Abstract: | This paper uses a neoclassical investment model extended with installation costs for capital, agency costs for investment financing, and the possibility of the firm being output constrained as a framework for an empirical analysis of investment behaviour in the Swedish manufacturing industry. The theory is implemented within a multivariate error-correction approach on data covering the time period 1951 to 1995, and we gain the following main results: (1) Tobins average Q is not the sole determinant of investment, neither in the short nor in the long run, and other variables like real output and capital gearing also affect investment activity; (2) the out-of-sample forecasts of the model track the evolution of actual investment growth quite impressively, especially at short- and medium-term horizons (1–2 years); (3) a relative equity-price variable is shown to constitute a good approximation of average Q, both for empirical modelling in general and forecasting in particular.Jel classification: C32, E22, E27We would like to thank Bob Chirinko, Stefan Palmqvist, Anders Vredin, seminar participants at Sveriges Riksbank, and two anonymous referees for helpful comments. Thanks also to Jan Södersten, Uppsala University, who provided us with most of the data for this analysis. All correspondence to Per Jansson.First version received: July 2000/Final version received: November 2002 |