Econometric modelling and forecasting of new retail development

Abstract

This paper provides a study of new retail development, proxied by the volume of new orders for retail building works, at the aggregate level in Great Britain. A regression model and a VAR system are estimated based on a theoretical framework that relates the level of new orders to changes both in real consumer expenditure and real retail rents. The former variable is used as a macroeconomic series to capture trends in the business of retailing and demand for retail space. Retail rents are included as an indicator of demand-supply imbalances in the market and potential profitability of retail projects. The empirical estimates confirm the significance of the proposed framework in the study of new retail construction. Tests also establish the reliability of the estimated regression and VAR models for forecasting purposes. In particular quarters the models predict less accurately owing to the significant quarterly volatility of the new orders series. Overall the regression model tends to produce more accurate forecasts. Ex ante forecasts for the period mid-1997 to 1999 based on the regression model suggest that the upward trend in the volume of new retail orders which began in 1992 will continue if real expenditure and real rents increase at an annual rate of four and six per cent respectively, but will reverse in 1998 if these variables exhibit growth rates of two per cent per annum. The VAR model predicts a further but moderate increase in new orders over this forecast period.