The tax reform literature, pioneered by Guesnerie [1977], uses static models but views tax reform as a dynamic process, i.e., as a policy-maker implementing incremental reforms over time. This paper studies tax reform in a dynamic version of the Diamond-Mirrlees-Guesnerie model and focuses on a specific aspect of the dynamic process, namely, the implications for tax reform of agents leaving bequests. The main idea is that a tax reform in one period will affect bequests and therefore endowments, equilibrium, and welfare in subsequent periods. Thus, the process of tax reform cannot be analyzed as a sequence of static economies; instead, the economies are linked by bequests. The paper undertakes a tax reform analysis a la Guesnerie, but with an added focus on welfare improving reforms for each generation. Second-best Pareto optima are then characterized, and these conditions are compared to the static optimal tax formulae derived in the literature. In particular, the key Diamond-Mirrlees result that production efficiency is desirable at second-best optima no longer holds in the presence of (effective) restrictions on the taxation of private savings. Restrictions on government savings (including balanced budget restrictions), however, do not disturb the desirability of production efficiency. Finally, the effects of certain political constraints on the tax reform process are also consider
