Political Economy and the Structure of Taxation∗

We study the constrained Pareto efficient allocations in a dynamic production economy in which the group in political power decides the allocation of resources. We show that Pareto efficient allocations take a quasi-Markovian structure and can be represented recursively as a function of the identity of the group in power and updated Pareto weights. For high discount factors, the economy ultimately converges to a first-best allocation in which there may be transfers between groups, but labor supply decisions are not distorted and the levels of labor supply and consumption do not fluctuate over time. When discount factors are low, the economy converges to an invariant stochastic distribution in which distortions do not disappear and labor supply and consumption levels fluctuate over time. In these allocations with distortions, the labor supply of individuals from groups that are not in power are taxed, while the labor supply of those in power is subsidized. The subsidies are useful to relax the political economy/sustainability constraints. We also show that the set of sustainable first-best allocations for high enough discount factors are independent of the Markov process for power change. This result contradicts a common conjecture that there will be fewer distortions when the political system creates a “stable ruling group”. The reason why this conjecture is incorrect is that social groups can be rewarded not only when they hold power, but also when they engage in production. Consequently, the probability of power switches do not directly affect “effective discount factors”. Nevertheless, it remains true that distortions decrease along sample paths where a particular group remains in power for a longer span of time. Finally, we demonstrate that the constrained efficient allocation can only be decentralized using distortionary taxes (even when the political system has access to lump-sum taxes), so that the results about fluctuations of distortions, consumption and labor supply levels correspond to fluctuations in taxes and redistribution. ∗We thank seminar participants at the MIT, University of Pennsylvania and Stanford.