COINTEGRATION ANALYSIS AND THE DETERMINANTS OF LAND PRICES - COMMENT

Research on testing for stationarity and cointegration has burgeoned in recent years and Hallam et al. (1992) should be commended for their clear introduction to the literature on the subject. This comment complements their paper by drawing attention to some recent developments and showing their relevance to the modelling of land prices. In particular, we highlight (1) the relevance of transforming data series prior to the pretesting for stationarity, (2) the importance of distinguishing between a difference-stationary process (DSP) and a trend-stationary process (TSP) in pretesting data series, (3) the fragility of tests for stationarity and the relevance of using alternative tests to cross-check for stationarity, and (4) the requirement that variables in the cointegrating regression have the same form of stationarity, since a DSP is incompatible with a TSP. In addition, we reject the empirical model favoured by Hallam et al. (1992) in favour of a recently reported econometric model of land prices which is consistent with both economic theory and the requirements of cointegration.