|
on Experimental Economics |
| By: | John Conlon; Peter Schwardmann |
| Abstract: | We study two hypotheses for how a human role in production could persist despite growing AI capabilities. Humans might have a durable advantage in taste, designing better products even as AI agents automate routine tasks, or consumers might intrinsically prefer that humans have a hand in production. In our experiment, participants act as producers, designing computer games with an AI coding agent, and as consumers who buy them. Across three studies (N = 1, 025), we find no evidence for either hypothesis. Games designed by an AI on its own outperform those of human producers, including both amateurs and professional creatives. Human producers do not anticipate the AI's advantage. Revealing whether a game is human-designed does not raise consumers' willingness to pay for the product. |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2610.18921 |
| By: | Michael Haylock; Patrick Kampkötter; Mario Macis; Susanne Seitz; Robert Slonim; Daniel Wiesen |
| Abstract: | Unavailability of stem-cell donors at the confirmatory typing (CT) stage, an operational milestone in the stem-cell donation process, reduces the effective supply of life-saving transplants for blood cancer patients worldwide. Although (potential) donors register voluntarily, requests may come much later, when circumstances and commitment have changed; keeping registered donors engaged and identifying who will follow through are therefore central managerial questions for registries. In a field experiment with a leading stem-cell registry, we randomize the framing of routine invitations to an engagement initiative. Relative to a business-as-usual letter, our experimental letters present the registry’s requests more saliently; two of them add identity framings inviting donors to join a “quickly available” or a “highly committed” donor team. The salience and identity framings significantly reduce initiative participation compared to the business-as-usual rate, yet intention-to-treat analyses show no detectable effect on average CT availability. Our central finding is that participation sorts donors by reliability: participants are unavailable in 12% of cases, versus 30% of non-participants. From a managerial perspective, opting in works as a reliability flag. Prioritizing participants in our experimental arms implies a relative increase of up to 7.5% more one-year survivors compared to not using this flag. Gains from flagging are likely to be larger for registries with lower availability. |
| JEL: | C93 D64 D82 D91 I12 |
| Date: | 2026–11 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:35998 |
| By: | Furlani, Irene (University of Passau); Grimm, Michael (University of Passau); Raya, Alia (Universitas Gadjah Mada); Ishadiyanto, Salim (Universitas Gadjah Mada); Widjanarko, Nadila (Universitas Gadjah Mada) |
| Abstract: | We conducted a cluster randomized controlled trial to evaluate a platform-based training intervention designed to integrate digital technologies into agricultural extension services and aimed to reduce information delivery costs while improving the efficiency of extension systems. The intervention targeted both extension workers and farmers and combined a digital literacy module with modules on digital agribusiness practices. We compared fully online delivery with a blended format combining online and in-person instruction, examining effects on platform adoption, digital literacy, and the use of digital agribusiness practices, as well as the cost-effectiveness of each modality. Both delivery modalities increased knowledge and adoption of the digital platform among extension workers and farmers and reduced digital literacy constraints among farmers, suggesting that digital platforms can facilitate information diffusion in agricultural extension systems. The intervention also increased the integration of digital agribusiness practices by extension workers, with evidence of adoption of some practices among farmers. Both interventions were relatively inexpensive compared to conventional extension training programs. |
| Keywords: | digital platforms, digital innovations, agricultural extension, millennial farmers, digital literacy, technology adoption, sustainable farming, Indonesia |
| JEL: | C93 O12 O13 O33 Q12 Q16 |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:iza:izadps:dp19041 |
| By: | Bruno Ferman; Lucas Finamor |
| Abstract: | Randomized experiments are often linked to administrative data to measure outcomes, but coarse identifiers can leave multiple administrative records consistent with each experimental unit. We show that unbiased estimation of the sample average treatment effect is possible without resolving these ambiguous links or specifying an outcome model. Our key linkage requirement is that the true administrative outcome is contained among the candidate outcomes for each experimental unit. We characterize the precision cost of linkage ambiguity and study two complementary ways to mitigate it: regression adjustment and reweighting based on observed linkage characteristics. We establish finite-sample and asymptotic properties of these estimators and develop valid randomization-based and large-sample inference. |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2610.15063 |
| By: | Yermone Sargsyan; Salim Turdaliev; Silvester van Koten |
| Abstract: | Using a long-term randomized field experiment in Yerevan, Armenia, we study heterogeneous household responses to behavioural energy reports in a post-Soviet transition-economy setting. Households were assigned to a control group, a consumption-comparison report arm, or a consumption-plus-cost report arm that added monetary bill information to the same peer comparisons. Reports were sent monthly during two report-delivery periods separated by a prolonged pause. Average treatment effects mask substantial heterogeneity by socioeconomic status (SES), female education, baseline electricity use, and electric heating. High-SES households respond differently to the added monetary information than low-SES households, while households with a tertiary-educated female respondent exhibit more conservationoriented responses. Both patterns are clearest during report-delivery months under the consumption-plus-cost reports. Households using electricity for heating show strong conservation-oriented responses, whereas low-baseline users can exhibit boomerang-type increases under consumption-plus-cost reporting. Responses during pause months also differ across groups, implying that intermittent reporting is not uniformly beneficial but depends on subgroup targeting and report content. |
| Keywords: | demand-side management, intermittent nudges, peer comparison, socioeconomic status, principal component analysis, heterogeneous treatment effects, post-Soviet economy, developing country, Armenia |
| JEL: | Q41 Q48 Q58 C93 D91 |
| Date: | 2026–09 |
| URL: | https://d.repec.org/n?u=RePEc:cer:papers:wp823 |
| By: | Hammar, Olle; Neubauer, Florian; Rose, Julian |
| Abstract: | Herskowitz (2021) investigates the relationship between unmet liquidity needs and the demand for gambling among sports bettors in Kampala, Uganda, using a combination of surveys, randomized field experiments, and lab-in-the-field experiments. The study finds that unmet liquidity needs drive gambling behavior, and financial interventions like providing savings mechanisms can reduce the demand for gambling. In this comment, we first successfully conduct a computational reproduction. Second, we conduct a number of robustness checks to check the robustness of the five main outcomes. The findings that anticipated expenditures increase targeted gambling payouts and that a randomized savings treatment reduces betting demand are robust to all analytical choices considered, although the latter remains statistically significant only at the 10% level used in the original paper. The finding that gambling winnings disproportionately increase lumpy expenditures is more sensitive: removing control variables renders the effect statistically insignificant, while alternative outlier treatment does not materially affect it. The two findings concerning the mechanisms underlying gambling demand are largely robust across specifications, although correcting two coding discrepancies reduces their statistical significance. Overall, the original results were found to be largely robust, supporting the paper's main conclusions. |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:i4rdps:339 |
| By: | Pietrzyk, Irena; Schneider, Juliana; Erdmann, Melinda; Helbig, Marcel; Jacob, Marita |
| Abstract: | In der Schule angebotene intensive Beratungsprogramme erhöhen nachweislich die Studienaufnahme benachteiligter Schüler:innen. Doch ist bislang unklar, ob sich diese Effekte auch langfristig halten und wie sie sich auf herkunftsbedingte Unterschiede in den Bildungswegen auswirken. Auf Basis einer randomisierten kontrollierten Studie haben wir die nachschulischen Bildungswege von Schüler:innen 4, 5 Jahre nach dem Abitur untersucht und heterogene Treatment-Effekte für Schüler:innen niedriger und hoher sozialer Herkunft geschätzt. Bei benachteiligten Schüler:innen erhöhte die Beratung die Wahrscheinlichkeit eines Hochschulwegs um 7, 5 Prozentpunkte. Bei privilegierten Schüler:innen verschob sie die Bildungswege in Richtung berufliche Ausbildung, auch wenn dieser Effekt über verschiedene Modellspezifikationen hinweg weniger stabil ausfiel. Diese gegenläufigen Effekte führten dazu, dass sich die herkunftsbedingten Unterschiede in den nachschulischen Bildungswegen deutlich verringerten. Die Studie liefert damit langfristige experimentelle Evidenz, die über die Betrachtung der Studienaufnahme hinausgeht. |
| Abstract: | High school counseling has been shown to increase higher education enrollment among disadvantaged students. However, it remains unclear whether these effects persist in the long run and how they shape origin-based differences in educational pathways. Drawing on a randomized controlled trial, we examine students' post-secondary educational pathways 4.5 years after high school graduation and estimate heterogeneous treatment effects for students from low and high social origins. Among disadvantaged students, counseling promoted higher education outcomes by 7.5 percentage points. Among privileged students, it shifted educational pathways toward vocational education and training (VET), although this effect was less stable across model specifications. These opposing effects led to a substantial reduction in origin-based differences in post-secondary educational pathways. The study thus provides long-term experimental evidence that moves beyond initial enrollment effects. |
| Keywords: | High school counseling, Educational inequality, Long-term effects, Social origin, Experiment, Beratung, Bildungsungleichheit, Langzeiteffekte, soziale Herkunft, Experiment, randomisiert kontrollierte Studie |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:wzbslm:344523 |
| By: | Robin Na; Kehang Zhu; Logan Cross; Michiel A. Bakker |
| Abstract: | Large language models (LLMs) are increasingly used to simulate human behavior, but which descriptions of people improve prediction when experimental designs change remains unclear. We develop a framework that draws candidate behavioral features from the literature and constructs profiles from earlier human experiments. An LLM-based profile generator assigns feature values from observed actions, messages, and surveys; a separate LLM simulates behavior using these profiles. By comparing simulated and observed outcomes in the training data, we select profile contents through forward feature selection, adding features while they improve prediction. We then evaluate the selected contents without modification against outcomes from new participants under new designs. Across 616 public-goods games, 144 multi-party bargaining games, and 1, 480 generalized 11--20 money-request designs, leading one-feature profiles predict human outcome distributions better than no profile or general-purpose personas. Profiles with more features provide no clear additional benefit, and features' relative predictive value largely persists across learning and validation. Using profiles from one game family to simulate another produces uneven gains and can make predictions less accurate than simulations without profiles. The framework makes the choice of behavioral information a testable part of simulation design, helping researchers assess which descriptions remain useful as experimental conditions change. |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2611.00661 |
| By: | Anders Jensen; Nikhil Kumar; Konstantin Poensgen |
| Abstract: | A property tax experiment in Ghana randomly provides tax education to property owners and frontline officials. The intervention raises tax knowledge, which spreads through social networks to untreated neighbors, but does not increase tax payments. Educated taxpayers and officials become more engaged in their interactions without developing more collaborative relationships. Taxpayers revise downward their beliefs about enforcement capacity and become more aware of legal ways to reduce their tax burden. Knowledge is a double-edged sword: it can stimulate citizens to engage with the tax system while exposing the state's limited capacity to enforce it. |
| JEL: | D83 H26 H31 O17 |
| Date: | 2026–11 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:36002 |
| By: | Sofia Amaral; Girija Borker; Nishith Prakash; María Micaela Sviatschi |
| Abstract: | Frontline state agents exercise substantial discretion over citizens’ access to public services, yet little is known about whether the attitudes and beliefs shaping this discretion can be changed among incumbent bureaucrats. Partnering with the Bihar Police - one of India’s largest states that operates under comparatively severe policing capacity constraints, we conduct a cluster randomized controlled trial across 417 police stations and roughly 2, 500 active duty senior field police officers to evaluate a three-day expressive arts intervention, grounded in the Theatre of the Oppressed, designed to change police responses to gender-based violence (GBV). Approximately 7 months after baseline, treated officers exhibit more favorable attitudes toward GBV, greater empathy, and stronger legal knowledge. These changes translate into observed policing behavior: in standardized victim audits conducted one year later, officers in treated stations are 15 percentage points less likely to blame victims, dismiss fewer complaints, and provide more referrals, while administrative records show a 41 percent reduction in registration delays and, over a longer horizon, increased GBV reporting by citizens. Exploiting variation generated by officer transfers, we find suggestive evidence that these effects persist when treated officers relocate and diffuse to untreated officers working alongside treated colleagues. The effects also extend to the workplace: junior female officers working with treated senior officers report lower harassment and anxiety. These findings show that the attitudes shaping bureaucratic discretion are malleable and that changing how incumbent state agents exercise authority can improve both citizen-facing service delivery and the broader workplace environment. |
| JEL: | O10 P0 |
| Date: | 2026–11 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:35986 |
| By: | Evie Thompson; Guy Michaels |
| Abstract: | US fertility has fallen well below replacement, even as stated fertility ideals have not. We ask how much money and time women are willing to trade off for additional children, a question that is difficult to answer using observational data or even natural experiments. We field a discrete choice experiment in which 1, 995 women choose between hypothetical futures varying the number of additional children, household income, and time use. Demand falls with both income and time prices, and rises with income faster than proportionally, consistent with larger families being a luxury good. Exploratory results show that full-time work, but not part-time work, is perceived as costly, and that baseline demand varies considerably by age, religiosity, and desired fertility even as price responses vary less. Notably, women on average require compensation, often significant, to accept a second additional child, including many who say they want two more. This finding complicates the standard reading of a stated ideal of two children as evidence of unmet demand: once its price is held fixed, we find little demand to reach it. |
| Keywords: | fertility, gender, discrete choice experiment, time use, willingness to pay |
| JEL: | J13 J16 J22 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:ces:ceswps:_13124 |
| By: | Juan P. et al. Aparicio |
| Abstract: | Generative AI is improving rapidly, but evidence on whether these advances translate into broad productivity gains across skill levels remains limited. We studied access to 2025–2026 generative AI among 276 researchers, ranging from undergraduates to senior academics, during a seven-hour research assignment spanning economics, political science, psychology, and management. Participants reproduced published results, searched for errors, proposed additional analyses, and wrote referee reports. AI assistance increased successful reproduction, the discovery of verified errors, the production of expert-approved robustness checks, and the quality of referee reports. The gains span computational and written work. The results contrast with our 2024 experiment, in which AI-assisted teams performed similarly to human-only teams on most outcomes. The studies differed in participants, team structure, and procedures. Generative AI assistance now appears capable of raising productivity across much of the expertise ladder and across multiple dimensions of scientific work. |
| Keywords: | AI, expertise, productivity, reproducibility, experiment |
| JEL: | O33 J24 C93 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:ces:ceswps:_13138 |
| By: | Raj Chetty; Jamie Fogel; Lawrence F. Katz; Kadeem Noray; Sonya Porter; James Reisinger |
| Abstract: | Modern program evaluation prioritizes evidence from randomized controlled trials (RCTs) because prior research found that observational estimates of program impacts often diverge substantially from experimental benchmarks. We revisit this finding --- which was based on small survey datasets --- in the era of large-scale administrative data. Using anonymized income tax records covering all American adults between 1999 and 2025, we estimate the effects of workforce training programs on earnings by using machine learning models to predict counterfactuals from rich data on prior work histories, demographics, and residential locations. Rather than testing whether our observational estimates are unbiased, we use 37 RCTs to estimate the distribution of their bias. We find that our machine learning estimates closely match the RCT estimates, with root mean squared error (RMSE) comparable to that of the RCTs themselves. Our method reduces RMSE by 60% relative to parametric estimators used in prior work because our larger samples and richer covariates permit more flexible, higher-dimensional models of the counterfactual. Applying our method to a broader set of sites and enrollment years beyond the experimental sample, we find that treatment effects on earnings vary substantially across locations and over time, even within a given program. Programs that place more participants into target-sector jobs generate the largest earnings gains, with impacts that persist even after participants leave their initial jobs. These findings provide guidance for improving job training programs and, more broadly, illustrate a scalable approach to program evaluation that uses observational data to construct experimentally validated ``digital twins.'' |
| Date: | 2026–11 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2611.03938 |
| By: | Jiayue Zhang; Yunyu Shu; Wycliffe Oluoch |
| Abstract: | Financial constraints often limit clean technology adoption in low- and middle-income countries; relaxing these constraints through credit contract design not only increases adoption but also shapes who adopts and how existing and newly adopted technologies are used. We randomize credit contract subsidies and elicit discrete choices over contracts to study firm demand for small-scale solar. Two subsidy designs carry the same amount but differ in allocation across contract terms: one reduces the down payment, while the other can flexibly lower the down payment or near-term repayments. |
| Keywords: | Financial constraints, Supply and demand, Credit, Choice, Green Technology |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:unu:wpaper:wp-2026-110 |
| By: | Nicola Gennaioli; John List; Matthias Rodemeier; Gregory Sun |
| Abstract: | Neoclassical demand obeys regularity: adding a good can only draw demand away from existing goods. In a natural field experiment with Lyft covering 7.2 million riders and 47 million choices, we randomize the presence and price of a premium ride. Contrary to regularity, the premium ride raises demand for the cheapest incumbent and for not buying. The effect grows with the premium price and with repeated exposure, and persists in sessions where the premium ride is not displayed. Rational inference and inattention cannot account for these facts. A model in which a surprisingly high price, cued on the app and retrieved from memory, draws attention to prices explains them. |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:feb:natura:00842 |
| By: | Cockx, Bart (Ghent University); Arni, Patrick (University of Bristol); Launov, Andrey (University of Kent) |
| Abstract: | How do unemployed workers form beliefs about their labor-market prospects, and how do these beliefs affect behavior, policy effectiveness, and welfare? We develop and estimate a structural job-search model with quasi-Bayesian updating, combining Swiss panel data on beliefs and job-search behavior with a randomized counseling-and-coaching intervention. Job seekers enter unemployment with overly optimistic beliefs about interview arrivals, but gradually become pessimistic as past signals fade from memory. The estimated updating process combines features consistent with both motivated beliefs and recency-weighted information processing. The model replicates the observed dynamics of applications, interviews, wages, and job finding, while decomposing treatment effects into changes in search efficiency, search costs, and beliefs. We find that belief formation is the principal channel through which the intervention accelerates transitions into employment. However, despite increasing job finding, the intervention reduces participant welfare because it operates primarily through lower reservation wages. |
| Keywords: | job search, biased beliefs, expectations, quasi-Bayesian learning, randomized control trial, structural econometric estimation, evaluation, active labor market policies |
| JEL: | C41 C51 C93 D04 D83 J64 J68 |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:iza:izadps:dp19047 |
| By: | Arni, Patrick; Cockx, Bart (RS: GSBE other - not theme-related research, ROA / Human capital in the region); Launov, Andrey |
| Abstract: | How do unemployed workers form beliefs about their labor-market prospects, and how do these beliefs affect behavior, policy effectiveness, and welfare? We develop and estimate a structural job-search model with quasi-Bayesian updating, combining Swiss panel data on beliefs and job-search behavior with a randomized counseling-and-coaching intervention. Job seekers enter unemployment with overly optimistic beliefs about interview arrivals, but gradually become pessimistic as past signals fade from memory. The estimated updating process combines features consistent with both motivated beliefs and recency-weighted information processing. The model replicates the observed dynamics of applications, interviews, wages, and job finding, while decomposing treatment effects into changes in search efficiency, search costs, and beliefs. We find that belief formation is the principal channel through which the intervention accelerates transitions into employment. However, despite increasing job finding, the intervention reduces participant welfare because it operates primarily through lower reservation wages. |
| Keywords: | Job search, biased beliefs, expectations, quasi-Bayesian learning, randomized control trial, structural econometric estimation, evaluation, active labor market policies |
| JEL: | C41 C51 C93 D04 D83 J64 J68 |
| Date: | 2026–11–10 |
| URL: | https://d.repec.org/n?u=RePEc:unm:umaror:2026004 |
| By: | Thompson, Evie; Michaels, Guy |
| Abstract: | US fertility has fallen well below replacement, even as stated fertility ideals have not. We ask how much money and time women are willing to trade off for additional children, a question that is difficult to answer using observational data or even natural experiments. We field a discrete choice experiment in which 1, 995 women choose between hypothetical futures varying the number of additional children, household income, and time use. Demand falls with both income and time prices, and rises with income faster than proportionally, consistent with larger families being a luxury good. Exploratory results show that full-time work, but not part-time work, is perceived as costly, and that baseline demand varies considerably by age, religiosity, and desired fertility even as price responses vary less. Notably, women on average require compensation, often significant, to accept a second additional child, including many who say they want two more. This finding complicates the standard reading of a stated ideal of two children as evidence of unmet demand: once its price is held fixed, we find little demand to reach it. |
| Keywords: | ARRAY(0x620706d32378) |
| JEL: | J13 J16 J22 |
| Date: | 2026–11–05 |
| URL: | https://d.repec.org/n?u=RePEc:ehl:lserod:141900 |
| By: | Andrea Orazio Spinello (CNR-IRCrES, Consiglio Nazionale delle Ricerche, Istituto di Ricerca sulla Crescita Economica Sostenibile, via dei Taurini 19, 00185 Roma, Italy); Valentina Carazzolo (CNR-IRCrES, Consiglio Nazionale delle Ricerche, Istituto di Ricerca sulla Crescita Economica Sostenibile, via dei Taurini 19, 00185 Roma, Italy); Ugo Finardi (CNR-IRCrES, Consiglio Nazionale delle Ricerche, Istituto di Ricerca sulla Crescita Economica Sostenibile, Strada delle Cacce 73, 10135 Torino, Italy); Stefania Frongia (Università di Sassari, Dipartimento di Scienze Umanistiche e Sociali, Via Roma 151, 07100 Sassari, Italy); Valentina Ghibellini (Università di Sassari, Dipartimento di Scienze Umanistiche e Sociali, Via Roma 151, 07100 Sassari, Italy); Emanuela Reale (CNR-IRCrES, Consiglio Nazionale delle Ricerche, Istituto di Ricerca sulla Crescita Economica Sostenibile, via dei Taurini 19, 00185 Roma, Italy); Andrea Vargiu (Università di Sassari, Dipartimento di Scienze Umanistiche e Sociali, Via Roma 151, 07100 Sassari, Italy) |
| Abstract: | This working paper outlines the methodological design employed by PLACES (Portraits and Landscapes of Academic Community Engagement Scholarship), a PRIN 2022 project, that investigates science-society interactions through Community Engagement (CE) initiatives in higher education institutions across Italy, France, and the United Kingdom. Drawing on a multiple case study methodology, this project aims to explore how and why universities engage with their communities and examines the conditions that enable or constrain such initiatives and their transformative potential. PLACES integrates macro-, meso-, and micro-levels of analysis across three countries to address the complexity and context-dependency of CE practices, encompassing policy and systemic frameworks, institutional practices, and individual scholars’ experiences. By investigating diverse institutional contexts and drawing from a combination of data sources including policy documents, quantitative indicators, and qualitative interviews, the project employs data triangulation to provide an in-depth investigation of the object of study. The cases selected by this study reflect different academic traditions and administrative models, offering a foundation for cross-national comparisons. Ultimately, by studying situated CE initiatives within broader systemic and institutional contexts, PLACES seeks to provide evidence-based insights for university managers, policymakers, and scholars and contributes to the ongoing debate on the role of universities as active agents of societal change in an evolving higher education landscape. |
| Keywords: | community engagement, universities, higher education, case study, multi-level study |
| URL: | https://d.repec.org/n?u=RePEc:csc:ircrwp:202506 |
| By: | Zuzanna Kowalik; Piotr Lewandowski; Fabrizio Pompei; Holger Stichnoth; Wojciech Szymczak; Laurène Thil |
| Abstract: | We study platform workers' preferences for social protection — in the form of sick pay and paid annual leave — distinguishing between workers who entered platform work out of economic necessity (the "pushed") and those attracted by its flexibility and advantages (the "pulled"). Using a discrete-choice experiment administered to nearly 800 taxi, delivery, and cleaning platform workers in Germany, Italy, and Poland, we find an average willingness to pay at 3.7% of earnings for sick pay coverage and 4.6% for paid annual leave. Preferences, however, differ sharply by motivation: pushed workers are willing to forgo on average 5.3% of earnings for sick pay, and 5.6% for paid annual leave, while pulled workers' valuations are substantially lower (3.1% and 4.2%, respectively). This difference is most pronounced among workers without access to these benefits, suggesting that the absence of protections is particularly consequential for those who entered platform work under greater economic constraint. The findings indicate that the welfare impact of extending social protections to platform workers — as proposed by the EU Platform Work Directive — might depend on the share of necessity-driven workers in the platform workforce. |
| Keywords: | platform work, willingness to pay, discrete choice experiment, social protection, job quality |
| JEL: | J32 J28 J81 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:ulp:sbbeta:2026-38 |
| By: | Liu, Tzu-Ping (University of Taipei); Yang, Tzu-Hsien |
| Abstract: | Political transparency promotes accountability only if citizens are willing to accept the information provided. As artificial intelligence (AI) increasingly mediates the production and evaluation of public information, an important question is whether AI-generated political information remains subject to partisan filtering. We examine this question using an online survey experiment in Taiwan that varies two potential partisan cues surrounding an AI-generated legislative-performance evaluation: the partisan camp of the legislator receiving a favorable evaluation and the partisan alignment of the media outlet associated with the AI system. The results show that acceptance is strongly shaped by congruence with the favorably evaluated legislator. Respondents are more accepting when the legislator belongs to their own partisan camp and substantially less accepting when the legislator belongs to an opposing camp. By contrast, congruence with the media outlet associated with the AI system has no statistically significant effect. Further, prior generative-AI usage is positively associated with acceptance, but this association is significantly weaker when the favorably evaluated legislator is politically congruent with the respondent. These findings demonstrate that AI-mediated transparency does not eliminate partisan filtering and suggest that the partisan implications of AI-generated outputs may matter more than the political alignment of the institutions associated with their production. |
| Date: | 2026–10–28 |
| URL: | https://d.repec.org/n?u=RePEc:osf:socarx:b3nzt_v1 |
| By: | Cristiana Firullo (Cornell University, Ithaca, New York); Camilla Cherubini (Stanford University, Stanford, California) |
| Abstract: | We build a field infrastructure to test how nutritional information should be displayed online. Its core is a browser extension that changes what Walmart.com shows next to each product and records what shoppers look at and buy, in their own accounts and with their own money. We use it to ask whether shoppers who leave nutritional information unread do so because reading it costs effort or because they would rather not know. In a pilot with 110 U.S. households, the extension shows participants nothing, a 0-100 nutritional score, the score with the nutrients and additives behind it, or the same label behind one click. We find that participants leave closed 96 percent of the labels one click away, and that they open least often the labels of products that score poorly and that they already buy. Participants who see the score alone fill their carts like those who see nothing. Participants who also see the nutrients and additives choose products that score 4.5 points higher within the same type, about 7 percent of the average score, and are 9 percentage points less likely to buy a product rated poor, while participants who have the same label behind one click do not. Participants attribute the same value to keeping the extension in every condition. We read these patterns as preliminary, since they come from a small pilot. Researchers and platforms could use this infrastructure to test health policy interventions, such as the design of nutritional labels, efficiently, in real purchases and before adopting them. |
| Keywords: | nutritional labels; information avoidance; rational inattention; online grocery; field experiment |
| JEL: | D83 D91 I12 L81 |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:net:wpaper:2610 |
| By: | Julian Roeckert; Lukas Mogge; Svenja Fluhrer; Kati Kraehnert |
| Abstract: | Using a randomized cash transfer experiment during an extreme winter and linking household panel survey with insurance customer data, we examine whether anticipatory humanitarian assistance crowds out demand for index-based livestock insurance in Mongolia. We find no evidence of charity hazard and can rule out large negative effects on insurance demand. Treatment effects differ by prior insurance experience. While previously uninsured households show no significant response, households with insurance experience are more likely to purchase insurance after the cash transfer. |
| Keywords: | Humanitarian assistance, Altruism, Climate adaptation, Insurance, Randomized controlled trial, Mongolia |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:unu:wpaper:wp-2026-119 |
| By: | Mariana Macedo (Jacobs University = Constructor University [Bremen]); Melanie Oyarzun (Corvinus University of Budapest); Cristian Candia (Universidad del Desarrollo, Northwestern University [Evanston]); César A Hidalgo (MBS - Manchester Business School - University of Manchester [Manchester], TSE-R - Toulouse School of Economics - UT Capitole - Université Toulouse Capitole - Comue de Toulouse - Communauté d'universités et établissements de Toulouse - EHESS - École des hautes études en sciences sociales - CNRS - Centre National de la Recherche Scientifique - INRAE - Institut National de Recherche pour l’Agriculture, l’Alimentation et l’Environnement, Corvinus University of Budapest) |
| Abstract: | Collective memory—shared information about the past that shapes a group's identity—is often examined through constructs such as familiarity or salience, conflating its local and global components. Here, we untangle these components using multilingual online attention traces for over 50, 000 Wikipedia biographies spanning 137 billion pageviews and a large-scale behavioral recognition experiment in which more than 41, 000 participants answered factual questions about famous biographies. Our experiment reveals a robust local memory premium, with people being around 10% points more likely to answer correctly questions about someone from their country and about five points more likely to answer correctly questions about biographies with whom they share a language. Finally, we study the impact of three major life events: Nobel Prizes, Oscars, and deaths in local and global online attention, finding these events produce brief but strong surges in local attention and slower but sustained increases in global attention. These findings add important spatial and temporal nuances to our understanding of collective memory. |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:hal:journl:hal-05852259 |
| By: | Shaun Da Costa;; Koen Decancq; |
| Abstract: | Multidimensional development indices such as the Human Development Index (HDI) embed value judgements about trade-offs between dimensions. Yet these are rarely tested against the valuations of the people they describe. We use a pictogram-assisted discrete choice experiment with 361 respondents in Tanzania’s Dodoma region to elicit trade-offs between income, schooling, and life expectancy. We estimate preferences using the 264 respondents who make compensatory choices and compare their valuations with the equal-weight trade-offs embedded in the HDI. The typical respondent values an additional year of schooling more highly than an additional year of life expectancy, consistent with the HDI’s qualitative ranking of dimensions. However, the implied schooling-life expectancy value ratio is around 1.5, compared with 7.2 under the HDI at Tanzania’s observed achievement levels. Allowing for observed and unobserved preference heterogeneity reveals substantial variation in valuations across respondents, which provides little support for treating any single weighting scheme as representative of this population’s valuations. The results show that the value judgements embedded in multidimensional indices can be tested empirically, and that even in a small, geographically concentrated sample the trade-offs implied by the HDI differ markedly from respondents’ valuations. |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:hdl:wpaper:2604 |
| By: | Sébastien Houde; Corinne Faure; Joachim Schleich |
| Abstract: | We define the concept of the moral cost of carbon (MCC): the internal carbon price individuals implicitly apply to their consumption decisions. We propose an experimental procedure to measure the distribution of the MCC and implement it in two experiments. Five facts emerge. (i) The MCC has a large mass at zero, but it has a long right tail, with a conditional median of about 500 euros per ton. (ii) It is similar on average across consumption domains, but not within individuals. (iii) It differs from the willingness to pay for carbon offsets. (iv) It is malleable with respect to information about peers' valuations. (v) It can be subject to motivation crowding-out when a carbon-framed fee is present in the decision environment. We illustrate the use of the MCC for the design of local climate policies. We consider a planner who must choose between a carbon fee and mandatory disclosure of the footprint of tourist stays. Whether disclosure should complement a carbon fee ultimately depends on whether moral costs should be accounted for in welfare. |
| Keywords: | morality, carbon pricing, carbon disclosure, motivation crowding-out |
| JEL: | H23 D62 D91 Q58 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:ces:ceswps:_13131 |
| By: | Senn, Julien (University of Zurich); Epper, Thomas (CNRS (French National Center for Scientific Research)); Fehr, Ernst (University of Zurich) |
| Abstract: | Workers care not only about pay and career opportunities but also about meaningful work and socially responsible employers. This heterogeneity matters for worker-firm matching, recruitment, and compensation, yet its preference-based origins remain poorly understood. Using a two-wave study representative of the Swiss population (N=1, 031), we study whether social preferences account for life goals, job preferences, and occupational sorting. We structurally estimate two facets of social preferences: altruism in the advantageous and disadvantageous domains. These dimensions are negatively correlated; failing to account for both substantially underestimates the effects of social preferences. Social preferences predict life goals, preferred job attributes, and willingness to work in the tobacco industry. Higher altruism is associated with a lower emphasis on wealth, career status, and pay, but a greater emphasis on corporate social responsibility and meaningful work. Lastly, altruism when ahead positively associates with employment in education, health, and social services. Our paper highlights the importance of social preferences for worker motivation and labor-market sorting. |
| Keywords: | social preferences, meaning-at-work, career choices, sorting |
| JEL: | D01 D91 J24 |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:iza:izadps:dp19037 |
| By: | Qingyuan Chai; Kevin Lang; Liang Zhong |
| Abstract: | We study the statistical credibility of randomized controlled trials (RCTs) in development economics and the optimal allocation of resources to replication. Using 745 principal-hypothesis tests from 190 articles in leading economics journals, we find that actual t-statistics peak well below the theoretical benchmark of 2.8 expected under 80 percent power, though above values associated with p-hacking. This suggests many RCTs are underpowered. Estimating an empirical Bayes mixture model that accounts for selective publication, we find that approximately 30 percent of published rejections are false discoveries, with average directional power at only 40 percent. At conventional significance thresholds, the posterior probability of a false discovery reaches 66 percent. Doubling original sample sizes lowers the average false discovery rate by five percentage points, whereas tightening the significance threshold to 0.005 yields a larger reduction at the cost of total discoveries. Finally, we characterize a funder’s optimal replication policy under a fixed budget and an evidence standard. Under standard regularity conditions, replications concentrate in an intermediate interval of original test statistics, and the optimal replication sample size is single-peaked in the original t-statistic. Illustrative calibrations show that successful replications reduce aggregate false discovery rates to 1.8–7.8 percent, depending on the evidence standard. |
| JEL: | C12 C18 O10 O12 |
| Date: | 2026–11 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:35999 |
| By: | Clist, Paul; Neubauer, Florian; Rose, Julian |
| Abstract: | Gibson et al. (2019) study whether migration to New Zealand changes Tongans' beliefs and preferences, by surveying winners and losers of a visa ballot. They find that around 10 years after the ballot, despite being much richer, migrants' preferences and beliefs are unchanged. We find that across 213 specifications, examining the robustness of five key (null) results, 211 remain insignificant. The results are highly robust to changes in dependent-variable construction and controls. While Gibson et al. explicitly consider statistical power and acknowledge that moderate effects cannot always be ruled out, interpreting the robust null findings ultimately depends on what effect sizes should reasonably be expected or considered substantively meaningful in this setting. |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:i4rdps:346 |
| By: | Maxime Saxena; Fabio Caccioli; R. Maria del Rio-Chanona |
| Abstract: | Decision Models (DMs), a new class of AI models, give fast responses and return a probability distribution over a fixed set of options. These properties make them candidate human surrogates in economic simulations and surveys, a role increasingly filled by Large Language Models (LLMs), which are costly at scale and under-represent the variation in human behaviour. We present a first evaluation of DMs as human surrogates in economics, comparing 10 DMs with 8 LLMs in three economic settings: a household spending survey on consumption, several classic economic games and a labour-market field experiment. Frontier DMs, those that score highest on general decision-making benchmarks, reproduce the variation in human responses more closely than LLMs in all three settings. They capture the modal human response as well as LLMs in the spending survey, but less well in the economic games. Frontier DMs are also roughly ten times faster than the fastest LLM and can be over a hundred times cheaper. Together, these results suggest that DMs capture the heterogeneity of human behaviour better than LLMs and make large-scale economic simulations more feasible. |
| Date: | 2026–11 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2611.05553 |
| By: | Haruo Kakehi; Yukitoshi Matsushita; Taisuke Otsu |
| Abstract: | This paper extends the finite-population asymptotic framework of Abadie et al. (2014, 2020) to the generalized method of moments (GMM) estimator for moment condition models. Motivating examples include instrumental variable regressions and regression adjustments under randomized controlled trials. We establish consistency and asymptotic normality of the GMM estimator and propose conservative variance estimators. We also study GMM weighting, misspecification, and over-identification testing under finite-population asymptotics. Notably, even though the optimally weighted GMM estimator is infeasible under finite-population asymptotics, the regression adjustment estimator in a randomized controlled trial is shown to be asymptotically equivalent to the optimally weighted GMM estimator. A simulation study and real data application illustrate the usefulness of our GMM theory. |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2610.15168 |
| By: | Rema Hanna; Emily Breza; Benjamin A. Olken; Abhijit Banerjee; Arun G. Chandrasekhar; Gideon Moore |
| Abstract: | Most government transfer programs aim to target the poor. Doing so implicitly assumes that the lowest-consumption households have the highest marginal social valuation, i.e., have the highest social welfare gain from an additional dollar. But this need not be the case. In 600 Indonesian villages, we ask over 5, 500 villagers to rank other villagers on how well off they are and how much a modest cash transfer would help them, and compare these rankings to consumption and self-assessed welfare. Targeting consumption misses 77% of households with the highest marginal social valuation. Targeting self-assessed welfare misses 57%. Even targeting those the community ranks worst off in levels, not margins, misses 25%. Divergences between different notions of neediness are statistically predictable. In a randomized controlled trial, proxy means tests best target consumption, self-targeting ordeals capture self-assessed welfare levels, and community targeting comes closest to marginal social valuation. Targeting strategies inescapably embed normative stances. |
| JEL: | I32 O2 |
| Date: | 2026–11 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:35984 |
| By: | Shuyuan Jia; Kangning Wang; Renzhe Xu |
| Abstract: | Classical Bayesian persuasion assumes that an information designer can condition signals on the entire state. We propose persuasion by parts. In this model, the receiver's payoff from acting is $X = \sum_{i=1}^n X_i$ for independent coordinates $X_1, \ldots, X_n$ observed by separate sources. The designer commits in advance to a product experiment, in which each source $i$ can condition its signal only on $X_i$. Given the signals, the receiver computes her posterior mean $y$ of $X$ and acts with probability $g(y)$, where $g\colon \mathbb{R} \to [0, 1]$ is a nondecreasing function. The designer aims to maximize the probability of action. Let $V_{\mathrm{prod}}(g)$ and $V_{\mathrm{joint}}(g)$ be the supremum probabilities of action under persuasion by parts and under joint persuasion, respectively. When the designer knows $g$, we show $$ V_{\mathrm{prod}}(g) \;\ge\; c\, V_{\mathrm{joint}}(g)^2 \qquad\text{for every nondecreasing } g\colon \mathbb{R} \to [0, 1], $$ where $c > 0$ is an absolute constant independent of $n$ and the prior; the exponent $2$ is tight. When $g$ is unknown, we construct a product experiment $\Pi$ from the prior alone, whose probability of action $V_\Pi(g)$ satisfies $$ V_\Pi(g) \;\ge\; c_\varepsilon\, V_{\mathrm{joint}}(g)^{2 + \varepsilon} \qquad\text{for every } \varepsilon > 0 \text{ and every nondecreasing } g\colon \mathbb{R} \to [0, 1], $$ where $c_\varepsilon > 0$ depends only on $\varepsilon$. |
| Date: | 2026–11 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2611.02845 |
| By: | Rogne, Adrian F.; Larsen, Edvard N.; Grødem, Anne Skevik; Kitterød, Ragni Hege |
| Abstract: | Faced with ongoing population ageing, many governments are aiming to prolong the careers of senior workers. In Norway this is done through diverse means, ranging from economic incentives and pension reforms to labour market policies and the support of campaigns promoting senior employment. Despite incentives and pressures to work longer, many seniors face barriers to prolonged employment and transitions in their late careers, which may undermine policy efforts. One such important obstacle is that managers may hold biased and stereotypical views concerning older workers, and discriminate against them. In this study, we assess age bias through vignette experiments conducted among 2, 264 managers in 1, 664 Norwegian firms and organizations. We first successfully replicate an international landmark study in the field (Lössbroek et al., 2021), confirming their finding that the average manager has biases against older job applicants, and that providing stereotype-rejecting information does not reduce this bias. We further assess the role of applicants’ gender, finding a slight advantage for younger women but substantively small gender differences in the impact of age. We also find limited heterogeneity among managers, except for more bias in industries associated with the private sector. Finally, we assess whether having more relevant work experience may compensate for seniority among job applicants. We find that older applicants with more relevant experience are preferred over older applicants with less relevant experience. However, employers on average still prefer younger applicants with less relevant experience over experienced seniors. |
| Date: | 2026–10–19 |
| URL: | https://d.repec.org/n?u=RePEc:osf:socarx:tva49_v1 |
| By: | Letao Wang (L2S - Laboratoire des signaux et systèmes - CentraleSupélec - Université Paris-Saclay - CNRS - Centre National de la Recherche Scientifique, CentraleSupélec, Université Paris-Saclay, CNRS - Centre National de la Recherche Scientifique); Abdel Lisser (L2S - Laboratoire des signaux et systèmes - CentraleSupélec - Université Paris-Saclay - CNRS - Centre National de la Recherche Scientifique, CentraleSupélec, Université Paris-Saclay, CNRS - Centre National de la Recherche Scientifique, Fédération de Mathématiques de CentraleSupélec - CentraleSupélec - Université Paris-Saclay - CNRS - Centre National de la Recherche Scientifique); Sreejith Sreekumar (L2S - Laboratoire des signaux et systèmes - CentraleSupélec - Université Paris-Saclay - CNRS - Centre National de la Recherche Scientifique, CNRS - Centre National de la Recherche Scientifique, CentraleSupélec, Université Paris-Saclay); Zeno Toffano (L2S - Laboratoire des signaux et systèmes - CentraleSupélec - Université Paris-Saclay - CNRS - Centre National de la Recherche Scientifique, CentraleSupélec, Université Paris-Saclay, CNRS - Centre National de la Recherche Scientifique, L2S - Laboratoire des signaux et systèmes - UP11 - Université Paris-Sud - Paris 11 - CentraleSupélec - CNRS - Centre National de la Recherche Scientifique) |
| Abstract: | Partial differential equations (PDEs) play a crucial role in financial mathematics, particularly in portfolio optimization, and solving them using classical numerical or neural network methods has always posed significant challenges. Here, we investigate the potential role of quantum circuits for solving PDEs. We design a parameterized quantum circuit (PQC) for implementing a polynomial based on tensor rank decomposition, reducing the quantum resource complexity from exponential to polynomial when the corresponding tensor rank is moderate. Building on this circuit, we develop a Quantum Physics-Informed Neural Network (QPINN) and a Quantum-inspired PINN, both of which guarantee the existence of an approximation of the PDE solution, and this approximation can be represented as a polynomial that incorporates tensor rank decomposition. Numerical experiments are conducted on the Hamilton--Jacobi--Bellman (HJB) PDE arising from the Merton portfolio optimization problem, which determines the optimal investment fraction between a risky and a risk-free asset. The results show that our quantum models achieve lower losses and approximation errors than a classical fully connected PINN while using substantially fewer trainable parameters. Our quantum models further outperform a classical PINN constructed to share a similar inductive bias, providing experimental evidence of quantum-induced improvement in the tested settings and highlighting a resource-efficient pathway toward classical and near-term quantum solvers for PDEs with exploitable solution structure. |
| Keywords: | Partial differential equation, Portfolio optimization, Quantum Physics-Informed Neural Network, Variational quantum algorithm, Quantum machine learning |
| Date: | 2026–11–03 |
| URL: | https://d.repec.org/n?u=RePEc:hal:journl:hal-05407711 |