Research

"All models are wrong, but some are useful." George E. P. Box

Publication

  1. S. Lee (forthcoming). A Consistent Variance Estimator for 2SLS When Instruments Identify Different LATEs. Journal of Business & Economic Statistics. [pdf]
    • summary: I provide a correct standard error formula for two-stage least squares estimators under treatment effect heterogeneity.
    • doi:10.1080/07350015.2016.1186555
    • STATA code and instructions
    • Replication of Tables IV, V, and VI of Angrist and Krueger (1991, QJE) using the correct standard error formula for 2SLS [data / MATLAB code (updated on Nov 17, 2015)] 
    • Replication of Table 7 Columns 4-6 of Angrist and Evans (1998, AER) using the correct standard error formula for 2SLS [data / MATLAB code (updated on Nov 17, 2015)]
    • Replication of Table 7 of Thornton (2008, AER) using the correct standard error formula (also cluster-robust) for 2SLS [data / MATLAB code (updated on Apr 15, 2016)]
    • Simulation code based on random subsample of Angrist and Evans (1998) dataset [MATLAB code (updated on Apr 15, 2016)] 
  2. S. Lee (2016). Asymptotic refinements of a misspecification-robust bootstrap for GEL estimators. Journal of Econometrics 192(1), 86-104. [pdf
    • summary: I propose a bootstrap procedure for the generalized empirical likelihood estimators for overidentified moment conditions models. The estimators include the empirical likelihood, the exponential tilting, and the exponentially tilted empirical likelihood. I establish asymptotic refinements of the bootstrap t tests and confidence intervals robust to misspecification of the moment condition model. Under correct specification, the refinement is for the true value. Under misspecification, the refinement is for the pseudo-true value. 
    • doi:10.1016/j.jeconom.2015.11.003
    • A longer version of Appendix (Lemmas and Proofs) [pdf]
    • Calculation of EL, ET, ETEL estimators, S.E's, and asymptotic/bootstrap CI's for simulation and application of Hellerstein and Imbens (1999, ReStat) [instruction (pdf))/MATLAB code]
  3. S. Lee (2014). Asymptotic refinements of a misspecification-robust bootstrap for GMM estimatorsJournal of Econometrics 178(3), 398-413. [pdf]
    • summary: I propose a bootstrap procedure for the generalized method of moments estimators for overidentified moment conditions models. I establish asymptotic refinements of the bootstrap t tests and confidence intervals robust to misspecification of the moment condition model. Under correct specification, the refinement is for the true value. Under misspecification, the refinement is for the pseudo-true value.
    • doi:10.1016/j.jeconom.2013.05.008
    • Supplementary Appendix [pdf]
    • MATLAB code: Coverage probability (Table 3 & Figure 1Table 4 & Figure 2), Size-corrected Power (Figure 3Figure 4)

Working Papers and Work-in-progress

  • Iterated GMM estimator (with Bruce Hansen).
  • A test for instrument validity with variable treatment intensity (with Michele De Nadai).
  • 2SLS with complete subset regressions in the first stage (with Youngki Shin).
  • On Robustness of GEL Estimators to Model Misspecification. [pdf]
  • Bootstrapping GMM Estimators under Local Misspecification. [pdf]

Ph.D. Dissertation

Conference and Invited Seminar Presentations