Estimating the elasticity of taxable income when earnings responses are sluggish

Estimates of the elasticity of taxable income (ETI) is conventionally obtained by "stacking" three-year overlapping differences in the estimation. In effect, this means that the ETI estimate is an average of first-, second-, and third-year effects. The present paper draws attention to this implication and suggests that if there is gradual adjustment the analyst should rather estimate the ETI by a dynamic panel data model. When using Norwegian income tax return data for wage earners over a 14-year period (1995-2008) in the estimation, an ETI estimate of 0.15 is obtained from the dynamic specification, compared to 0.11 for the conventional approach. Importantly, the conventional approach fails to render a long-term elasticity estimate by increasing the time span of each difference.