How Much Should We Trust Regional-Exposure Designs?

68 Pages Posted: 28 Jul 2023 Last revised: 14 Nov 2023

See all articles by Jeremy Majerovitz

Jeremy Majerovitz

Federal Reserve Bank of St. Louis

Karthik Sastry

Princeton University - Department of Economics

Date Written: November 13, 2023

Abstract

Many studies use panel data to implement a regional-exposure design, interacting aggregate shocks with heterogeneous exposures. We show how unobserved aggregate shocks complicate inference in this setting and induce substantial under-coverage when clustering by region. We suggest two-way clustering, potentially with an autocorrelation correction, and randomization inference as solutions, and develop a feasible optimal instrument to improve efficiency. In an application to estimating regional fiscal multipliers, valid 95% confidence intervals cannot reject near-zero multipliers, although 90% intervals are informative. The feasible optimal instrument doubles power. Our results suggest that the precision promised by regional data may disappear with correct inference.

Keywords: Applied Econometrics, Regional Data, Shift-Share Instruments

JEL Classification: C12, C18, C21, C23, C26, F16, R12

Suggested Citation

Majerovitz, Jeremy and Sastry, Karthik, How Much Should We Trust Regional-Exposure Designs? (November 13, 2023). Available at SSRN: https://ssrn.com/abstract=4433676 or http://dx.doi.org/10.2139/ssrn.4433676

Jeremy Majerovitz

Federal Reserve Bank of St. Louis ( email )

411 Locust St
Saint Louis, MO 63011
United States

Karthik Sastry (Contact Author)

Princeton University - Department of Economics ( email )

Princeton, NJ 08544-1021
United States

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
119
Abstract Views
584
Rank
438,797
PlumX Metrics