Structured Music as Human Capital: Thresholds, Complementarities, and Simulation Evidence
Description
This repository contains the complete simulation and replication package accompanying the article “Structured Music as Human Capital: Thresholds, Complementarities, and Simulation Evidence”. It documents how structured cultural participation—specifically the Enescu Festival’s Children’s Concert Series—is modeled as an input in a Becker–Heckman style skill-formation framework featuring: (i) a pedagogical quality threshold, (ii) home and school complementarities, (iii) age-sensitive returns, and (iv) a joint cognitive–affective (“health-like”) channel. Contents: 1. Simulation code implementing the human capital production function, including: • Intensity of exposure (M), pedagogical quality (Q), and an age shifter g(A) • Effective inputs with complements: M_eff = M(1 + φ), Q_eff = Q(1 + ψ) • A quality threshold Q* distinguishing the high-quality scenario (S1) from the below-threshold scenario (S2) • An optional health-like state S_t that scales returns and moderates outcome variance 2. Generated datasets (CSV) used to compute all reported statistics: • Group means and difference-in-differences (DiD) contrasts at baseline, post-test, and follow-up • First-stage, OLS, and IV (LATE) estimates, including Anderson–Rubin confidence sets • Heterogeneity analyses by age, baseline executive function quartiles, and health-like proxy • Cost-effectiveness grids reporting RON per 0.1 SD gain under alternative cost and effect-size assumptions 3. Figure and table outputs: • Diminishing returns to intensity (M) • Quality threshold patterns comparing S1 and S2 • DiD trajectories over time • Robustness checks (age curvature, threshold smoothing, exclusion violations, spillovers, practice effects, attrition) • Joint-investment channel plots linking cognitive and affective components 4. A documentation file linking: • Each script to the corresponding tables and figures in the manuscript • Key parameter choices to the rationale summarized in the appendices All data are simulated and contain no identifiable information. The repository is designed to ensure transparency and reproducibility, and to serve as a modular template for researchers planning field experiments or quasi-experimental evaluations of cultural or educational programs. Users can reproduce every table and figure from the paper or adjust parameters (M, Q, φ, ψ, g(A), S_t) to explore alternative program designs, thresholds, and targeting strategies.
Files
Steps to reproduce
This repository provides all code and configuration needed to replicate the Monte Carlo simulations, main tables, and figures reported in the manuscript on structured music exposure and human capital formation. System requirements Use Python 3.9+ with the following core packages installed: numpy, pandas, matplotlib, (optionally) scipy. The code is designed to run on a standard laptop; full Monte Carlo runs are computationally intensive but feasible. Project structure Place all files in a single working directory, preserving the relative paths: – enescu_all_in_one.py: main simulation, calibration, figures, and tables. – enescu_appendix_only.py: supplementary robustness simulations for Appendix S1–S8. Both scripts automatically create an outputs/ folder with subdirectories for CSV and figure exports. Install dependencies From the project directory, install required packages: pip install numpy pandas matplotlib scipy Run the main simulation Execute enescu_all_in_one.py: python enescu_all_in_one.py This script: (a) calibrates the production-function scale parameter K so that the high-quality scenario (S1) yields an IV LATE and DiD effects in the targeted range; (b) simulates scenarios S1 (high quality with complements) and S2 (below-threshold quality); (c) estimates OLS, IV, DiD, and weak-IV-robust Anderson–Rubin sets; (d) generates main tables (effect estimates, DiD means, cost-effectiveness) as CSV and LaTeX; (e) produces core figures (diminishing returns, quality threshold, time dynamics, cost grid). Run appendix robustness simulations Execute enescu_appendix_only.py: python enescu_appendix_only.py This script reuses the production structure to generate Appendix panels S1–S8: – age-sensitivity of g(A), – hard vs. soft quality threshold under measurement noise, – exclusion restriction violations, – spillovers and randomization units, – first-stage strength under varying compliance, – practice effects and attrition, – heterogeneity by baseline skills and health-like state. Outputs are saved as CSV files and PNG figures in outputs/appendix/ and outputs/figures/appendix/. Interpretation All effect sizes are standardized on the simulated baseline EF distribution and correspond to the estimates reported in the manuscript. Users can modify parameters in the scripts (e.g., quality, costs, compliance, curvature) to explore alternative calibrations; regenerated outputs will remain directly comparable to the published specification.
Institutions
- Universitatea de Vest din Timisoara Facultatea de Economie si de Administrare a Afacerilor