An important result of economic theory is that the efficiency loss associated with taxation of income is higher the more progressive the tax system is. I.e. the higher the marginal tax rate compared to the average tax. The reason for this is that the tax on the last earned euro is interfered with the players’ marginal adjustment in the labour market. When a worker must decide whether he wants to work a few extra hours each week, so there is the additional tax he has to pay for these additional euros causing undesirable biases in the adaptation, not the level of taxation. This reasoning is based on a basic assumption that all workers are free to choose the schedule that is best for them. In practice, we know that many are rationed in the labour market, and that some jobs have more or less given working hours. To the extent that the tax affects adaptation to persons who must choose between jobs with given working hours, it is the total tax burden associated with each of these choices that are relevant, not the marginal tax rate.
The argument is also based on an assumption that all people are equal, i.e. they have the same preferences and that they face the same hourly wages. Formally speaking, this is usually reflected in economic models describe a “representative agent”. This assumption is of course not done because anyone really believes that all people are equal, but because it provides a simpler model. This is a technique that in many cases are completely “harmless” in the sense that the “representative agent” can help us to reveal relationships that will also apply under more general assumptions. But when it comes to the analysis of tax efficiency it shows that the assumption representing is anything but harmless. It may instead give rise to totally erroneous conclusions about how taxes affect individuals’ employment.
If people are different with respect to the degree of substitutability between leisure and consumption it will not be effectively to exposing all individuals for exactly the same marginal tax rate. If we move some of the taxes from a person with flat indifference curve to a person with a steeper indifference curve would it obviously be possible to reduce the overall deadweight loss. We may naturally not expose various individuals for various tax systems based on knowledge of their indifference curves. But if it should prove to be a systematic correlation between wages on the one hand and indifference curve curvature on the other, this could be an argument for exposing different income levels for different marginal tax rates.
Ole Andreas Lågøen