EMPIRICAL ECONOMICS  Springer-Verlag 2001 Empirical Economics 2001) 26:391±405 Unobserved components in an error-correction model of consumption for Southern European countries* Nicholas Sarantis, Chris Stewart Department of Economics, London Guildhall University, 31 Jewry Street, London, EC3N2EY, UK E-mail: stewart@lgu.ac.uk) First version received: January 1999/Final version received: June 2000 Abstract. In this paper we show how the potential misspeci®cation of the consumption function can be ameliorated by approximating any unmodelled long run variation with an unobserved component in the form of a time- varying trend. This methodology is applied to Greek, Portuguese and Spanish consumption functions during the post-second World war period. The empir- ical evidence suggests that there are many determinants of long-run con- sumption in these countries, in addition to income and in¯ation, and these unobserved long-run e¨ects are captured by a nonstationary stochastic component. The long-run elasticity of consumption with regards to the un- observed component is greater than unity in all countries. Key words: Unobserved Components; Consumption; Southern Europe JEL classi®cation: D12, C51, C52 1. Introduction The error-correction approach to modelling consumption and other macro- economic relationships has been widely adopted since Davidson et al 's 1978) path-breaking paper DHSY hereafter). The major feature of the error- correction model ECM) is that it distinguishes between short term and long term e¨ects. The speci®cation of long-run components in the consumption *) This paper was presented at the 1998 Annual Conference of the Money, Macro and Finance Research Group, Imperial College, London, September 1998. We are grateful to the participants of this conference, Andrew Harvey, Guy Judge, two anonymous referees and the editor of this journal for their constructive comments and suggestions on previous versions of this paper. We are responsible for any remaining errors.