Tax Neutrality in the Presence of Adjustment Costs

Abstract

It is well-known that expensing of capital expenditures will lead to tax neutrality with respect to investment, provided that interest payments are not tax deductible [Brown, 1948]. However, if interest payments are deductible, then allowing firms to deduct economic depreciation will lead to a neutral tax system [Samuelson, 1964]. These neutrality results have been derived in models without costs of adjustment (see also, Hall and Jorgenson [1971], Stiglitz [1973], and King [1974]). However, recent research in investment theory has emphasized the role of costs of adjustment [Lucas, 1967; Gould, 1968; Hayashi, 1982; Abel, 1982a,b]. Using a separable adjustment cost technology in which adjustment costs depend only on the rate of investment, Abel [1982a,b] has shown that the two neutrality propositions above continue to hold. However, using a general nonseparable adjustment cost technology, Hartman [1978] showed that only one of the two neutrality propositions remains valid in the presence of adjustment costs. Expensing continues to lead to neutrality if interest payments are not deductible, but Hartman finds that economic depreciation and interest deductibility do not lead to neutrality in the presence of adjustment costs. In this paper we analyze investment behavior in an adjustment cost framework and construct a neutral tax system with interest deductibility. The neutral tax system allows for economic depreciation but requires some adjustments to taxable income which recognize the fact that capital can implicitly produce net revenue by reducing adjustment costs. Hartman found economic depreciation to be nonneutral because he did not make this necessary adjustment to taxable income and because he calculated economic depreciation incorrectly using the price of uninstalled capital rather than the value of newly installed capital. In the presence of costs of adjustment, true economic depreciation is calculated using the shadow price of installed capital rather than the observable market price of new uninstalled capital. It has been shown by Hayashi [1982] that under appropriate linear homogeneity and price-taking assumptions, the shadow price of capital can