Segmentation, Tightness, and Credit Reallocation: Evidence from Korea

Description

This repository contains the official replication package for the following manuscript: - Paper Title: "Segmentation, Tightness, and Credit Reallocation: Evidence from Korea" - Author: Hyunjoon Lim - Journal: Economic Modelling ### 1. Overview This package provides the complete code and data structures required to replicate the empirical and structural results reported in the paper. It consists of: 1) Empirical Analysis: Stata code (`lp_panel.do`) implementing the bank-firm panel local projections. 2) Structural Model: MATLAB and Dynare codes (`run_ProfitMax_KR_segSMM_shareStable.m`, `HeteroBank_ProfitMax_KR_shareStable.mod`) implementing the structural SMM/IRF-matching estimation and counterfactual simulations. ### 2. Data Availability & Confidentiality Notice (IMPORTANT) The empirical analysis in this paper utilizes confidential, highly restricted bank-firm loan-level microdata obtained from commercial and regulatory databases in Korea (KODATA, FISIS, and ValueSearch). Due to strict legal and institutional privacy restrictions, the raw confidential analysis dataset cannot be publicly shared or redistributed. To fully comply with the journal's data sharing policy while respecting data confidentiality, this repository provides a complete "synthetic pseudo-data" set within the `data/` folder. This pseudo-data preserves the exact variable names, labels, formats, identifiers, and multi-dimensional panel structures of the original confidential file. [NOTE TO REPLICATORS]: Replicators can execute the Stata script (`lp_panel.do`) seamlessly using this synthetic data to verify that the entire econometric pipeline runs without errors. However, because the data values are synthetically generated to protect privacy, the resulting regression coefficients and standard errors will differ from the empirical estimates reported in the manuscript. For the structural model block, the SMM estimation utilizes the empirical moments reported in the final text. These targets are directly hard-coded within the MATLAB wrapper. Therefore, the structural estimation, parameter tables, baseline IRFs, and counterfactual figures can be replicated exactly and identically without requiring access to the restricted microdata. ### 3. Software & Dependencies - Stata: Version 16.0 or higher (requires user-contributed packages: ftools, reghdfe, winsor2) - MATLAB: R2020a or higher - Dynare: Version 4.6.4 or 5.x ### 4. Replication Instructions Please refer to the enclosed `README.md` file within the root archive for detailed, step-by-step replication instructions, required directory setups, and a comprehensive summary table mapping each script to the respective tables and figures in the manuscript. ### 5. Contact For inquiries regarding institutional pathways to access the original confidential microdata or technical questions about code execution, please contact the author at the email address provided in the manuscript.

Institutions

• Chonnam National University

  Gwangju, Gwangju

Categories

Licence