Shift-Share Designs: Theory and Inference*

By Rodrigo Adão, Michael J. Kolesar, Eduardo Morales. Quarterly Journal of Economics, 2018

531 citations recorded by the source corpus.

We study inference in shift-share regression designs, such as when a regional outcome is regressed on a weighted average of sectoral shocks, using regional sector shares as weights. We conduct a placebo exercise in which we estimate the effect of a shift-share regressor constructed with randomly generated sectoral shocks on actual labor market outcomes across U.S. commuting zones. Tests based on commonly used standard errors with 5% nominal significance level reject the null of no effect in up to 55% of the placebo samples. We use a stylized economic model to show that this overrejection problem arises because regression residuals are correlated across regions with similar sectoral shares, independent of their geographic location. We derive novel inference methods that are valid under arbitrary cross-regional correlation in the regression residuals. We show using popular applications of shift-share designs that our methods may lead to substantially wider confidence intervals in practice.

Subfields: Economics and Econometrics

View Paper: https://www.semanticscholar.org/paper/164973ae4897b3e519aa33791ce57c5128756f8e

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Economics and Econometrics

Citations in the corpus

Trade, pollution and mortality in China (2016, unknown): “due to the similarity of industry structure across prefectures that are not geographically proximate (Adão et al. 2018). As a robustness check, we follow Borusyak et al. (2018) and conduct the analogous analysis at the industry level, which overcomes such concerns about statistical inference of shif”
Venting Out: Exports During a Domestic Slump (2018, methodology): “In column 6, the we first report standard errors clustered by province, and below we report standard errors that apply the formula in Adão et al. (2019). All specifications include firmlevel log differences in TFP and log differences in average wages as additional controls (estimates not included to”
The Slope of the Phillips Curve: Evidence from U.S. States (2020, background): “So our standard errors are not subject to the concerns about inference with shift-share instruments raised by Adão, Kolesár, and Morales (2019).”
Quasi-Experimental Shift-Share Research Designs (2018, unknown): “As discussed above we also draw on the inferential framework of Adao et al. (2018), who derive valid standard errors in shift-share designs with a large number of idiosyncratic shocks.”
Benign Effects of Automation: New Evidence from Patent Texts (2018, unknown): “This last point also addresses the overrejection problem of shift-share designs as discussed by Adao et al. (2019).”
The Adjustment of Labor Markets to Robots (2021, methodology): “Robot exposed labor markets tend to have slightly higher income (GDP) per capita (but the standard error is relatively large when using the inference suggested by Adão et al. (2019)).”
Stock Market Wealth and the Real Economy: A Local Labor Market Approach (2019, unknown): “Clustering by time allows for areas with high or low stock market wealth to experience other common shocks and accords with the recommendation of Adão et al. (2018) in the special case of a single national shifter.”
Income Growth and the Distributional Effects of Urban Spatial Sorting (2019, unknown): “See, for example, Ado et al. (2018), Borusyak et al. (2018), and Goldsmith-Pinkham et al.”
Tasks, Automation, and the Rise in US Wage Inequality (2021, methodology): “In particular, we compute standard errors as in Adao, Kolesár, and Morales (2019) and Borusyak, Hull, and Jaravel (2022), which are robust to the presence of unobserved industry shocks that affect all workers or workers in routine jobs in an industry.”