A structural time series test of the P-star model: evidence from the middle east

Abstract

In this article, a structural time series test of the P-star model is conducted using quarterly data for the Middle Eastern countries of Egypt, Jordan and Morocco. The conventional P-star model is altered to obtain an equation for the price level that consists of a stochastic trend and the actual levels of output and velocity. The empirical results obtained are highly supportive of the model and show the perils of modelling output and velocity as deterministic rather than stochastic trends. Estimates of the dynamic relationship between the price gap and the inflation rate are also highly supportive of the adjustment mechanism inherent in the P-star model, producing a large and significant coefficient of adjustment for each country.

Notes

¹ In his study of the P-star model, Christiano (1989) heavily criticises its use as a forecasting tool. He claims that it does no better or worse than a conventional 3-month Treasury Bill model using quarterly observations for the period 1970 : 4-1989 : 3. As such, he concludes that the P-star model is not the inflation forecaster's ‘holy grail’.

² Harvey (1989, 1990, 1993) and Koopman et al . (2000) provide details of the state space representation and estimation of these models.