EBs 22.5 18.7 9.1 4.6 Office 23.8 20.8 13.1 7.7 Retail 22.3 19.9 14.4 9.5 6.0 Industrial 21.9 19.3 13.4 9.0 5.8
MEAN ABSOLUTE PERCENTAGE ERROR 36M 24M 12M 6M 3M All : 101.1% 89.1% 60.2% 38.0% 24.0% PAs 103.0% 90.7% 61.2% 39.1% 24.9% FMs 98.2% 86.5% 57.9% 33.9% 19.5% EBs 101.9% 84.7% 60.3% 41.0% 20.8% Office 106.2% 92.8% 58.4% 34.4% 20.5% Retail 98.9% 88.2% 63.5% 42.0% 26.6% Industrial 103.4% 91.0% 63.4% 42.2% 27.4%
THEIL U1 STATISTIC 36M 24M 12M 6M 3M All : 0.80 0.70 0.44 0.25 0.14 36M 24M 12M 6M 3M All : 0.80 0.70 0.44 0.25 0.14 PAs 0.81 0.72 0.45 0.26 FMs 0.79 0.69 0.43 0.22 0.11 EBs 0.66 0.28 0.12 Office 0.82 Retail 0.47 0.15 Industrial 0.16
AVERAGE RANGE 36M 24M 12M 6M 3M All : 10.30 10.70 10.40 10.20 9.50 PAs 36M 24M 12M 6M 3M All : 10.30 10.70 10.40 10.20 9.50 PAs 7.10 7.40 7.50 5.60 FMs 8.40 8.30 7.70 6.60 EBs 5.70 7.00 7.30 6.80
“BEST” FORECASTER: MAE 36M 24M 12M 6M 3M All : 16.70 14.10 7.80 3.70 0.60 PAs 18.70 16.00 9.30 5.10 2.10 FMs 17.30 14.80 8.60 4.00 1.10 EBs 19.60 15.40 10.10 5.90
“BEST” FORECASTER: MAPE 36M 24M 12M 6M 3M All : 75.50% 63.70% 35.40% 16.70% 2.70% PAs 84.60% 72.60% 42.10% 23.10% 9.50% FMs 78.30% 67.10% 38.70% 17.90% 5.00% EBs 88.90% 69.60% 45.50% 26.50% Theil 0.58 0.48 0.25 0.12 0.01
“BEST” FORECASTER Groups: Individuals: PAs: 0% - FMs: 75% - EBs:25%
“WORST” FORECASTER: MAE 36M 24M 12M 6M 3M All : 27.0 24.8 18.3 13.9 10.1 PAs 25.8 23.6 17.0 12.7 7.7 FMs 23.1 16.3 10.5 EBs 25.3 22.4 17.4 12.6
“WORST” FORECASTER: MAPE 36M 24M 12M 6M 3M All : 122.20% 112.00% 82.60% 62.90% 45.70% PAs 116.90% 106.60% 76.90% 57.50% 34.80% FMs 116.50% 104.60% 73.70% 47.50% EBs 114.70% 101.30% 78.50% 57.00% Theil 0.91 0.86 0.67 0.47 0.30
“WORST” FORECASTER Groups: Individuals: PAs: 42% - FMs: 0% - EBs:58%
PROPERTY FORECASTING IMPLICATIONS Accuracy re: 2008 property forecasts Uncertainty versus disagreement Conservative bias Improvements over time : 36M 3M Critical times Group differences Sector differences Other issues re: changes in forecasts - impact of news - expected returns (IPD monthly) - anchoring 2009 property forecasts?