Estimating Latent-Variable Panel Data Models Using Parameter-Expanded SEM Methods 

Journal of Business & Economic Statistics, 43(2), 324–337, 2024  Paper · Slides · Supplement · Code

Parameter-expanded stochastic EM methods for dynamic panel data models with latent variables, covering factor, discrete-choice, and persistent–transitory quantile models, designed to improve global convergence and reduce computation time from hours to minutes. 

Income Risk Inequality: Evidence From Spanish Administrative Records 

with Manuel Arellano, Stéphane Bonhomme, Micole De Vera, and Laura Hospido 

Quantitative Economics, (13)4: 1747-1801, 2022 Paper 

Using Spanish Social Security records from 2005–2018, we construct individual measures of income risk and document substantial heterogeneity across male workers, with greater risk among younger, lower-income workers and during recessions.

Estimating Choice Models with Piecewise Smooth Objective Functions: Application to Joint Retirement 

with Tatiana Rosa Draft

Under review

We show that a class of choice models with piecewise-smooth objectives admits a latent-variable representation, linking piecewise optimization to incomplete-data econometrics and enabling efficient estimation by stochastic EM. Applied to joint retirement among European couples, the method offers computational advantages and reveals complementarities between spouses' retirement decisions.

Income, Employment and Health Risks of Older Workers 

Draft    

I develop a model in which job-specific mismatch shocks help explain older workers' transitions to lower-wage, shorter-hour bridge jobs before retirement. Estimated using HRS data, mismatch shocks account for about 55% of movers' average wage decline, while flexible job changes and re-entry provide substantial welfare value.

Modeling Noisy Subjective Probabilities with Application to Heterogeneous Income Expectations 

with Manuel Arellano Slide

Presentations: CEMFI Workshop on Applied Microeconometrics and Panel Data 2025, Calabria Workshop on Frontiers in Measurement and Survey Methods 2026