About
I am a PhD candidate in Economics at Utrecht University, expecting to complete my degree in Spring 2026. My research focuses on understanding labor markets, with particular attention to how job amenities, firm organization, and technological change shape worker outcomes.
I combine administrative employer-employee data with surveys and modern machine learning and econometric methods to study questions about worker mobility, wage determination, and labor market inequality.
Secondary Fields: Applied Machine Learning, Microeconometrics
Job Market Paper
Why do workers separate from their jobs? I study the role of a wide range of amenities in the decision of workers to separate from their firm. Using administrative data and a survey on amenities, I match similar workers and measure their separation response to differences in amenities. I show that amenities are quantitatively important: when additionally considering firm-level amenity differences, workers are up to 3 times less responsive to wages. I find that separations responses vary depending on the amenity: for example, whether a worker's supervisor listens or stimulates them to learn has a strong effect, while being able to work remotely and fixing one's own schedule is less important. These findings have important implications for market structure: if workers are not as reactive to wages as previously thought, firms could be playing a bigger role when setting wages. When decomposing amenities to understand if the differences come from between firms or within firms, I find most of the variation arise within firms, but there is a considerable firm-level component. Moreover, I find that considering amenities makes measured labor market inequality higher.
Keywords: Job amenities, Separation elasticities, Matched employer-employee data
JEL: J2, J31, J32, J42, M50
Research
Equalizing the Effects of Automation? The Role of Task Overlap for Job Finding
This paper investigates whether task overlap can equalize the effects of automation for unemployed job seekers displaced from routine jobs. Using a language model, we establish a novel job-to-job task similarity measure. Exploiting the resulting job network to define job markets flexibly, we find that only the most similar jobs affect job finding. Since automation-exposed jobs overlap with other highly exposed jobs, task-based reallocation provides little relief for affected job seekers. We show that this is not true for more recent technological shocks, such as Artificial Intelligence, for which exposure is positively associated with job finding rates.
Firm Organization and Worker Outcomes: The Role of Occupational Specialization
Using matched employer-employee data from Portugal, we show that firms differ starkly in their occupational employment composition, even within detailed industries, with some firms employing workers across a broad range of occupations and others being much more specialized. These differences are robustly predictive of wages: a worker employed in an occupationally-specialized firm earns less than that same worker employed in a less specialized firm. The wage penalty for working in an occupationally-specialized firm is observed across occupations and industries of all skill levels, and is distinct from the penalty associated with working in a firm with fewer organizational layers. Specialization helps account for the role of firms in inequality, as specialization is strongly negatively related to estimated AKM firm fixed effects. Around 50-60% of the wage penalty from specialization is explained by differences in firm productivity. Specialized firms, however, also engage in lower rates of rent-sharing conditional on productivity, accounting for up to one-quarter of the difference in wage premia between high- and low-specialization firms. Finally, we show that being employed in a specialized firm is also associated with worse longer-term career outcomes for workers.
Automatically Classifying Job Titles into Occupations Using FastText and Socio-demographic Information
I test the effectiveness of FastText, a word embedding algorithm that considers sub-word structure, to classify millions of unique job title write-ins into a standardized list of Census occupations. This is a novel approach compared to the dictionary approaches, which take every word as an object, that have been used in the literature so far. Furthermore, I study the incorporation of worker characteristics to enhance the classification results. To do so, I develop a deep learning classification algorithm that combines pretrained word embeddings with other numerical variables. I find that in its best configuration FastText achieves a macro F1 score of 0.73, while the deep learning classification algorithm that uses individual characteristics improves this score to 0.85 when classifying into 222 detailed occupations. This points to the potential for adding extra information for text classification in general and specially for short text. The deep learning algorithm presented in this paper is flexible enough to be applied to a wide range of contexts.
Teaching
I have extensive teaching experience across institutions in the Netherlands and Chile, spanning both graduate and undergraduate levels. At Utrecht University, I have served as Teaching Assistant for Econometrics and Labor Economics. At the Tinbergen Institute, I taught graduate courses in Natural Language Processing and Asymptotic Theory.
My teaching spans a broad range of subjects in economics and quantitative methods. At VU Amsterdam and Amsterdam University College, I assisted with Macroeconomics, Monetary Economics, and Econometrics courses. Earlier in my career, at Universidad de Chile, I was the primary lecturer for Advanced Mathematical Methods at the undergraduate level, and assisted with graduate courses in Quantitative Methods and Macroeconomics. I also have experience teaching foundational courses including Calculus, Algebra, and Introduction to Economics at Pontificia Universidad Catolica de Chile.