Data and Code for: Valid t-ratio Inference for IV
Principal Investigator(s): View help for Principal Investigator(s) David Lee, Princeton University; Justin McCrary, Columbia University; Marcelo Moreira, FGV EPGE; Jack Porter, University of Wisconsin
Version: View help for Version V1
| Name | File Type | Size | Last Modified |
|---|---|---|---|
| ReplicationClean | 06/22/2022 10:19:PM |
Project Citation:
Lee, David, McCrary, Justin, Moreira, Marcelo, and Porter, Jack. Data and Code for: Valid t-ratio Inference for IV. Nashville, TN: American Economic Association [publisher], 2022. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 2022-09-23. https://doi.org/10.3886/E164502V1
Project Description
Summary:
View help for Summary
Data and code for "Valid t-ratio Inference for IV"
Abstract
In the single-IV model, researchers commonly rely on <i>t</i>-ratio-based inference, even though the literature has quantified its potentially severe large-sample distortions. Building on Stock and Yogo (2005), we introduce the <i>tF</i> critical value function, leading to a standard error adjustment that is a smooth function of the first-stage <i>F</i>-statistic. For one-quarter of specifications in 61 AER papers, corrected standard errors are at least 49 and 136 percent larger than conventional 2SLS standard errors at the 5-percent and 1-percent significance levels, respectively. <i>tF</i> confidence intervals have shorter expected length than those of Anderson and Rubin (1949), whenever both are bounded.
Abstract
In the single-IV model, researchers commonly rely on <i>t</i>-ratio-based inference, even though the literature has quantified its potentially severe large-sample distortions. Building on Stock and Yogo (2005), we introduce the <i>tF</i> critical value function, leading to a standard error adjustment that is a smooth function of the first-stage <i>F</i>-statistic. For one-quarter of specifications in 61 AER papers, corrected standard errors are at least 49 and 136 percent larger than conventional 2SLS standard errors at the 5-percent and 1-percent significance levels, respectively. <i>tF</i> confidence intervals have shorter expected length than those of Anderson and Rubin (1949), whenever both are bounded.
Scope of Project
JEL Classification:
View help for JEL Classification
C10 Econometric and Statistical Methods and Methodology: General
C12 Hypothesis Testing: General
C26 Single Equation Models: Single Variables: Instrumental Variables (IV) Estimation
C46 Specific Distributions; Specific Statistics
C10 Econometric and Statistical Methods and Methodology: General
C12 Hypothesis Testing: General
C26 Single Equation Models: Single Variables: Instrumental Variables (IV) Estimation
C46 Specific Distributions; Specific Statistics
Related Publications
Published Versions
Report a Problem
Found a serious problem with the data, such as disclosure risk or copyrighted content? Let us know.
This material is distributed exactly as it arrived from the data depositor. ICPSR has not checked or processed this material. Users should consult the investigator(s) if further information is desired.