Replication materials for “Inference for the Cluster Jackknife Instrumental Variables Estimator with Many Instruments”

Description

This repository contains the Python replication materials and saved output for the paper “Inference for the Cluster Jackknife Instrumental Variables Estimator with Many Instruments.” The package reproduces the Monte Carlo simulations and the Miami-Dade County empirical application reported in the paper. It includes Python programs implementing the cluster jackknife instrumental variables estimator, the proposed variance estimator, the conventional cluster-robust benchmark, documented random seeds, and CSV files containing the reported results. The individual-level Miami-Dade court-record data are not redistributed in this repository. The data are publicly available from Chyn, Frandsen, and Leslie (2025), “Data and Code for: Examiner and Judge Leniency Designs in Economics: A Practitioner’s Guide,” openICPSR, Version V1: https://doi.org/10.3886/E209883V1. Users should download `clean_Miami_data.dta` from that repository and follow the instructions in the README file. The empirical program reproduces an analysis sample of 91,421 observations. The Monte Carlo results are based on 500 replications for each design point.

Institutions

• Beijing Normal University

  Beijing, Beijing
• University of Alberta

  Alberta, Edmonton

Categories

Funders

• National Natural Science Foundation of China

  Beijing, Beijing

  Grant ID: 12322112

Licence