Doing Wonders with an Egg: Optimal Re‐distribution When Households Differ in Market and Non‐Market Abilities

Abstract

The paper studies non-linear income taxation and linear commodity taxation in a household production context with households differentiated by market and non-market ability. In such a setting, there is an efficiency motive for re-distribution which is independent from the usual equity motive, and operates also when the social planner is indifferent to utility inequality. As a consequence, some of the policy prescriptions applicable to the case in which households differ in market ability only do not hold when households differ also in nonmarket ability. For instance, re-distribution is not necessarily from high-to low-wage households, and it is not necessarily true that the marginal rate of income tax should be zero for high incomes and positive for low incomes. In some cases, re-distribution may accentuate rather than lessen utility inequality, and can reverse the direction of income inequality relative to the laissez-faire equilibrium. Furthermore, contrary to Atkinson-Stiglitz, it may be optimal to use indirect and direct taxation simultaneously even when the utility function is separable in commodities and labour.

Direct versus Indirect Taxation: Trends, Theory, and Economic Significance

“…Cremer et al (2001) show that when individuals differ in several unobservable characteristics (productivity and endow ments), differential commodity taxes do have a role to play as instruments of optimal tax policy -an optimal (general) income tax will not suffice, while the optimal commodity tax rates follow traditional Ramsey rules. Papers by Saez (2002) and Balestrino et al (2003) make contributions along similar lines. Naito (2004) finds that using a commodity tax can increase social welfare in the presence of a non-linear income tax system when human capital accumulation is endog enous.…”

Optimal Utilitarian Taxation and Horizontal Equity

“…Under certain circumstances, both constraints could even be binding at the same time (generating what Brito et al (1990) call a "self-selection cycle"). Notice that the mentioned properties are a common feature of all models that introduce heterogeneity along more than one dimension (see Balestrino, Cigno and Pettini (1999) and Cremer, Pestieau and Rochet (2001)). Finally, notice also that although the single crossing condition holds in our model, it will not generally do so in models with more than one differentiating characteristic of the agents.…”