projects

Core work from my PhD research

A central puzzle in labor market research is understanding why certain sociodemographic groups—older workers, the highly educated, white men—are not only better paid on average but also show much greater internal wage variation. This overrepresentation at the top of the wage distribution hints at deeper structural processes at work.

My research explores how workers sort into firms and occupations in ways that amplify existing advantages. I show that those who already have labor market advantages tend to end up in the highest-paying positions, while the disadvantaged cluster in lower-wage opportunities. This sorting mechanism doesn't just create inequality between groups—it also drives inequality within them. Using administrative data, I quantify how much of wage inequality can be attributed to this sorting process by comparing actual wage distributions to counterfactual scenarios where sorting is artificially eliminated.

I'm also developing an agent-based model to understand how social networks shape labor market sorting. The model examines two key mechanisms: how workers' wage expectations are influenced by their peers, and how job information circulates through networks with varying levels of homophily. By calibrating the model on real-world distributions, I explore under what conditions social networks generate the empirical patterns of sorting and inequality we observe.

A related strand of this work examines wage trajectories over workers' early careers. Conventional wisdom and major stratification theories predict that initial advantages should compound over time—those who start high should pull further ahead. Yet empirical studies consistently find the opposite: workers with higher starting wages appear to experience slower wage growth. I demonstrate that this paradox stems from a widespread methodological issue in how researchers model wage trajectories. When properly accounting for the statistical properties of longitudinal data, the evidence actually supports the theoretical prediction: early advantages do accumulate, consistent with both cumulative advantage and human capital frameworks.

Related papers: One paper [R&R], other two are WP.

To what extent are life outcomes predictable? And what role does luck—unexpected, random events—play in shaping life trajectories? These questions sit at the intersection of sociological theory, which increasingly recognizes the importance of chance in achievement, and empirical research, which finds that many life outcomes are surprisingly hard to forecast.

Our work in this area takes multiple approaches to understanding predictability and randomness in the life course. One strand examines whether there exist deep, learnable patterns in how lives unfold across multiple domains—education, employment, health, family formation, and so on. Drawing inspiration from how foundation models in AI learn general patterns from large datasets, we investigate whether similar models can capture the structured nature of life sequences. Using comprehensive population registry data, we test whether a model trained on entire life trajectories can perform well on specific predictive tasks, such as forecasting fertility decisions.

Another strand develops methods for representing entire social networks at the population scale. Working with the full network of a country's population presents both opportunities and technical challenges. We've worked on techniques for creating meaningful representations of these massive, multi-layered, time-evolving networks that can then be used to improve predictions across a range of demographic and social outcomes.

We're also contributing to a broader theoretical synthesis that brings together several literatures: sociological theory on the role of luck in achievement, empirical work on the unpredictability of life outcomes, and causal studies that estimate the effects of random life events. This review aims to provide a unified framework for understanding how chance shapes life trajectories.

Related papers: van de Rijt, A. Bernardi, F. Foley, W., and Lucas Sage. (Forthcoming). "Luck, predictability and the life course." Annual Review of Sociology.

Pial, T., Macanovic, A., Sage, L., Hassan, E., Hafner, F., Skiena, S., van de Rijt, A., Emery, T. "Fertility Prediction with Foundational Large Language Models Trained on Population Registry Data." Submitted.

Collaborators: Arnout van de Rijt, Steven Skiena, Tom Emery, Tanzir Pial, Ana Macanovic, Fabrizio Bernardi, Dakota Handzlik, William Foley, Flavio Hafner

Addressing climate change requires unprecedented levels of collective action and cooperation. The relationship between inequality and climate action is theoretically and empirically ambiguous. On one hand, there are reasons to believe inequality might have beneficial effects on aggregate emissions. On the other hand, inequality may fundamentally undermine our capacity to cooperate toward shared environmental goals—when people perceive that burdens are distributed unfairly, they become less willing to support climate policies or make personal sacrifices.

Building on these insights, we develop a broader framework that directly tests the competing theoretical predictions about inequality's net effects on climate action. Our work aims to quantify whether inequality's potential benefits for emissions are outweighed by its corrosive effects on social cooperation, clarifying whether addressing inequality represents an obstacle to climate action or a prerequisite for it.

Related papers: Two pre-registered experiments in progress.

Collaborators: Giulia Andrighetto, Eva Vriens

Related papers: Work in progress.

Collaborator: Sylvain Fontaine