"Gender Differences in Economics Seminars," with Pascaline Dupas, Amy Handland, Alicia Sasser Modestino, Muriel Niederle, Haoyu Sheng, and Justin Wolfers

[Draft, May 2025] (AER, conditionally accepted)

Abstract: We assess whether men and women are treated differently when presenting their research in economics seminars. We collected data on every interaction between presenters and audience members across thousands of seminars, job market talks and conference presentations, leveraging both human judgment and audio processing algorithms to measure the number, tone and type of interruptions.  Within a seminar series, women are interrupted more than men, and this finding holds when controlling for characteristics of the presenter and their paper topic. This differential treatment appears to reflect in part greater engagement with female speakers - resulting in larger attendance and seminar engagement, and in part greater hostility toward female speakers - resulting in more negative, mid-sentence, and declarative interruptions.

"Room for happiness? Cultural Patterns in How We Rate and Evaluate," with Koen Decancq

[Draft, October 2025] [Slides]

Abstract: How comparable are self-reported evaluations across countries? We explore this question using a unique data set of over four million online hotel reviews from guests of 109 nationalities who rated identical rooms. Despite evaluating the same objective experiences, we find systematic cross-country differences in how individuals map internal evaluations onto response scales—a process we refer to as the reporting function. We introduce two indices capturing key dimensions of this variation: stringency (how critically people rate) and moderateness (how much they avoid extreme scores). These indices correlate with countries’ socioeconomic, cultural, and religious characteristics, and remain robust across model specifications. Importantly, they also predict patterns in other domains: for example, our stringency and moderateness measures are associated with the level and inequality of self-reported life satisfaction across countries. Our findings highlight that response scale use is shaped by cultural conventions, with implications for cross-national comparisons of subjective data in fields ranging from well-being research to management and consumer behavior.

Presented at London School of Economics, University College London, Paris School of Economics, Pompeu Fabra University, University of Antwerp, the Italian Association of Development Economists Meeting 2023, and the National Institute of Statistics and Economic Studies of Luxembourg (STATEC)

"The Times They Are Changing: How the BLM and Me Too Movements Shifted Attitudes on American TV," with Ruben Durante

[Draft, October 2025] [Slides], 

Abstract: We study how the Black Lives Matter (BLM) and Me Too movements reshaped representation and discourse on U.S. television by linking transcripts, audio, and guest-level demographic profiles for more than 13,000 CNN and Fox News shows with text analysis, image recognition, and audio-based tone classification. We find that BLM triggered a sharp rise in the visibility of Black guests, while Me Too expanded women’s presence, with both movements increasing not only who appeared but also how much space they occupied once on air. The composition of invited guests shifted as well: before the movements, Black and female participants tended to be younger and less experienced, whereas afterward they appeared older, more experienced, and more often drawn from political, academic, and executive roles. On-air treatment also changed, with presenters engaging Black guests in a more positive tone after BLM and addressing women in a more polarized mix of positive and negative tone after Me Too, while text analysis shows that pre-existing gaps in patronizing and condescending language narrowed, especially for Black guests. Together, these results demonstrate that large social movements reshape media gatekeeping by expanding representation, elevating more authoritative voices within underrepresented groups, and recalibrating the tone of high-visibility public discourse.

Presented at Monash-Paris-Warwick-Zurich-CEPR Text-As-Data Workshop.

"Don't Stop Me Now: Gender Attitudes in Academic Seminars Through Machine Learning"

[Draft, December 2023] [Slides] Featured in Financial Times Jun 24 and Financial Times May 25

Abstract: This paper focuses on the interactions between peers in the academic environment, building on a widely cited finding in the gender literature that men interrupt women more often than women do. Utilizing a unique dataset from economic seminar audio recordings, this study investigates gender-based peer interactions and uncovers four key findings: (i) Female speakers are interrupted more frequently, earlier, and differently than males; (ii) the extra interruptions largely stem from female, not male, audience members; (iii) Female presenters receive more questions, particularly from males and more comments from female audience members; (iv) audience members of both genders interrupt female speakers with a more negative tone. These findings, persistent even after controlling for variables such as the presenter's seniority, seminar series, or topic of the presentation, contribute in a novel way to the growing literature on the disparate treatment of women economists.

Presented at European Bank for Reconstruction and Development (EBRD), The National Institute of Economic and Social Research (NIESR), Sapienza University of Rome, Text-as-Data workshop at the University of Liverpool, Ifo Institute, University College London, Royal Economic Society Annual Conference, Pompeu Fabra University, ETH Zurich, Aix Marseille School of Economics, the Latin American Economic Association meeting, 2023 Winter School on Inequality and Social Welfare Theory in Canazei (Italy), University of Antwerp, University of Bordeaux, Italian Association of Development Economists, ORT University and Universidad de la República (Uruguay)

"Living with Reduced Income: An Analysis of Household Financial Vulnerability Under COVID-19",  with Catarina Midoes (Social Indicators Research) [PDF] Featured in VoxEU

Abstract: The COVID-19 crisis has led to substantial reductions in earnings. We propose a new measure of financial vulnerability, computable through survey data, to determine whether households can withstand a certain income shock for a defined period of time. Using data from the ECB Household Finance and Consumption Survey (HFCS) we analyse financial vulnerability in seven EU countries. We find that, out of the 243 million individuals considered, 47 million are vulnerable to a three month long income shock (the average length of the first wave COVID-19 lockdown), i.e., they cannot afford food and housing expenses for three months without privately earned income. Differences across countries are stark. Individuals born outside the EU are especially likely to be vulnerable. Being younger, a single parent, and a woman are also statistically significant risk factors. Through a tax-benefit microsimulation exercise, we look into the COVID-19 employment protection benefits, the largest income support measure in the countries considered. Considering as our sample individuals in households where someone receives a salary, we derive household net income when employees are laid-off and awarded the COVID-19 employment protection benefits enacted. Our findings suggest that the employment protection schemes are extremely effective in reducing the number of vulnerable individuals. The relative importance of rent and mortgage suspensions, (likewise, widespread COVID-19 policies), in alleviating vulnerability, is highly country dependent.

Presented at University of Pisa (online), Society for the Study of Economic Inequality Conference (ECINEQ - online) and University of Antwerp (online)

• "With a Little Help from My Bot? The Impact of LLMs on Research Inequality" 

• “The Maternity Leave Gap in Academia: Evidence from Big Data on Conferences”, with Maddalena Grignani

• "Ledgers and Links: building high-dimensional worker–firm panels from company accounts and public web data", with T. Kemeny, C. Ozgen, M. Nathan, G. Pialli, A. Rosso, and A. Sivropoulos-Valero

• "The Environmental Cost of Academic Conferences" with Pedro Rojo-Salas