Working Paper: NBER ID: w14434
Authors: Aviv Nevo; Adam M. Rosen
Abstract: Dealing with endogenous regressors is a central challenge of applied research. The standard solution is to use instrumental variables that are assumed to be uncorrelated with unobservables. We instead assume (i) the correlation between the instrument and the error term has the same sign as the correlation between the endogenous regressor and the error term, and (ii) that the instrument is less correlated with the error term than is the endogenous regressor. Using these assumptions, we derive analytic bounds for the parameters. We demonstrate the method in two applications.
Keywords: endogeneity; instrumental variables; econometrics; parameter identification
JEL Codes: C30; C31; C33
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Imperfect Instrumental Variable (IIV) correlated with the error term (C36) | Direction of bias in OLS estimator ascertained (C51) |
| Correlation between IIV and error term same as correlation between endogenous variable and error term (C26) | Bounds on parameter of interest (C51) |
| Use of multiple IIVs (C36) | Tightening of identification region for parameters (R15) |
| Assumptions regarding correlations of instruments and endogenous variables with error term (C36) | Partial identification of parameters (C30) |
| Improved bounds on parameter of interest (C51) | Incorporating additional assumptions about correlations (C10) |