Endogenously censored median regression with an application to benefit elasticity of US unemployment duration

- We suggest a new inference method for median regression.
- The method is robust to assumptions on the censoring mechanism.
- The method performs well in Monte Carlo simulations.
- We use our method to obtain bounds for benefit elasticity of unemployment durations.
- Using SIPP data we find that this elasticity is between −1.1% and +7.9%.

We suggest a new inference method for median regression which is robust to assumptions on the censoring mechanism. Applying the method on the SIPP data we obtain that the benefit elasticity of median unemployment duration is between -1.1% and +7.9%.