CAGE Working Papers March 2026
CAGE Working Papers March 2026
Friday 27 Mar 2026CAGE research papers draw on our global academic network of research associates and address topics aligned to our four core themes.
Contact cage.centre@warwick.ac.ukLink opens in a new window for more information on submitting research to our working paper series or to be added to our mailing list.
Authors: Josh Lerner, Junxi Liu, Jacob Moscona, David Y. Yang
Theme: Designing and Building Institutions
Summary: This paper examines the global impact of China鈥檚 rise as a hub for innovation and entrepreneurship. Using international data, it shows that growth in China鈥檚 venture capital sector stimulated entrepreneurship in other emerging markets, particularly in sectors closely aligned with Chinese industries. The effects are driven by local investors and firms resembling Chinese counterparts. This expansion increased innovation, serial entrepreneurship, and broader economic outcomes, suggesting that emerging economies benefit from more context-appropriate technologies and business models.
798 Greenwashing or Pragmatism?Link opens in a new window
Authors: Junxi Liu, Shaoting Pi, Ao Wang
Theme: Designing and Building Institutions
Summary: This paper examines the rise in shareholder support for environmental and social proposals between 2010 and 2020, focusing on changes in proposal content. It documents a shift from 鈥渂ig-ask鈥� proposals demanding operational changes to 鈥渟mall-ask鈥� proposals focused on disclosure, which drives increased support rates. Using a structural model of proposer and voter behavior, the study shows this shift reflects an equilibrium adjustment rather than greenwashing, indicating a more pragmatic approach to advancing environmental and social objectives.
Authors: Sonia Bhalotra, Matthew Ridley
Theme: Gender, Health and Wellbeing
Summary: This paper examines the prevalence and consequences of workplace sexual harassment in the UK and public perceptions of related policies. Using population surveys, it provides unified estimates of harassment prevalence, harms, legal awareness, and policy effectiveness, while also documenting citizens鈥� beliefs about these issues. An information experiment tests how evidence on prevalence, harms, and policy performance shapes support for policy and civil society action. The study also compares policymakers鈥� beliefs with those of the public.
Authors: Victor Lavy, Moses Shayo
Theme: Designing and Building Institutions
Summary: This paper studies how teachers鈥� grading rule violations influence students鈥� ethical behavior. Using administrative data linking teacher-assigned internal scores with externally graded national exam scores, it tracks grading violations and subsequent student cheating. Exploiting within-student variation in teacher exposure, the study finds undergrading increases cheating, consistent with imitation and negative reciprocity. Overgrading has mixed effects: it raises cheating in heterogeneous communities but reduces it in homogeneous ones, highlighting the role of community structure and reciprocity norms.
Authors: Victor Lavy, Moses Shayo
Theme: Designing and Building Institutions
Summary: This paper investigates why parents systematically under-report children鈥檚 socio-emotional difficulties relative to children鈥檚 self-reports using data from Luxembourg, the United Kingdom, and Australia. A theoretical framework and novel survey design disentangle information frictions from reporting-style differences. Evidence from Luxembourg shows that about 70% of the discrepancy stems from information frictions. A randomized information intervention shifts parental beliefs among those with weak priors, highlighting the role of second-order beliefs and the potential for targeted information policies.
794 Social Media vs. Democracy: Evidence from the January 6th InsurrectionLink opens in a new window
Authors: Karsten M眉ller, Carlo Schwarz, Zekai Shen
Theme: Designing and Building Institutions
Summary: This paper examines how social media can be used by political elites to undermine democratic institutions, focusing on the January 6, 2021 Capitol insurrection. Exploiting plausibly exogenous variation in Twitter usage, it shows that greater exposure predicts participation in the attack, support for election fraud claims, and related donations. Trump鈥檚 tweets triggered spikes in online mobilization, while his account suspension reduced toxic public expression but had limited effects on private beliefs or financial support.
Authors: Mahyar Habibi, Dirk Hovy, Carlo Schwarz
Theme: Designing and Building Institutions
Summary: This paper develops and validates a method to measure how content moderation distorts online discourse using text embeddings from computational linguistics. Applying it to 5 million U.S. political tweets, the authors find that removing toxic content substantially alters the semantic composition of discussion, with distortions comparable to randomly eliminating several major topics. The results show distortions stem from removing topic content, not just toxic language. Rephrasing tweets with generative language models reduces toxicity while preserving discourse integrity.