Purchasing Power Parity and Spanish Provinces, 1940-1992

In this paper we empirically assess the validity of the PPP hypothesis from a new point of view. Instead of examining international data, we analyse PPP for Spanish provinces, eliminating many factors frequently offered to explain the failure of PPP. To that end, we make use of annual CPI data covering the 1940-1992 period. We apply time series analysis that allows for the presence of breaks in the trend of that series. We examine all possible bilateral relationships between provincial CPIs, running 2,450 regressions and computing 7,450 statistical tests. Our results are very suggestive since the PPP hypothesis is rejected by the data, even though in our case potential factors explaining the failure of PPP do not apply. Indeed, using robust versions of traditional cointegration techniques, although we find long-run relationships between provincial CPIs in many cases, such relationships do not satisfy in general the restrictions imposed by the PPP. On the other hand, our results indicate long-run relationships for almost all bilateral combinations of provincial CPIs when structural breaks are taken into account, being such breaks statistically significant. In this way, we illustrate how the formal consideration of eventual structural breaks may be useful for a more correct specification of an econometric model. Nevertheless, even with structural breaks the restrictions implied by PPP do not hold.