Abstract
This paper provides sufficient and necessary conditions for when the elasticity of taxable income (ETI) estimand can be given a causal interpretation as a positively weighted average of heterogeneous individual elasticities of taxable income. We show how the ETI estimand can be used to learn about compensated and uncompensated elasticities by constructing bounds, or obtaining point estimates by either imposing homogeneity assumptions on elasticities or using external estimates of income effects. We apply our results to analyze a reform of the Norwegian tax system that reduced the marginal tax rates on middle and high incomes. Our results show small elasticities of taxable income for middle-income individuals that increase rapidly with income. By combining the ETI estimates with estimates of income effects obtained using lottery winners, we find that (un)compensated elasticities are small for middle incomes but increase steadily with income. Notably, our estimates imply that the Norwegian top-income tax rates exceed the revenue-maximizing one.
