In order to assess the efficiency of a tax, we should examine its effect on the behavior of individuals. In general, the less a tax affects behavior, the more efficient it is thought to be. The standard example of a non-distorting tax is a lump-sum tax, which does not change with the behavior of the taxpayer. However, this article demonstrates that behavioral distortions can and do arise from a change in even a lump-sum tax. The only truly non-distortionary tax would be one based on utility itself. Utility, which has been used as a norm for distributional analysis, is also the ideal base for efficiency analysis. In fact, any reasonable attempt to describe a minimally distortive basis of taxation will significantly resemble the notion of a tax on utility. Therefore, utility itself is the best basis for evaluation of the efficiency of a tax. Such a tax has many additional features which make it more useful for analytical purposes than lump-sum taxes.
Income taxes and consumption taxes have assumed a very prominent place in public finance in large measure because they are thought to be relatively efficient bases for taxation (Tresch, 2003). These taxes are considered efficient because it is believed that the behavioral response to these taxes is considered to be less elastic to marginal increases in the rate of taxation when compared with other bases of taxation such as excise taxes on goods and services (Ramsey, 1927; Tresch, 2003). This lower behavioral response is commonly believed to result in lower losses to welfare than most other potential tax bases. However, as explored in this article, these conclusions are based on a number of assumptions about the utility functions of individuals which may not be correct.
The primary intent of this article is to put forth three very simple points. First, the standard arguments about how one measures the efficiency of income and other taxes are incomplete. This does not mean they are not useful, but they ignore many important aspects of behavior. Second, these lacunae can cause our analysis to be blind to important distortionary elements of a tax. If we do not understand what causes behavior, it will be difficult to prevent from distorting behavior unintentionally. Third, it is therefore necessary for any welfare analysis of taxation to study the actual utility functions of individuals. It is here that behavioral economics, experimental economics and other empirical economic research can have a significant impact on our analysis. In particular, this article will discuss how research tends to indicate that wealth or income enters the utility function not as single number (or scalar), but rather in multiple parts (i.e., as a vector).