Evaluating forecasting models of farmland prices

A set of rigorous diagnostic tools is used to evaluate the forecasting performance of five farmland value models. The models are two variations of the present-value model, an ARIMA, a vector autoregression, and an error-correcting model. One- and three-period-ahead out-of-sample forecasts are evaluated in terms of forecast accuracy (root mean-squared error) and the ability to predict turning points (Henriksson-Merton test). By the Henriksson-Merton test, the error-correcting model generates superior forecasts at both forecasting horizons. The vector autoregressive model performs poorly by root mean-squared error criterion, as does ARIMA in predicting turning points in the three-period-ahead forecast