A reproducibility assessment of "Immigration Restrictions as Active Labor Market Policy: Evidence from the Mexican Bracero Exclusion"

Clemens et al. (2018) examine the impact of the exclusion of Mexican immigrant workers, known as braceros, from US agricultural labour markets in the 1960s. The explicit goal of this policy was to boost wages and/or employment for local agricultural workers by decreasing labour supply. The authors present a theoretical framework which argues that this policy may not have the desired impacts if firms can adjust capital inputs in response. Their main finding is that there is no effect of the bracero exclusion on the wages or employment of local agricultural workers. They provide evidence of increased mechanisation in response to decreased labour supply. We conduct a computational and robustness reproduction of the main results presented in Clemens et al. (2018). Using the reproduction package provided by the original authors, we successfully computationally reproduce the main results. Our robustness reproduction comprises three distinct robustness checks. We test the sensitivity of the results to changing the base year used to define treatment. We implement a more recently proposed difference-in-difference estimators in the presence of a continuous treatment variable. Finally, we change how missingness is dealt with in one outcome variable-domestic seasonal employment. Our results do not provide compelling evidence to challenge the findings of the original paper. However, we note that large confidence intervals around point estimates make it difficult to draw conclusions of a precise null effect.