Appendix: Is the EITC as Good as an NIT? Conditional Cash Transfers and Tax Incidence

Appendix Figures 1 and 2 present estimates of the net transfer to workers, separately by marital status and the presence of children, under a 27 different elasticity parameter combinations. These extend the results from Figure 5 of the main text showing the total transfer pooled across all workers. B Appendix: Incidence with income effects The specifications in Section 2.4 assume that income effects are zero. In this appendix, I derive formulas for the incidence of tax increases when income effects are non-zero. I begin by defining two behavioral elasticities corresponding to income effects on the extensive and intensive margin. Let r be non-labor income (including husband’s earnings, which are treated as exogenous) less the taxes that would be owed on this income if the woman did not work. Let v be virtual income, the intercept of the linear segment of the budget constraint on which a family with a working woman lies. These are related by v = r+wL(MT R−ATR), where L = ph is the woman’s labor supply, wL is her earnings, MT R is the marginal tax rate, and ATR is the average tax rate on the woman’s earnings. A common definition of the income elasticity of labor supply is η ≡ NYL ∗ ∂L ∂v, whereNY is net-of-tax total income (including earnings; NY ≡ v + wL(1−MTR) = r + wL(1−ATR)). Conventionally, the second fraction is the derivative of labor supply with respect to virtual income but the first uses net income instead. A natural extension of this to allow for distinct extensive and intensive margins is to define the participation elasticity as ηe ≡ NYp ∗ ∂ p ∂ r and the elasticity of hours conditional on participation as ηi ≡ NY h ∗ ∂h ∂v. Note that although these are defined with respect to different income concepts, a $1 increase in the family’s exogenous income raises r and v by the same amount. A drawback to these definitions is that non-labor and virtual income need not be positive. For example, single mothers in the phase-in segment of the EITC are likely to have quite low non-labor income (and for this group v = r). Small changes in tax parameters can therefore induce enormous changes in lnr and lnv, if indeed these are well defined. As an alternative, I focus on quasi-elasticities: η̃e ≡ ∂ ln p ∂ r = 1 N