Lorenz Gschwent

Lorenz Gschwent

PhD student @ University of Duisburg-Essen

Welcome

I am a PhD student in Economics at the University of Duisburg-Essen and a member of the Research Training Group 2484 Regional Disparities and Economic Policy. My main research interests are in Political Economy, Regional and Innovation Economics.

Publications

The Rise of Health Economics: Transforming the Landscape of Economic Research (with Björn Hammarfelt, Martin Karlsson, and Mathias Kifmann). Health Economics

Abstract | Paper This paper explores the evolving role of health economics within economic research and publishing over the past 30 years. Historically, largely a niche field, health economics has become increasingly prominent, with the share of health economics papers in top journals growing significantly. We aim to identify the factors behind this rise. Using a combination of bibliometric methods and natural language processing (NLP), we classify abstracts to define health economics. Adapting NLP methods to evaluate the novelty, impact, and quality of academic papers, we demonstrate that the mainstreaming of health economics is driven by innovative, high-quality research, with two notable waves in quality ratings that highlight the emergence and impact of distinct subfields within the discipline. We find a strong positive correlation between citations and quality ratings, with health economics papers receiving fewer citations for their quality compared to other economics fields. Pandemic-related research received a high number of citations during 2020 and 2021; however, our findings indicate that this work was not systematically more novel or impactful than prior studies within the same subfield.


Current projects

Censorship in Democracy (with Marcel Caesmann, and Matteo Grigoletto). Submitted

Abstract | Working Paper | VoxEU Column Democracies increasingly use censorship to counter foreign propaganda, yet evidence on its consequences remains scarce. We exploit the European Union’s 2022 ban on Russia Today (RT) and Sputnik as a natural experiment, using a triple-difference design that compares users connected to the banned outlets against unconnected users in EU and non-EU countries. We analyze a daily panel of 677,780 tweets from 146,633 Twitter users in seven European countries. Pro-Russia output declines by 21.7% among connected EU users; in a difference-in-differences comparison with non-EU users, total pro-Russia output among EU users falls by 13.6%. Alternative suppliers do not fill the gap: neither their pro-Russia output nor the engagement they receive rises after the ban. Consistent with an agenda-setting role of the banned outlets, the share of EU users’ tweets covering the outlets’ daily top-five topics decreases by 17%. A survey experiment offers suggestive evidence that such censorship can come at a cost to the very norms it is meant to defend.


Quality, Scope, and Leniency: Strategic Application Behavior at the US Patent Office

Abstract Whether an invention receives a US patent depends partly on a lottery: applications are quasi-randomly assigned to examiners whose grant rates differ widely. Using a novel pre-grant, text-based measure of invention quality covering 5.4 million USPTO applications, I estimate that approximately 4.7% of granted patents – roughly one in twenty – would have been rejected under a different examiner draw, four times the best prior estimate of patent office inconsistency. The share of these lottery-decided grants rises steeply with the filed breadth of the application while declining gently in invention quality: the lottery prices breadth, not merit. Marginal grants are therefore quality–scope mismatches – broad rights on inventions whose quality does not stand out. A structural model of patent prosecution as a multi-round negotiation over scope explains why examiner noise concentrates on ambitious applications. The composition of the marginal grants has two policy implications. First, marginal treatment effect curves show that existing LATE estimates in the examiner leniency literature understate average treatment effects of patent granting. Second, structural counterfactuals indicate that examiner leniency dispersion – not the length of prosecution – is the dominant source of allocative inefficiency at the USPTO: equalising examiner leniency reallocates granted breadth away from low-quality inventions, and because breadth carries a social cost but little private value, examiner consistency is welfare-improving under any plausible social cost of scope.