Abstract
A growing literature in the field of econometrics is on the treatment of seasonal variables. However, so far, very few studies in India have applied advanced seasonal modelling techniques to important macroeconomic variables. This paper examines the seasonal properties of Indian monthly WPI inflation and their usefulness in modelling the series more efficiently. Monthly WPI inflation was found to be a periodic process with 18 lags and periodic integration of order two. A comparison between the performances of a PAR (18) and an AR (18) model showed that the former performed substantially better in terms of R2, AIC, in-sample predictive ability and residual properties. However, the out-of-sample forecasts from the PAR model were only reasonable. The best forecasts obtained for a horizon of 22 months, however, had good ‘direction of change’ predictions. The model could also produce interval forecasts of modest accuracy.
Keywords
Get full access to this article
View all access options for this article.
