Simulations: Difference Estimation for RD with Multiple Ordered Thresholds

This repository contains a small, reproducible R setup that generates the simulation results of our paper “Difference Estimation for Regression Discontinuity with Multiple Ordered Thresholds” (Moghani & Badri, 2025). The code compares three estimators for the boundary difference $\delta = m(1)-m(0)$: a boundary estimator, an integrated-derivative estimator, and their inverse-variance combination. It also saves CSV outputs and a diagnostic figure.

Theory in brief: the integrated-derivative estimator integrates local slope estimates over the interior; it is asymptotically independent from the boundary estimator and admits a feasible CLT. This permits efficiency gains by inverse-variance combination.

---

What’s inside

• simulate.R — single-file R script that runs the simulations end-to-end (Colab-friendly) using a polynomial DGP.
• simulate2.R — single-file R script that runs the simulations end-to-end (Colab-friendly) using a cosine DGP.
• README.md — this file.
• LICENSE — MIT.

The script is written to work locally or in Google Colab with an R runtime.

---