This study on overinvestment differs from the existing literature in that investment in machinery and equipment is modelled as a structural vector autoregression with identification achieved by imposing long-run restrictions, as in Blanchard and Quah (1989).