Seminarium: Ola Vestad, SSB

Abstract: We study a mid-2000s reform that lowered the top marginal tax rate on labor income. We first show that the conventional elasticity-of-taxable-income (ETI) estimand fails to recover a positively weighted average of individual elasticities, leading to a biased estimate. Correcting for this bias raises the ETI from 0.16 to 0.23. We then derive the restrictions that the corrected ETI imposes on (un)compensated earnings elasticities and on income effects and show how they can be point- or partially identified under different (sets of) parametric assumptions. Common parametric specifications are rejected by the data. Finally, we combine the tax reform with exogenous shocks to unearned income from lottery wins to point identify income and substitution effects without parametric assumptions. The estimated compensated elasticity is almost twice the conventional ETI estimate. Taken together, our results imply that the efficiency costs of taxation can be substantial even when the conventional ETI appears small.