iv-estimation

Fix endogeneity by identifying valid instruments and running two-stage least squares (2SLS) estimation. This skill provides templates for instrument selection, first-stage F-statistics, and interpreting LATE to solve endogeneity problems in econometric models.

After 2SLS estimation, run first-stage F-statistics to detect weak instruments, perform Wu-Hausman tests for endogeneity, and apply Sargan-Hansen or Anderson-Rubin overidentification tests to validate instrument strength and model specifications.

Use propensity score matching (PSM) instead of instrumental variables when selection on observables is credible. PSM estimates the ATT with balance checks, whereas IV is necessary when unobserved confounding causes endogeneity.

Yes, you can perform IV estimation and 2SLS workflows across Python, R, and Stata. The skill provides practical code templates, test interpretations, and reference materials to guide end-to-end instrumental variables analyses in these environments.

Avoid 2SLS when you cannot find valid instruments or when instruments are weak, violating exclusion restrictions. In such cases, LIML alternatives or propensity score matching may provide more reliable causal inference estimates for your analysis.