On tests for linearity against STAR models with deterministic trends

Linearity testing against smooth transition autoregressive (STAR) models when deterministic trends are potentially present in the data is considered in this work. Our findings show, in contrast to results recently reported in Zhang (2012), that linearity tests against STAR models lead to useful results in this setting.

► We study linearity tests against STAR when a deterministic trend is present in the data. ► Detrending the data a priori leads to a good performance of the tests. ► An alternative auxiliary regression using a lower polynomial order is shown to have high power.