Working Paper: NBER ID: w20673
Authors: Isaiah Andrews; Matthew Gentzkow; Jesse M. Shapiro
Abstract: We propose a local measure of the relationship between parameter estimates and the moments of the data they depend on. Our measure can be computed at negligible cost even for complex structural models. We argue that reporting this measure can increase the transparency of structural estimates, making it easier for readers to predict the way violations of identifying assumptions would affect the results. When the key assumptions are orthogonality between error terms and excluded instruments, we show that our measure provides a natural extension of the omitted variables bias formula for nonlinear models. We illustrate with applications to published articles in several fields of economics.
Keywords: parameter estimates; sensitivity; structural models; transparency; identifying assumptions
JEL Codes: C1; C52
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| sensitivity measure (C52) | predict asymptotic bias (C51) |
| assumption of households giving $10 (D19) | estimated social pressure biased upward (C92) |
| violations of separability in consumption and leisure (D10) | biases in estimated parameters governing consumption behavior (D11) |
| violations of exclusion restrictions (C24) | significant biases in estimated markups (L11) |
| sensitivity of estimates to various moments (C51) | predicted biases can be quantitatively assessed (C52) |
| local perturbations in the model (C62) | derive corresponding biases (D91) |