Descriptive econometrics for non-stationary time series with empirical illustrations

Recent work by the author on methods of spatial density analysis for rime s eries data with stochastic trends is reviewed. The methods are extended to include processes with deterministic trends, formulae for the mean spatial density are given, and the limits of sample moments of non-stationary data are shown to take the form of moments with respect to the underlying spatia l density, analogous to population moments of a stationary process. The met hods are illustrated in some empirical applications and simulations. The em pirical applications include macroeconomic data on inflation, financial dat a ori exchange rates and political opinion poll data. It is shown how the m ethods can be used to measure empirical hazard rates for inflation and defl ation. Empirical estimates based on historical US data over the last 60 yea rs indicate that the predominant inflation risks are at low levels (2-6%) a nd low two-digit levels (10-12%), and that there is also a significant risk of deflation around the -1% level. Copyright (C) 2001 John Wiley & Sons, L td.