Maria Cristina Maurizio

The Temporary Impact of Permanent Employment Incentives: Evidence from Italy

M. Cantarella, M.C. Maurizio, F. Serti

Job Market Paper

This paper evaluates the short- and medium-term effects of payroll tax reductions for young workers hired on, or converted to, permanent contracts. Using administrative data from Tuscany covering full employment histories, we exploit the age-based eligibility rule introduced in 2018. The exemption did not draw non-employed individuals into permanent jobs; its effect is concentrated among the employed and runs mainly through conversions of temporary contracts, which rise by about one third in the first year. Employers also bring conversions forward ahead of the worker's loss of eligibility. Over four years the gain in permanent employment persists, but shifts away from the original employer toward other firms, following involuntary separations rather than quits. Worker-based hiring subsidies thus reshape careers across firms rather than expanding employment.

Machine Learning for Profiling in the Italian Public Employment System

M.C. Maurizio, F. Serti

Public Employment Services are increasingly replacing rule‑based profiling of jobseekers with models that predict the risk of long‑term unemployment. In Italy the national approach uses one regression. In this research I build models for Tuscany using administrative data from 2018 to 2021. I compare regression, Lasso, Random Forest and CatBoost looking at both accuracy and how well the predictions match reality. Machine learning gives modest accuracy gains while richer variables matter more, than the choice of algorithm. When I validate the models on years predictions drift as labour market conditions change so the models need periodic recalibration. Shapley values highlight which individual characteristics drive the predictions helping Public Employment Services target training and activation measures.

Beyond the Observed Disability Pay Gap: Accounting for Selection

P. Albarran, M.C. Maurizio

This study explores the disability pay gap in Spain from 2014 to 2022 with data from EU‑SILC taking into account the self‑selection of people into jobs. Because employment rates differ sharply between people with disabilities and people without disabilities estimating the disability pay gap among workers can mislead. The research uses the decomposition framework developed by Pereda‑Fernández in 2025, which expands on the selection model of Arellano and Bonhomme from 2017. This method lets us separate the disability pay gap across the wage distribution into parts due to endowments, coefficients, selection and participation. By comparing corrected and uncorrected numbers the study shows how much of the disability pay gap is caused by selection, into employment.