Equity, incentives, and taxation (Q1187685)

This monograph addresses itself to the following questions; in the presence of production, which states are fair? Are these states attainable by means of an income tax? What states are attainable if other tax systems are used? Are there other tax systems which are better suited to implement more allocations? How effective are taxes as an instrument for insurance? Answers to these questions are provided within the framework of a Mirrlees type model where there are two goods - consumption and labor; labor produces the consumption good under constant returns to scale. Agents differ in tastes and their ability (except while tackling the insurance question, when all agents are identical), where ability is measured by a parameter n i.e., an agent with ability n working for h hours supplies nh efficiency units of labor which can be converted to nh units of output. The study begins by looking at equity criteria suggested in \textit{L. Foley} [``Resource allocation and the public sector'', Yale Economic Essays (1967)] and \textit{H. R. Varian} [J. Econ. Theory 9, 63-91 (1974)] and shows how in the presence of production, fair states according to the former may not exist; whereas the second criterion leads to non-trivial fair states when there is a finite number of agents. In the presence of a continuum of agents, however, there is a unique fair allocation - the Laissez-Faire state. Also it is shown that income taxes serve to attain such states only if the second criterion is used in the context of finite economies; moreover, by using an income tax, it is not possible to implement an equitable allocation, if there are at least two agents with identical preferences but differences in ability. Since income taxes depend only on the quantity y, viz., income, other taxes would involve knowledge of other characteristics (signals). Within the context considered, there are the following possibilities: the signal n (or wage rate), the signal h (or hours worked) or a combination of y and h, say. Taxes depending on single signal and on the combination y and h are considered. The purpose is to find out which of these signals is `more informative'. If the tax based on a signal s induces an allocation then the allocation is said to be s-implementable. And a signal \(s_ 1\) is more informative than another signal \(s_ 2\) if all \(s_ 2\)- implementable states are \(s_ 1\)-implementable. No general results are obtained in this connection except that the joint signal (y,h) is shown to be more informative than single signals apart from trivial cases. The final section, on insurance, considers a stochastic model with uncertain income; moral hazard aspects are considered. The method of shaping an optimal tax in such situations is also considered.