Corporate taxation and investment: Explaining investmentdynamics with firm-level panel data

Using a ¯rm-level panel data set I assess whether dynamic models of in-vestment provide an empirically fruitful framework for analyzing tax e®ectson changes in capital stock. In particular I estimate a one-step error correc-tion model (ECM) complementing the usual estimation of a distributed lagmodel. A correction term accounts for non-random sample attrition, whichhas not been considered in previous studies on investment even though most(if not all) panel data sets on ¯rms are incomplete. Both, ECM and dis-tributed lag model, suggest that user cost of capital and output have aneconomically and statistically signi¯cant in°uence on capital formation. Inthe ECM, however, estimates are larger in size and match theoretical pre-dictions more closely. My preferred estimate of -1.3 implies that a decreasein the user cost of capital by 10 percent will increase the ¯rm's capital stockby 13 percent, on average.[...]