Approximately Median-Unbiased Estimation of Autoregressive Models.

This paper introduces approximately median-unbiased estimators and confidence intervals for univariate AR(p) models with time trends. The methods are applied to the Nelson-Plosser macroeconomic data series, the extended Nelson-Plosser series, and some annual stock dividend and price series. The results show that most of the series exhibit substantially greater persistence than least squares estimates and some Bayesian estimates suggest. For example, for the extended Nelson-Plosser data, eight of the fourteen series are estimated to have a unit root, while six are estimated to be trend stationary. In contrast, the least squares estimates indicate trend stationarity for all of the series.