Specialised Human Capital Supply and Sectoral Expansion - Joint with Britta Hecker, Friederike Hertweck and Lukas Jonas (updated: 16 July 2026)
How do local labour markets absorb expansions in specialised skill supply? We construct a novel dataset of computer science programme openings across West German higher education institutions from 1971 to 2022 and link it to administrative data on workers and establishments. Openings sub- stantially increase local employment of computer scientists without reducing wages. Employment gains are concentrated almost entirely in the ICT sector and are accompanied by a near doubling of ICT establishments, while we find little evidence of increased patenting or investment. We develop a model with free firm entry that rationalises this pattern: entry pins wages independently of labour supply, so skill expansions are absorbed through the sectoral composition of firms, an adjustment margin that can operate alongside endogenous technical change, but requires no technological up- grading by incumbents. Targeted expansions of specialised skill supply may therefore function more as a sectoral industrial policy than as a catalyst for broad technological upgrading.
Do Workers Sort to Firms or Occupations? - Joint with Luke Heath Milsom (updated: 19 February 2026)
In this paper, we find that high-wage workers sort mainly to high-wage occupations and not to high-wage firms, and that at least half of the previously documented sorting to firms can be attributed to the segregation of occupations across firms. To reach these conclusions, we leverage the universe of matched employee-employer data from France and Germany and estimate a flexible two-way worker-job fixed effects model of log wages. We then isolate worker sorting to firms by studying the within-occupation across-firm covariance between worker and job fixed effects
Presented at the 9th Warwick Economics PhD Conference (2021), Royal Economics Society Annual Conference (2022), AASLE (2023), EWMES (2023), SOLE (2024), EALE (2024) (by coauthor)
Individual and Aggregate Mismatch in Higher Education (updated: 19 October 2023)
In many economies, many graduates are observed to work in non-graduate occupations. Does this mean that the level of education is inefficiently high? I propose a model that features uncertainty about the causal effect of HE and incorporates general equilibrium wage effects in a matching model of the labour market. My model highlights that there are two distinct notions of overeducation. Workers could be individually mismatched in the sense that they would be better off if they made alternative education choices. There could also be aggregate mismatch if average welfare would be higher if the average level of HE attendance changed. I find that uncertainty about returns generates the former, while endogenous education choice in a matching market generates the latter. Structurally estimating my model on UK data, I calculate that 32.9% of the population individually mismatched. I find in policy simulations that policy-makers can improve aggregate welfare by reducing HE share by 1.6 percentage points. Reducing uncertainty about returns is unambiguously good for workers but increases income inequality and worsens firms’ outcomes by making labour more expensive.
Presented at the 10th Warwick Economics PhD Conference (2022), University of Surrey PhD Workshop in Microeconomics (2022), Young Economist Symposium (2022), Asian & Australasian Society of Labour Economics Conference (2022), European Winter Meeting of the Econometric Society (2022), Society of Labour Economists Conference (2023), Leuven Summer Event (2023), EEA-ESEM Congress (2023), European Association of Labour Economists Conference (2023)
Determinants of University Subject Choice in the UK - Draft available on request (updated: 22 February 2021)
Students applying to British universities have to choose to specialise in a particular subject at the point of application. What factors play into their choice of subject of study? I analyse students' subject choices in university using a cohort panel dataset (LSYPE), and in particular, whether those choices respond to their expected wages conditional on studying the subject. I find that earnings, defined as earnings at age 25 or as lifetime earnings, are positively associated with choice probability, although the quantitative impact of expected earnings on subject choice is small. I also find that there are substantial differences in choice probability between students of different sexes, ethnicities and to a lesser extent, socio-economic classes.
Labours of Love: Search for Occupational Meaning and Worker Welfare. With Hannah Zillessen and Luke Heath Milsom.
How Do Parents Affect the College Education Choice of Their Children?. With Adrian Lerche
A Decision-Theoretic Approach to Debates about University Admission. With James Baraniak