Abstract
This paper reconsiders the equilibrium correction model of nondurable consumption in the UK by Davidson et al. (1978), denoted DHSY. The DHSY model fails outside the original observation period and several studies claim that this is due to neglected nonlinearities or time-varying parameters. This paper will take both features into account simultaneously by using the methodology of smooth transition regressions (STR). The study is performed both for the original sample and for an extended one. It turns out that nonlinearities are present in both samples when keeping the original model specification. The resulting consumption functions are characterised by time-varying parameters rather than nonlinear relationships between the explanatory variables. The estimated nonlinear models encompass, and dominate in variance, their linear equivalents for each sample.