Angrist, J.D., & Evans, W.N. (1998). Children and Their Parents' Labor Supply: Evidence from Exogenous Variation in Family Size. The American Economic Review, 88(3). 450-477.
Angrist and Evans (1998) focuse on the endogeneity problem arising when trying to estimate the labor-supply consequence of childbearing for women. The authors use an instrumental variable approach by exploiting the parental preferences for a mixed sibling-sex composition (i.e., children of different gender) to construct instrumental variables using same sex and twin variables. The findings of the paper is that IV estimates statistically significant and smaller than OLS, suggesting that OLS might exaggerate the effect of childbearing on labor-supply outcomes. For samples of all women and married women, the estimated coefficients of having a third child are negative, which suggest that childbearing has a negative effect on labor-supply.
The replication is carried on the main notebook from this repository. The auxiliary folder contains different functions made for the cleaning of 1980 and 1990 Census Extracts, data preparation, summary statistics, regressions and tables, and plots. It also contains pickled files for the results of simulations using causal forests as part as the proposed research extensions for estimating heterogenous treatment effects across different subgroups of schooling levels for women. To ensure the reproducibility of the proyect, the code for creating the pickled files can be found on the notebook as a code block.
General structure of the notebook:
- Introduction
- Identification Strategy
- Empirical Methodology
- Replication Angrist & Evans (1998)
- Extensions on Angrist and Evans (1998) using 1980 Census Extracts
- Critical assesment
- Extension I: Exploration on the probability of additional childbearing conditional to women's education
- Extension II: Causal Forest application for an alternative estimation of heterogeneity effects
- Extension III: Testing for selection on unobservables or "violation of exclusion restriction"
- Conclusions
- Appendix: Variable Dictionary
The notebook that contains my replication of the results of Angrist and Evans (1998) can also be visualized using nbviewer and mybinder under the following badges:
Also, my repository is supported by GitHub Actions Continuos Integration (CI). The state of my workflow can be found here:
-
Angrist, J., & Evans, W. (1998). Children and Their Parents' Labor Supply: Evidence from Exogenous Variation in Family Size, 88(3), The American Economic Review.
-
Athey, S., and Wager, S. (2019). Estimating treatment effects with causal forests: An application. Observational Studies, 5(2), 37-51.
-
Black, D., Joo, J., Lalonde, R., Smith, J., Taylor, E. (2015). Simple Test for SelectionB Bias: Learning more from instrumental variables. IZA Discussion Papers 9346, Institute of Labor Economics (IZA).
-
Cunningham, S. (2021). Causal Inference: The Mixtape. Yale University Press
-
Naushan, H. (2021). Causal ML for Data Science: Deep Learning with Instrumental Variables.
-
Stoffi, F., and Gnecco, G. (2019). Estimating heterogeous causal effects in the presence of Irregular assignment mechanisms. arXiv:1808.04281
-
Winship, C., & Morgan, S. L. (2007). Counterfactuals and causal inference: Methods and principles for social research. Cambridge, England: Cambridge University Press.