Replication file "Endogenous sunk costs and market structure in Vocational Higher Education"

Replication package for "Endogenous Sunk Costs and Market Structure in Vocational Higher Education" Paola Bordón (Universidad de Chile) and Bernardo Lara (Universidad Andrés Bello) This package contains the Stata code and data needed to reproduce every table and figure in "Endogenous Sunk Costs and Market Structure in Vocational Higher Education." The paper asks why Chile's vocational higher education (VHE) sector came to be dominated by four institutions (INACAP, DUOC UC, AIEP and Santo Tomás) instead of splitting into many smaller ones. This happened during a large, policy-driven rise in demand between 2005 and 2018. Contents in zip files code/: 17 Stata do-files. There is one master script (00_run_all.do), four data-build scripts (CNED INDICES panel, accreditation, student mobility, advertising) and 12 analysis scripts. data/processed/: analysis-ready datasets, including the CNED INDICES program-institution-campus-year panel (2005–2021), CNA accreditation status, advertising expenditure (2011–2018), student mobility aggregates, and projected demand by province. data/public_raw/: public source files. These are the raw CNED INDICES export and workbook, the mifuturo.cl employability and income workbooks (2012–2022 vintages), the AEQualis/IADB 2011 survey, and the authors' compilation of advertising expenses from filed financial statements. output/: reference copies of every figure and table used in the paper, so replicators can compare them with freshly generated files. Data sources: CNED INDICES, CNA accreditation records, MINEDUC enrollment microdata, mifuturo.cl (SIES/MINEDUC), the AEQualis survey (IADB, 2011), and financial statements filed with the regulator. The individual-level MINEDUC enrollment files are public but not included because of their size. Their de-identified aggregates are included, so every exhibit can be reproduced without them. The README gives download links for anyone who wants to rebuild from the raw files. Requirements: Stata 17 or later, plus outreg2, estout and did_imputation (installed automatically). A full run takes under an hour on a standard laptop. The package is about 650 MB. Usage: Set the global $root in code/00_run_all.do and run it. README.pdf maps each exhibit in the paper to its script and output file. License: MIT for the code and CC-BY 4.0 for the data included here.