The Earnings Cost of the Ethnicity Degree Awarding Gap: Evidence from Five Cohorts of English Graduates: Replication Material

Description

Stata replication code for 'The Earnings Cost of the Ethnicity Degree Awarding Gap: Evidence from Five Cohorts of English Graduates' (Oxford Bulletin of Economics and Statistics). The package contains three do-files implementing the full analysis pipeline — data import and cleaning (00_import.do), sample construction and variable preparation (01_data_prep.do), and the three-stage regression analysis and mediation decomposition (02_analysis.do) — together with a README documenting software requirements, directory structure, variable definitions, and equation-to-code correspondence, and a codebook defining all analysis variables. The underlying data are a bespoke extract of HESA student records linked to Graduate Outcomes survey data obtained via Jisc and are not publicly available; see the data availability statement for access information.

Steps to reproduce

## Data Availability The data used in this paper are not publicly available. They comprise a bespoke extract of Higher Education Statistics Agency (HESA) student records linked to Graduate Outcomes survey data, obtained via Jisc under a Data Sharing Agreement. Access to these data requires: 1. An application to Jisc/HESA for a bespoke data extract 2. ONS Accredited Researcher status (or equivalent) 3. Approval from the relevant data governance bodies Researchers wishing to replicate this study should contact Jisc (https://www.jisc.ac.uk) regarding access to Graduate Outcomes data. General information about HESA data access is available at https://www.hesa.ac.uk/data-and-analysis. --- ## Code The replication package contains three Stata do-files that implement the full analysis pipeline in sequence: ### 00_import.do Imports the raw Graduate Outcomes CSV extract and applies initial cleaning: - Imports raw data from `287114_Item1_Data.csv` - Defines value labels for all numeric-coded variables - Handles missing value codes - Saves cleaned dataset as `grad_outcomes_clean.dta` **Input:** `287114_Item1_Data.csv` (raw HESA/Jisc extract) **Output:** `grad_outcomes_clean.dta` ### 01_data_prep.do Constructs the analytic sample from the cleaned dataset: - Applies sample restrictions (full-time, UK-domiciled, first-degree qualifiers at English HE providers) - Harmonises ethnicity coding across cohorts (F_ETHNIC for 2018/19–2021/22; ETHNIC for 2022/23) - Constructs UCAS tariff band dummies - Constructs work location regional variable from Local/Unitary Authority codes - Constructs degree classification indicator (good degree = First or Upper Second Class) - Constructs earnings variables (log salary, trimmed) - Saves analytic dataset as `go_analysis.dta` **Input:** `grad_outcomes_clean.dta` **Output:** `go_analysis.dta` ### 02_analysis.do Implements the full three-stage analysis: - **Stage 1:** Descriptive distributions of UCAS tariff scores by ethnic group (Figure 1, Table 2) - **Stage 2:** Sequential conditioning regression for degree classification gaps (equations 1–4; Table 3, Figure 2) - **Stage 3:** Sequential conditioning regression for earnings gaps with group-specific returns (equations 5–8; Tables 4–5) - **Decomposition:** Mediation decomposition of the raw ethnic earnings gap (equation 9; Table 6) - **Appendix:** All appendix tables and figures, including cohort stability robustness check (Figure A1) **Input:** `go_analysis.dta` **Output:** Tables (CSV/XLSX), figures (EPS/PNG), saved estimates (STER) --- ## Software Requirements - **Stata** (code written and tested in Stata 17) - **reghdfe** package (Correia, 2017): install via `ssc install reghdfe` - **ftools** package (required by reghdfe): install via `ssc install ftools` - **esttab/estout** package: install via `ssc install estout` To install all required packages, run: ```stata ssc install reghdfe ssc install ftools ssc install estout ```

Institutions

• Manchester Metropolitan University

  England, Manchester

Categories

Licence