Fabio Cerina

Welcome to my home page! I am full professor in Economic Policy (13A2) at the Dipartimento di Scienze Economiche e Aziendali of the University of Cagliari.

I am director of the Master's degree program in Economics, Finance and Data Analysis at the University of Cagliari.

I am also Researcher at CRENoS (Center for North South Economic Research) based in Cagliari and Sassari. 

I am an economist with research interests at the intersection of macroeconomics, labor economics and urban economics. Currently my work focuses on the spatial and occupational dynamics underlying job polarization and its geographic dimension and on the labor market impact of Generative Artificial Intelligence and remote work.

UPDATES

• NEW WORKING PAPER! In "Urban Elevators and the Geography of Labor Market Polarization" (with Elisa Dienesch, Alexander Monge-Naranjo and Alessio Moro), CRENoS WP 26/18, we ask why large cities gain abstract jobs and lose routine jobs faster than small ones. Using French administrative data (1984–2021), we show that the divergence is produced inside cities, not by workers moving between them: large cities move workers from routine to abstract jobs at about twice the small-city rate, and this flow more than accounts for the urban gap in abstract employment. The pattern predates the 1994 onset of French polarization and holds within occupations, where staying in a large city raises upgrading by about 40% for college and non-college workers alike. Unlike in the US, the urban routine decline is absorbed by upgrading rather than by downgrading into low-paid services. Check it out here

• NEW COLUMN: The VoxEU column on our work "The Anatomy of Polarization: Evidence from Worker Flow" is out. Check it out here

• NEW VERSION!  A substantially revised version of "Skill-Biased Remote Work and Incentives" (with Luca G. Deidda and Simone Nobili) is available [here]. Using more than 230 million U.S. job postings, we document the joint post-pandemic rise of remote work and performance pay, especially in high-skill jobs, and rationalize it with a moral-hazard model in which monitoring effort is harder at a distance. The model predicts that a decline in the efficacy of remote monitoring reduces remote work among low-skill workers only, a prediction confirmed by evidence exploiting variation in New York State's electronic-monitoring regulation. The revision sharpens the argument, strengthens the empirical analysis, and adds an online appendix ruling out a purely technological explanation.  

• NEW WORKING PAPER!  In "Generative AI and Posted Labor Demand" (with Simone Nobili and Marco Rosso), CRENoS WP 26/13 we answer two main questions: 1) Did ChatGPT reduce the volume of job postings in exposed occupations? Yes. Across 368 million U.S. job postings from Lightcast (2016–2025), and using an exposure measure that reweights AI's theoretical capabilities by actual usage (built on Anthropic data), posted demand in the most exposed occupations fell by about 8.6% relative to less-exposed occupations in the same local labor market and month. 2) Did ChatGPT hurt job postings in junior positions the most? Not really. Junior postings in exposed occupations did decline more, but the shift away from junior roles had already begun in 2021–22, before mass LLM deployment, and shows no further break at the release of ChatGPT, consistent with organizational frictions (e.g. the rise of remote work) rather than AI substitution. Permanent updated version here