Consistently Bounding Parameter Values with One Instrument and Two Endogenous Explanatory Variables: With an Application to the Effect of Fast-Food Availability on Obesity

26 Pages Posted: 30 Oct 2009

See all articles by Richard A. Dunn

Richard A. Dunn

University of Connecticut - Department of Agricultural and Resource Economics

Date Written: October 30, 2009

Abstract

The current paper considers a linear regression framework with two endogenous regressors, but only one instrument that is correlated with both. I demonstrate that under reasonable conditions, these different sources of endogeneity act in opposing directions and hence IV regression can generate economically meaningful bounds. The method is then used to bound the effect of fast-food availability on weight outcomes when the proposed instrument is the number of interstate exits in the county of residence. No economically meaningful relationship is found, suggesting that how other firms respond to the presence of interstates does not affect parameter estimates.

Keywords: endogeneity, instrumental variables, bounding, obesity, fast-food, under-identification

JEL Classification: C31, C12, I10

Suggested Citation

Dunn, Richard A., Consistently Bounding Parameter Values with One Instrument and Two Endogenous Explanatory Variables: With an Application to the Effect of Fast-Food Availability on Obesity (October 30, 2009). Available at SSRN: https://ssrn.com/abstract=1497142 or http://dx.doi.org/10.2139/ssrn.1497142

Richard A. Dunn (Contact Author)

University of Connecticut - Department of Agricultural and Resource Economics ( email )

1376 Storrs Road, Unit 4066
Storrs, CT 06269
United States
860-486-0111 (Phone)

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
88
Abstract Views
634
Rank
520,585
PlumX Metrics