Unit root testing against an ST-MTAR alternative: Finite-sample properties and an application to the UK housing market

Abstract

A class of smooth transition momentum-threshold autoregressive (ST-MTAR) tests is proposed to allow testing of the unit root hypothesis against an alternative of asymmetric adjustment about a smooth non-linear trend. Monte Carlo simulation is employed to derive finite-sample critical values for the proposed test and illustrate its attractive power properties against a range of stationary alternatives. The empirical relevance of the ST-MTAR test is highlighted via an application to aggregate house price data for the UK. Interestingly, house prices are found to exhibit structural change characterised a fitted logistic smooth transition process, with the newly proposed ST-MTAR test providing the most significant results of the alternative smooth transition unit root tests available.