An Assessment of Property Performance Forecasts: Survey versus Econometric

Property performance forecasts in the UK are based both on econometric and qualitative methodologies and are produced by in house teams of firms and consultancies. These forecasts are for client use and hence they are not broadly circulated. In the last seven years the Investment Property Forum (IPF), a leading property industry body in the UK, has published point and range forecasts of rents, capital values and total returns for UK property for one and two year horizons. This is a survey of independent forecasts in the industry. The contributors to this survey range from property advisors and consultancies to fund managers and equity brokers. This initiative brings the property industry in line with similar surveys of economic and financial series forecasts (HM-Treasury and Consensus Economics surveys). The IPF forecast survey results are communicated to IPF's wide membership and the survey certainly represents a barometer of the market. Evaluating the accuracy of these forecasts is critical if they are to be used in decision making. The quality of forecasts (either consensus or from an individual source and either econometric or otherwise) has been the subject of extensive research in economics. Among other issues, questions of interest in this literature relate to how efficient the forecasts are in the sense of fully incorporating the information available at the time, the size and bias of forecast errors, how the forecast horizon and accuracy interact and whether consensus forecasts perform better than any individual forecast. Statistical measures of forecast performance beyond the usual measures (such as mean absolute errors, root mean squared errors, regression tests of forecast efficiency and bias and the Theil statistics) continue to develop in the forecast accuracy literature providing the basis for assessing property performance forecasts. The aim of this paper is two fold. First, it assesses the IPF consensus forecasts using established methodologies and criteria. Second, the IPF forecasts are compared to those obtained from econometric models and more naOve techniques. The econometric forecasts are out of sample using data available to the analyst at the time of the forecast. In particular tests are carried out to examine whether one forecast dominates another in terms of information content and whether any benefits can be obtained from the combination of the survey and econometric forecasts. Based on recent developments in forecast evaluation, the analysis is furthered to address two additional issues: (i) whether revisions to the forecasts are predictable and (ii) pooling information over forecast horizons rather than analyse the forecasts separately for each horizon. The second point is pertinent to this study given the small number of observations.