Source code and content for pascalmichaillat.org

Portal for the course "Unemployment" at UCSC [ECON 182]

Estimate an AKM-style two-way fixed effects model using Canadian matched employer-employee data.

An interpretable system that models the future of work as an equilibrium under AI-driven forces. Instead of predicting job loss, it decomposes workforce disruption into automation pressure, adaptability, skill transferability, demand, and AI augmentation to explain stability, tension, and transition paths by 2030.

Julia implementation of the task-based production function from the paper Supply, Demand, Institutions, and Firms by Daniel Haanwinckel

Official Technical Stack & Economic Engine for the NUPA Framework. Authored by Brandon Anthony Bedard (Nov 2025). Featuring the 40/40/20 Recursive Reinvestment Model and FASL Protocol

Define labor markets using methods from Natural Language Processing (NLP) to cluster occupations based on their task descriptions.

promotional material for our work on Eckstein-Keane-Wolpin models

Source code for the data analysis and models, including the FewSOC prompting framework for O*NET-SOC classification, used in the paper "Leveraging Large Language Models for Career Mobility Analysis: A Study of Gender, Race, and Job Change Using U.S. Online Resume Profiles."

A Rust library for performing Oaxaca-Blinder decomposition on Polars DataFrames, with support for categorical variables and bootstrapped standard errors.

Regression analysis in Stata to assess the impact of Medicaid expansion on employment outcomes using the IPUMS database.

Interactive map of AI's technical capability to replace human labor across 13 territories, 1800–2041. A research and art project.

The Uncomfortable Coexistence of Job Destruction and Labor shortages

Occupational Technology Skill Shares (OTSS) for Germany, 2012–2023

This is a Calibrated Simulation of Individual Level Data for the Basic National Pension Scheme (Tier 1) in Ghana for the Periods 2015 to 2024.

Interactive dashboard comparing labor market trends between Norway and the United States using SSB and BLS government data.

Replication of Card & Krueger (1994) using a two-period difference-in-differences design in R.

Computational Economics: O*NET 27.3 → AI Automation Exposure Modeling. 3,040 tasks across 80 Management/Entrepreneurship occupations with 12 ML features (AI_Risk_Score, Future_Proof_Score). Labor economics research → production datasets for sklearn/XGBoost. 62% management jobs HIGH AI vulnerability.