Forecasting rental rates for Norwegian commercial real estate

Abstract

This study seeks to identify key determinants of rents of commercial real estate in Oslo and formulate econometric models capable of describing and predicting their movements. Such a model will improve the precision of property valuations and be a useful aid in making real estate related investment decisions. The study finds real rental rates to be a function of previous periods’ rents, employment rates, real interest rates and vacancy rates. The forecast models examined are a classical linear regression model, an autoregressive moving average (ARIMA) model and a vector autoregressive (VAR) model. The performance of these are evaluated using root mean squared errors (RMSE), mean absolute errors (MAE), mean absolute percentage errors (MAPE) and Theil’s u-stat as well as variance decomposition and the percentage of correct signs predicted by the model compared to the actual values. The study concludes that given the available data, the classic linear regression model is able to produce the most precise forecasts, although the precision is not satisfactory. None of the forecasts are at present able to consistently beat a random walk, but a clear trend of improvement in forecast accuracy is detected when gradually increasing the estimation sample.