Unobservable Selection and Coefficient Stability: Theory and Evidence
By E. Oster. 2019
4,152 citations recorded by the source corpus.
A common approach to evaluating robustness to omitted variable bias is to observe coefficient movements after inclusion of controls. This is informative only if selection on observables is informative about selection on unobservables. Although this link is known in theory in existing literature, very few empirical articles approach this formally. I develop an extension of the theory that connects bias explicitly to coefficient stability. I show that it is necessary to take into account coefficient and R-squared movements. I develop a formal bounding argument. I show two validation exercises and discuss application to the economics literature. Supplementary materials for this article are available online.
Subfields: Demography, Economics and Econometrics, Safety Research
View Paper: https://www.semanticscholar.org/paper/387a49a51ef3ed0ea5818769e2e413c9790182ee
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Demography
- M. Tabellini
- Ashenafi Biru
- Roni Michaely
- Alexandre Poirier
- Ernest H. O’Boyle
- Adrien Matray
- David Wuepper
- José-Luis Peydró
In Economics and Econometrics
- Jessie Handbury
- Dmitry Taubinsky
- Benjamin B. Lockwood
- Hunt Allcott
- Leah E M Bevis
- Oscar A. Mitnik
- Robert Mahlstedt
- M. Tabellini
In Safety Research
- M. Tabellini
- Ashenafi Biru
- Roni Michaely
- Alexandre Poirier
- A. Hill
- Adrien Matray
- David Wuepper
- José-Luis Peydró
Citations in the corpus
- People Management Skills, Employee Attrition, and Manager Rewards: An Empirical Analysis (2020, positive): “To assess coefficient sensitivity, we use the Oster (2019) test.”
- Executive Extraversion: Career and Firm Outcomes (2018, positive): “We explore the potential severity of omitted variable bias using the framework of Oster (2016), who demonstrates that omitted variable bias is less likely to be a problem when (1) the coefficient of interest is stable after including a wide range of controls, and (2) including controls…”
- The Rise of Shadow Banking: Evidence from Capital Regulation (2018, result): “This 5However, our point estimates are very similar if we exclude bank fixed e↵ects, which indicates that our main result is orthogonal to unobserved lender characteristics (Altonji, Elder, and Taber 2005; Oster 2019).”
- Inference on breakdown frontiers (2017, background): “There is a large literature on this, including important work by Rosenbaum and Rubin (1983), Robins, Rotnitzky, and Scharfstein (2000), and Rosenbaum (1995, 2002).”
- Regressive Sin Taxes, with an Application to the Optimal Soda Tax* (2019, background): “This limited coefficient movement in columns (3) and (4) is consistent with the idea that unobservables also have limited impacts on τ̂ , although this is certainly not dispositive due to the low R2 values (Oster 2017).”
- Do Index Funds Monitor? (2021, positive): “Across all specifications, our estimates remain highly stable, suggesting our results are not sensitive to influence by omitted confounders (Oster, 2019).”