Research

Research

Sebastian Buhai’s Research in Economics

Research Profile

My research combines behavioral insight with analytical rigor to understand socio‑economic mechanisms and to inform evidence‑based policy. In this sense it follows David Blackwell’s focus on understanding and Jean Tirole’s ambition to use economics for societal improvement:

“Basically, I’m not interested in doing research and I never have been, I’m interested in understanding, which is quite a different thing. And often to understand something you have to work it out yourself because no one else has done it.” — David Blackwell (1983)

“I was trying to understand society. I liked the rigor of mathematics and physics, and I was deeply interested in the human and social sciences[…] Economics not only documents and analyzes individual and collective behavior, it also aspires to recommend better public policy.” — Jean Tirole (2017)

My primary expertise lies in Microeconometrics and Labor Economics, complemented by a wide-ranging engagement with Industrial Organization, Social & Economic Networks, and Applied Microeconomics more broadly. Thematically, my work studies labor and industrial market dynamics, addressing questions such as human capital formation, worker career trajectories, bargaining and rent-sharing mechanisms; persistent wage and employment inequalities, job amenities, firm productivity, and worker welfare, as well as dynamic disclosure, reputational dynamics in information networks, product release timing, and capital structure. Bridging theoretical frameworks including game theory, real options, search-and-matching models, and mean field games, my methodological toolkit also encompasses empirical approaches grounded in structural econometric modeling, with cross-sectional and longitudinal analyses of large-scale surveys and administrative registers playing a central role.

For a deeper dive, you should explore my research profile in 5 slides or detailed research statement.

Research Output

Original Research Journal Articles (peer-reviewed)

Returns to Tenure or Seniority? (with Miguel Portela, Coen Teulings and Aico van Vuuren), March 2014, Econometrica, 82 (2), pages 705–730, DOI: 10.3982/ECTA8688 . See also the corresponding Web Appendix or the old July 2009 version (includes theory framework).
Synopsis. This paper shows that a worker’s tenure relative to her co‑workers is a distinct determinant of both layoff risk and wages, over and above standard worker and firm heterogeneity. We define a firm‑specific seniority index, prove identification of its effect in flexible duration models of job exit and in wage equations with rich unobserved heterogeneity, and estimate these effects using longitudinal Danish and Portuguese LEED. The results imply that part of the usual “return to tenure” or “return to firm size” is more accurately interpreted as a return to relative position in the firm’s tenure hierarchy.
Keywords: wage dynamics, tenure, seniority, Last-In-First-Out
JEL codes: J31, J41, J63

Tenure Profiles and Efficient Separation in a Stochastic Productivity Model (with Coen Teulings), May 2014, Journal of Business and Economic Statistics, 32:2, pages 245-258, DOI:10.1080/07350015.2013.866568 . See also the corresponding Web Appendix or the accepted, non-gated, November 2013 version. Trivia: The paper has earlier had a 2nd R&R at the Review of Economic Studies (and a change of the editor-in-charge before the 2nd resubmission).
Synopsis. This paper develops a real‑options model of worker–firm matches in which productivity evolves stochastically, hiring requires an irreversible investment, and separations occur efficiently when surplus falls below an endogenous threshold. We identify the structural parameters of this model from joint information on individual job durations and wage histories, and estimate them on US panel data. The estimates show that concave wage–tenure profiles are driven largely by selection on surviving matches, while the causal return to tenure itself is small.
Keywords: random productivity growth, efficient bargaining, job tenure, inverse gaussian, wage-tenure profiles, option theory
JEL codes: C33, C41, J31, J63

How Productive is Workplace Health and Safety? (with Elena Cottini and Niels Westergaard-Nielsen), October 2017, Scandinavian Journal of Economics, 119(4), pages 1086-1104, DOI:10.1111/sjoe.12184. See also the corresponding Online Appendix or the non-gated, December 2015 version. NB. Earlier WP versions of this article were entitled “The Impact of Workplace Conditions on Firm Performance”.
Synopsis. This paper estimates the impact of workplace health and safety conditions on firm productivity. Using Danish LEED merged with representative establishment‑level work‑environment surveys, we estimate augmented Cobb–Douglas production functions that control for simultaneity and unobserved inputs. The results show that improving specific dimensions of the work environment, such as internal climate and monotonous repetitive work, has large and robust positive effects on total factor productivity, comparable in magnitude to the marginal product of capital per worker.
Keywords: occupational health and safety, work environment, firm performance, production function estimation
JEL codes: C33, C36, D24, J28, L23

A Social Network Analysis of Occupational Segregation (with Marco van der Leij), February 2023, Journal of Economic Dynamics and Control, vol. 147, 104593, DOI:10.1016/j.jedc.2022.104593. The accepted, non-gated Dec 2022 version; you can download it also as arXiv preprint (NB. you can also google and easily find several old or very old versions, in, e.g., various discussion paper series: the gist of the model and the main conclusions have never ever changed.).
Synopsis. This paper develops an applied theory framework for occupational segregation when jobs are obtained through referrals in homophilous social networks. We model two groups that differ only in access to “good” and “bad” occupations through their contacts, calibrate the multi‑stage Nash equilibrium to network data, and conduct a social‑welfare analysis. The model shows that partial segregation can arise endogenously and can be utilitarian‑optimal, so that even first‑ and second‑best policies designed to maximize welfare may preserve some segregation rather than eliminate it.
Keywords: social networks, homophily, occupational segregation, labor market inequality, social welfare
JEL codes: J24, J31, J70, Z13


Downloadable (Unpublished) Research Working Papers

Who Counts as Young? Matching Rules for Social Discounting; current stage: working draft. The replication file package is available on the project’s GitHub repository. You can also download the arXiv preprint.
Synopsis. Social discounting often compares the young and the old. In a heterogeneous life-cycle economy, that comparison is not defined by age alone. Young and old agents may differ in assets, earnings, constraints, continuation values, and private marginal values of current resources; changing the matched cells can change the welfare object before any ethical view about age has been imposed. This paper proposes current-cell matching: fix the represented margin, the private unit, the current information that prices the margin, and the common support before reading an age sign as residual social priority. The theory separates raw resource weights from residual weights expressed in private value units and shows that imposed local discount labels are valid precisely when they are induced by a potential on a graph. A calibrated OLG audit then reports which sign claims are certified, and which instead rest on support, units, wedges, or anchors.
Keywords: social discounting, intergenerational welfare, overlapping generations, heterogeneous agents, welfare weights, common support, current-cell matching, graph rationalizability
JEL codes: D61, D63, D15, D52, E21, H43, C63

The Geometry of Heterogeneous Extremes: Optimal Transport and Entropic Design; current stage: working draft. You can also access it as arXiv preprint or as IZA discussion paper.
Synopsis. This paper develops a theory of heterogeneous extremes in economics, built around the idea that top outcomes reflect not only tails but also unequal access to opportunities. In a mixed Poisson search setting, it shows how the distribution of opportunity access shapes normalized maxima, yields order comparisons, and provides a benchmark against the homogeneous economy. Its central contribution is a geometric framework, based on transport ideas, that turns differences in heterogeneity into bounds for the whole law of extremes, the full quantile curve, and disciplined counterfactual comparisons across economies. The paper also separates classical extreme value approximation error from heterogeneity effects, studies an entropy regularized design problem under a mean constraint, and illustrates the framework through a stylized labor market network application to the distribution of top wages.
Keywords: extreme value theory, heterogeneous search, optimal transport, Wasserstein distance,
entropic design, labor market networks
JEL codes: C46, C61, D83, D85, J31, J64

Reputation and Disclosure in Dynamic Networks (with corresponding Supplementary Online Appendix); current stage: revised working draft. You can also access it as arXiv preprint.
Synopsis. Hard evidence often travels through intermediaries whose actions are observed before the evidence itself is revealed. This paper studies institutions with a public docket: a dated sealed record is known to exist, custody and review dates are public, and content is observed only at final disclosure. In such settings, delay is evidence. At a review, retention rules out the states in which the holder would have relayed or disclosed the record; the public posterior is an exact censored Gaussian likelihood, summarized by finitely many endpoints under interval strategies. A quadratic Ornstein–Uhlenbeck benchmark verifies Markov perfect Bayesian equilibria and gives resolution in finite time under reputational discipline. Certified copies add a network force: retention on one route restricts what is feasible on another before it acts, creating planner value that decentralized link formation can miss.
Keywords: hard evidence, dynamic disclosure, informative retention, reputation, network formation, common-record censoring
JEL codes: C72, C73, D62, D82, D83, D85

Wage Dispersion, On-the-Job Search, and Stochastic Match Productivity: A Mean Field Game Approach; current stage: working draft. You can also access it as arXiv preprint or IZA discussion paper;  and download recent presentation slides: 45 min talk (Feb 26) giving mostly the intuition, or 90 min talk (Mar 26) with more detail, or 60 min talk (Apr 26) with some revisions relative to the initial draft.
Synopsis. This paper develops a continuous‑time equilibrium search model, cast as a stationary mean field game, in which match‑specific surplus follows a diffusion, workers choose on‑the‑job search and separation, and firms post state‑contingent wages. The cross‑sectional surplus distribution endogenously generates outside options and the job ladder, separations follow a surplus cutoff, and, under standard monotonicity, the stationary equilibrium is unique. I solve and calibrate the coupled HJB-Kolmogorov system to micro evidence on stochastic match productivity, wages, and mobility, structurally decompose wage dispersion into selection, search, and wage‑policy feedbacks, and run counterfactuals on firing costs, search subsidies, and match‑productivity volatility.
Keywords: wage dispersion, on-the-job search, job ladder, stochastic match productivity, mean field games
JEL codes: C73, D83, J31, J63, J64

Real Option AI: Reversibility, Silence, and the Release Ladder; current stage: working draft. See it also on arXiv; and you can download some relatively recent 30 min presentation slides (Oct 2025) for it.
Synopsis. This paper develops a dynamic disclosure and impulse‑control model in which a firm manages two reset options (a cheap, reversible patch and a costlier, less reversible pivot) together with a public publication‑frequency clock. Short, announced clock‑off windows create local silence that shuts down the martingale part of public beliefs, selecting pure S‑s policies: a two‑rung release ladder with no interior mixing and a clean boundary‑value characterization. Leverage distorts timing only through irreversibility, and the model rationalizes observed telemetry around software and AI releases while providing a measurement blueprint based on firm‑authored disclosure traces.
Keywords: dynamic disclosure, real options, costly reversibility, AI product releases, capital structure & leverage
JEL codes: C61, C73, D83, G32, L86

Firm downsizing, public policy, and the age structure of employment adjustments (with Hans-Martin von Gaudecker); current stage: revised version in preparation
Synopsis. This paper links firm labor demand to early‑retirement institutions using Danish matched employer‑employee data covering the universe of mass layoffs over two decades. We develop a model in which distressed firms facing publicly financed early‑retirement schemes disproportionately shed low‑skilled workers who are eligible for those schemes, and we test its implications with individual exit hazards and firm‑level displacement shares. Exploiting multiple reforms to the early‑retirement system, we show that early‑retirement eligibility systematically shapes the age and skill composition of downsizing, highlighting how social insurance design interacts with firm‑side adjustment.
Keywords: early retirement, labour demand, employment adjustment, mass layoffs, LEED
JEL codes: H32, H55, J26, J65

Selected Work in More or Less Progress (includes also titles for some *temporarily* stagnant projects)

Synopsis. This paper develops a microfounded theory of intrafirm rent sharing in which a firm that invests in worker‑specific know‑how sets wages through sequential bilateral bargaining in a fixed queue. The bargaining protocol generates order‑dependent wage shares, with early bargainers capturing more, so that seniority premia reflect the marginal value of specific human capital; comparative statics hinge on replacement costs and bargaining patience. The model predicts that wage pass‑through from firm shocks rises with seniority, concentrates on incumbents, and weakens when bargaining order is randomized, and matched employer‑employee data with shift‑share instruments, within‑firm rank shocks, and primitives from first‑passage models support these predictions.
Keywords: rent sharing; multilateral bargaining; specific human capital; seniority; wage pass–through; LEED
JEL codes: J31; J41; J63; C78; D21

  • Job Hazard Premia and Worker Risk Preferences; stage: presentation mode

Synopsis. This paper revisits compensating wage differentials for job hazards using Danish longitudinal employer‑employee data merged with a representative worker survey on health, safety, and other disamenities. I estimate models that account for unobserved worker and firm heterogeneity and for heterogeneity in risk preferences, proxied by behaviors such as smoking and family composition. Comparing valuations from hedonic wage regressions with those implied by mixed proportional hazard models of job exit, I show that once sorting and risk attitudes are modeled explicitly, duration‑based valuations of key hazards and amenities are large and broadly consistent with the estimated wage premia.

  • Performance Pay, Wage Dispersion, and Job Separation (with Miguel Portela); stage: presentation mode

Synopsis. This paper contributes to the literatures on incentive pay, inequality, and labor‑market dynamics by estimating the causal effects of performance‑related pay on wage growth, earnings dispersion, and worker‑firm separation. Using Portuguese matched employer‑employee data with detailed information on the variable component of wages, we identify performance‑pay schemes and address endogenous compensation policies and worker selection into performance‑pay firms. The results show how performance pay reshapes wage trajectories, within‑ and between‑firm inequality, and mobility, and a matching model with imperfect monitoring and learning on the job provides a coherent interpretation of the empirical patterns.

Published Books, Monographs, and Research Reviews

Essays on Labour Markets: Worker-Firm Dynamics, Occupational Segregation and Workplace Conditions (digital version also accessible via the EUR online repository), PhD Thesis 2008, Erasmus University Rotterdam/ Tinbergen Institute, published as book by Thela Thesis -Academic Publishing Services, ISBN 978-90-5170-921-6, in the Tinbergen Institute Research Series (no. 431), Amsterdam, The Netherlands; November 2008

Wages, Seniority and Separation Rates in a Stochastic Productivity Model: A Comparative Perspective, MPhil Thesis 2003, Tinbergen Institute, published as monograph by the “Lumen” Scientific Publishing House (Editura “Lumen”), ISBN-10 973-7766-51-2, Iasi, Romania; February 2006

Quantile Regression: Overview and Selected Applications, Ad-Astra Journal (Young Romanian Scientists’ Journal), Vol. 4, 2005

Note on Panel Data Econometrics, Netherlands Network of Economics (NAKE) “Nieuws”, 15 (2), December 2003

Selected unpublished work from my graduate student days (surveys, reports, course papers, etc.)

On Risk in Educational Choice: Brief Overview and Research Note, December 2003

Investigating Reciprocal Motivation in Experimental Labor Markets, June 2003

Incomplete Contracts and the Theory of the Firm, January 2003

Job Search and Contact Networks, April 2002

Check out also my IZA – Institute of Labor Economics research profile. And you could also look at my ORCID, Scholar.Google, CienciaID, ResearcherID, IDEAS.RePEc, etc., all (at least partly) ‘automated’ profiles, even though all those are less complete/ correct (with some being truly egregious in that regard) than this webpage w.r.t. organizing/ depicting my research output. My ORCID profile links also to other relevant research-related work, otherwise accessible via different pages of this website (e.g., recorded public lectures/ talks, various science popularization essays, policy analyses; etc.); as for automatic citations retrieval, Scholar.Google outperforms the others while still underestimating my true citation counts* (* do note that relying only—or mostly—on raw citations, for whatever purpose, is naive at best).

For more details and further information about my past or ongoing research, including placement in my wider research context/ projects/ plans, links to presentation slides etc., please see my detailed research statement. For publications/ dissertations in other scientific fields/ disciplines than Economics, please see the corresponding section of my Curriculum Vitae page; for works targeting a wider (both academic and sometimes non-academic) audience, please see my Essays page and/or my Media-Coverage page; you can also consult the list of courses followed during my graduate programs at the Tinbergen Institute, inter alia with (no longer updated) links to their instructors’ webpages (for external, specialized summer schools/ workshops, see the corresponding section in my extended CV in PDF). Last but not least, you should also read about my (mostly research-based) teaching activities.

NB. Annoying to occasionally find new studies—including newly published articles—citing very old/ outdated drafts of my research papers, even in cases where there are newer or even *published* versions by now. I do see it a professional obligation to search for a particular work’s latest version before citing it (or, alternatively, to have valid, well argued, reasons to refer to older drafts). Feel free to contact me if in doubt.

Once more, don’t forget to check out my research profile — either in a nutshell or as detailed statement.