Does labour income react more to income tax or means-tested benefits reforms?

I provide estimates of compensated elasticities of labour income with respect to the marginal net-of-tax rate on the 2006–15 period for France, exploiting not only income tax reforms but also means-tested benefits reforms. I use semi-parametric graphical evidence and a classic two-stage least-squares estimation applied to a rich data set including both financial and socio-demographic variables. I obtain an estimated compensated elasticity on the intensive margin around 0.2–0.3 in response to income tax reforms, and around 0.1 in response to in-work benefit reforms, while I found no statistically significant response to family allowance reforms. I show that the difference between elasticities contradicts the theoretical prediction of the classical labour supply model. These asymmetric responses are very robust to a large number of robustness checks. The most plausible explanation is that income tax reforms are more salient and better perceived than benefit reforms. I also highlight an average compensated elasticity of 0.1 for all transfers on the intensive margin and provide heterogeneous elasticities depending on types of people, which could be used for optimal tax analyses.