Scorrrsh Journal zyxwvutsrq of zyxwvutsrqpon Polifical Economy, zyxwvutsrq Vol. zyxwvut 40, No. 2. May 1993 zyxwvut SJ Scottish Economic Society 1993. Published by Blackwell Publishers, 108 Cowley Road, Oxford zyxw OX4 IJF. UK and 238 Main Streel. Cambridge. MA 02142. USA zyxwvutsr A STRUCTURAL VECTOR AUTOREGRESSION MODEL OF THE UK BUSINESS CYCLE PAUL MICHAEL TURNER* School of Business and Economic Studies, University of Leeds I INTRODUCTION There has been a definite increase in the intensity of research effort devoted to the analysis of business cycles in recent years. Prominent in this has been the real business cycle (RBC) literature surveyed by Plosser (1989). However, there have also been theoretical contributions from a Keynesian perspective in the work of Mankiw (1985) and Akerlof and Yellen (1985). In terms of the empirical work there has been increased attention towards the analysis of busi- ness cycle ‘facts’ through the work of Kydland and Prescott (1990) and, in a British context, Blackburn and Ravn (1990). In addition there have been models of the cyclical process which have adopted an explicitly Keynesian structure (Blanchard and Watson, 1986; Blanchard and Quah, 1989; Blanchard, 1989). In this paper I seek to apply the Keynesian modelling strategy of Blanchard et al. to the UK economy. The plan is as follows: In Section I1 I review and compare the Cowles Commission approach to structural model building and the ‘theory free’ approach of Vector Autoregressions. The model to be esti- mated is described in Section 111. In Section IV the procedures used to estimate the model are discussed with particular emphases on the treatment of the trend component of the series and on the identification of the structure of contem- poraneous interactions between variables. The estimates themselves, coupled with simulations of the model, follow in Section V, with Section VI containing conclusions. *Thanks are due from the author to Terry Mills for many helpful comments on several drafts of this paper, to Keith Blackburn for providing an awareness of the Hodrick-Prescott filter, to Edward Prescott for the Fortran algorithm used to compt,e it, to participants in a seminar at the Money, Macro and Finance Research Group conference and to two anonymous referees for numerous useful comments. All opinions and errors in the paper remain those of the author. Date of receipt of final manuscript: 3 August 1992. 143