|
on Experimental Economics |
| By: | Marina Agranov; Federico Echenique; Kota Saito |
| Abstract: | We investigate whether risk and time preferences differ when individuals make decisions for others compared to making decisions for themselves. We introduce a novel ``skin in the game'' experimental design, where choices for others incur a direct cost to the decision-maker, ensuring a genuine trade-off between self-interest and surrogate allocation. The modal outcome is that participants are more risk-averse and impatient when choosing for others than for themselves. Our methodology reveals significant heterogeneity, successfully identifying selfish types often missed by the more standard ``no skin in the game'' approaches. The message is nuanced, as even non-selfish participants behave differently when they have skin in the game. Furthermore, our framework yields more consistent behavior and superior out-of-sample predictive power. |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2601.14489 |
| By: | Bejarano, Hernan; Busso, Matías; Santos, Juan Francisco |
| Abstract: | We study how individuals in six Latin American countries value public versus private provision of education and healthcare using a survey experiment. Respondents were randomly assigned to vignettes that vary income, service quality, and provider type. Perceived quality is the main driver of choices: the probability of selecting a private provider roughly doubles when public quality falls from 80 to 20 percent, while income has a smaller effect. Higher institutional trust lowers the likelihood of switching to private providers but does not affect willingness to pay once individuals choose private provision. The multi-country design supports external validity and reveals similar behavioral responses across contexts. The results show that improving service quality and rebuilding institutional trust can reduce reliance on private provision. |
| Keywords: | Stated Preferences;willingness to pay;Public versus Private Provision;service quality |
| JEL: | D12 H42 I21 I18 O54 |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:idb:brikps:14477 |
| By: | Brit Grosskopf (Department of Economics, University of Exeter); Yangfei Lin (School of Economics, Zhejiang University) |
| Abstract: | In a controlled laboratory experiment, we study how leader gender affects both the willingness of group members to contribute ideas and the informational quality of those contributions. Participants are randomly assigned to groups of three, consisting of one merit-based leader and two group members. Group members submit answers to a general knowledge task and report their willingness - on a scale from one to five - to have their answer selected as the group response. In the baseline condition, gender information is not revealed; in the treatment condition, the gender of the leader and group members is disclosed. We find that men become significantly more willing to contribute when led by a female leader, while women’s willingness does not depend on leader gender. However, this increase in male participation is accompanied by a decline in the accuracy threshold at which men are willing to step forward. In contrast, women raise their accuracy threshold under female leadership, contributing only when they are highly confident in the correctness of their answers. As a result, conditional on stating the highest willingness level, men are substantially less accurate under female leadership, whereas women are more accurate. We refer to this asymmetric pattern as a sisterhood effect. We find no corresponding brotherhood effect: men do not exhibit higher conditional accuracy under male leadership relative to the no-gender benchmark. On the selection side, leaders strongly weight stated willingness when choosing whose answer represents the group. When gender is revealed, female leaders are more likely to select female group members, whereas male leaders show no systematic gender-based selection. This selection behaviour mitigates the lower quality of highly willing male contributions under female leadership and preserves group performance. Overall, leader gender shapes collective decision-making not by altering underlying ability, but by changing how private knowledge is translated into expressed willingness and how that willingness is filtered into group choices. |
| Keywords: | leader gender, information aggregation, participation decisions, confidence thresholds, group performance |
| JEL: | C91 D83 J16 |
| Date: | 2026–01–08 |
| URL: | https://d.repec.org/n?u=RePEc:exe:wpaper:2601 |
| By: | Barro, Tjantana; Marencak, Michal; Nghiem, Giang |
| Abstract: | We provide causal evidence that the economic framing of a structural policy changes households' macroeconomic expectations. In a randomized survey experiment in the Bundesbank Online Panel of Households, all participants first read an identical neutral primer about climate policy measures and are then randomly assigned to receive no further text or an additional narrative interpreting the policy primarily as a negative demand or supply shock. Both narratives reduce expected growth. However, only the supply-shock framing raises inflation expectations, while the demand-shock framing does not reduce them-contrary to a simple demand-channel benchmark. These findings suggest that communication that makes different macro channels salient can materially shape expectations, with implications for economic policy communication during structural transitions. |
| Keywords: | climate change, expectations, survey experiments, RCT. |
| JEL: | C33 D84 E31 E52 Q4 |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:han:dpaper:dp-743 |
| By: | Ngoc-Thao Noet; Serge Blondel |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:tep:teppwp:wp26-02 |
| By: | Boltz, Marie; Bustelo, Monserrat; Díaz, Ana M.; Suaya, Agustina |
| Abstract: | This paper studies whether gender gaps in womens labor-market outcomes are sustained by systematic misperceptions about social and spousal support for maternal employment. Using a representative sample of 1, 732 cohabiting couples with young children in Bogotá, we show that while support for working mothers is nearly universal, both women and men substantially underestimate others supportparticularly fathersand frequently misperceive their partners views. We then implement a randomized information intervention that provides personalized feedback on prevailing local attitudes. The intervention reduces these misperceptions without altering individuals own attitudes. Treated men become 9 percentage points more likely to prioritize their wife for a scarce career-building opportunity, while womens choices change little. In the short run, treated women report more intensive job search and treated men place greater weight on workfamily balance. Effects are concentrated among women already active in the labor market, highlighting both the potential and the limits of norm-correcting information. |
| Keywords: | Gender norms;Female Employment;Pluralistic ignorance;RCT. |
| JEL: | C93 D91 J21 J16 |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:idb:brikps:14486 |
| By: | Simon Gaechter (University of Nottingham); Dominik Suri (University of Bonn); Sebastian Kube (University of Bonn) |
| Abstract: | AI-driven systems are rapidly moving from decision support to directing human behavior through rules, recommendations, and compliance requests. This shift expands everyday human–AI interaction and raises the possibility that AI may function as an authority figure. However, the behavioral consequences of AI as an authority figure remain poorly understood. We investigate whether individuals differ in their willingness to comply with arbitrary rules depending on whether these rules are attributed to an AI agent (ChatGPT) or to a fellow human. In a between-subject design, 977 US Prolific users completed the coins task: they could earn a monetary payoff by stopping the disappearance of coins at any time, but a rule instructed them to wait for a signal before doing so. There are no conventional reasons to follow this rule: complying is costly and nobody is harmed by non-compliance. Despite this, we find high rule-following rates: 64.3% followed the rule set by ChatGPT and 63.9% complied with the human-set rule. Descriptive and normative beliefs about rule following, aswell as compliance conditional on these beliefs, are also largely unaffected by the rule’s origin. However, subjective social closeness to the rule setter significantly predicts how participants condition their behavior on social expectations: when participants perceive the rule setter as subjectively closer, conditional compliance is higher and associated beliefs are stronger, irrespective of whether the rule setter is human or AI. |
| Keywords: | artificial intelligence; AI-human interaction; ChatGPT; rule-following; coins task; CRISP framework; social expectations; conditional rule conformity; social closeness; IOS11; online experiments |
| Date: | 2026–02 |
| URL: | https://d.repec.org/n?u=RePEc:not:notcdx:2026-02 |
| By: | Marc Witte (Vrije Universiteit Amsterdam and Tinbergen Institute); Johanna Roth (Sciences Po); Morgan Hardy (New York University Abu Dhabi); Christian Johannes Meyer (University of Oxford) |
| Abstract: | We present findings from an at-scale randomized trial of a government program providing public employment services in Addis Ababa, Ethiopia, with up-to-date vacancy information. Before the program, women with relatively less education searched more narrowly with worse labor market outcomes than the rest of our representative sample of relevant job seekers. These women also have lower direct intervention take-up than the rest of the sample. However, only these women significantly increase applications, receive more offers, shift from household enterprise work to wage employment, and experience higher earnings in response to the intervention. These employment impacts are larger than can be explained by vacancies directly curated through the intervention. Instead, these women adjust search behavior, expectations, and employment aspirations more broadly. Notably, offers come through friends and family networks, their modal baseline search method, underscoring the potential role of social networks in disseminating employment information to the most marginalized job seekers. |
| Keywords: | Public Employment Services, Labor Market Frictions, Marginalized Job Seekers, Randomized Controlled Trial (RCT) |
| JEL: | J08 J16 J64 O15 |
| Date: | 2025–07–25 |
| URL: | https://d.repec.org/n?u=RePEc:tin:wpaper:20250044 |
| By: | Fabien Giauque; Mehdi Farsi |
| Abstract: | Dynamic social norms have been recognized as a promising approach to promote energy sufficiency. By highlighting trends and future shifts rather than current states, dynamic norms allow for a better focus on emerging norms that are not widely adopted. While existing studies predominantly examine behavioral outcomes, the underlying processes and trade-offs remain to be explored. This paper uses a discrete choice experiment (DCE) combined with a randomized controlled trial to study electricity saving preferences under various dynamic norms. An emphasis is placed on the rationale for the norm changes. The results show that dynamic norms framed in terms of growing concerns about energy supply security positively affect electricity saving goal, whereas those framed around climate change do not. The heterogeneity analyses suggest that dynamic norms shape behavior through two complementary mechanisms: they generate new preferences while simultaneously reinforcing existing ones. The concluding analysis identifies four distinct groups that vary systematically in their preferences for electricity sufficiency. |
| Keywords: | Electricity saving; Dynamic Norms; Energy supply security; Climate change; Discrete choice experiment; Latent Class Model; Mixed Logit Model; Value-Belief-Norm Theory |
| JEL: | D12 D91 Q48 |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:irn:wpaper:26-01 |
| By: | Carter, Michael; Sunderland, Mark |
| Keywords: | Research and Development/Tech Change/Emerging Technologies, Risk and Uncertainty |
| URL: | https://d.repec.org/n?u=RePEc:ags:canzdp:263701 |
| By: | Pietro Bini; Lin William Cong; Xing Huang; Lawrence J. Jin |
| Abstract: | Do generative AI models, particularly large language models (LLMs), exhibit systematic behavioral biases in economic and financial decisions? If so, how can these biases be mitigated? Drawing on the cognitive psychology and experimental economics literatures, we conduct the most comprehensive set of experiments to date—originally designed to document human biases—on prominent LLM families across model versions and scales. We document systematic patterns in LLM behavior. In preference-based tasks, responses become more human-like as models become more advanced or larger, while in belief-based tasks, advanced large-scale models frequently generate rational responses. Prompting LLMs to make rational decisions reduces biases. |
| JEL: | D03 G02 G11 G4 G40 G41 |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:34745 |
| By: | Keefer, Philip; Ronconi, Lucas |
| Abstract: | Why are inefficient policies popular? One explanation is incomplete voter information. Evidence from survey experiments in Argentina points instead to inattention. The experiments explore voter evaluations of two anti-inflation policies, price controls and limits on monetary emission. Inattentive individuals should favor policies with simple, easily validated narratives, such as price controls; survey respondents considered price controls to be at least as effective in controlling inflation as limits to monetary emission. Two experimental treatments encouraged respondents to consider more complex policy narratives. Both increased respondent evaluations of the effectiveness of monetary policy relative to price controls. Inattention better accounts for these results: effects were independent of respondent knowledge and beliefs; larger for the evaluation of the more complex policy; and strongest for respondents who were more attentive to the survey. Treatment effects are also independent of the strength of partisan identity and ideological beliefs, indicating that these are low-cost cues for inattentive voters rather than signals of immutable beliefs regarding appropriate policies. The results underscore the role of attention in the spread of political narratives and their influence on voter behavior. |
| Keywords: | rational inattention;cognitive effort;attention allocation;Voter Behavior;Populism |
| JEL: | C90 D72 D80 D83 E71 P30 P50 |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:idb:brikps:14480 |
| By: | Simon Gaechter (University of Nottingham); Dominik Suri (University of Bonn); Sebastian Kube (University of Bonn); Johannes Schultz (University of Bonn) |
| Abstract: | Democratic societies depend on citizens following rules even when those rules are set by political opponents. Rising polarization may threaten this behavior. We test the impact of polarization on rule compliance in the United States across three pre-registered waves (May and November 2024; April 2025; n = 8, 340) using the “coins task”, which is a non-political, generic rule-following task, where breaking the rule increases payoffs. Participants were randomly assigned to follow rules set by the experimenter, a political co-partisan, a political opponent, or a non-partisan US citizen. Rule compliance ranged from 52.3% to 57.8%, and equivalence testing indicates no meaningful differences across waves or partisan rule-setter identities. However, greater affective distance from partisan rule setters is associated with lower compliance and weaker descriptive and normative beliefs about rule-following. These findings suggest that rule compliance is resilient to the rule-setter’s identity. While affective polarization may erode this behavior somewhat, substantial compliance remains: the human tendency to follow rules, even when incentivized to break them, survives the “stress test” of partisan rule-setting in highly polarized times. |
| Keywords: | Political polarization;affective polarization;rule-following;coinstask;norms;online experiments; political identity; equivalence testing; replication |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:not:notcdx:2026-01 |
| By: | James C. Cox; Vjollca Sadiraj |
| Abstract: | Paying salient rewards to subjects is expensive. Is it worth it? We describe previous data from otherwise identical treatments with hypothetical or salient payoffs that indicate opposite conclusions about the central research question in a paper. If we are going to pay salient rewards, how should we? The answer depends on the theoretical model underlying hypotheses being tested with the data, hence the choice of payoff protocol is an essential feature of experimental design. In recent years there has been renewed interest in theoretical and experimental research on properties of payoff protocols. We present an extension and critical discussion of the literature intended to promote uptake. |
| Keywords: | Salient payoffs, Payoff protocols, Incentive compatibility, Random selection, Experimental procedures |
| JEL: | C7 C9 D9 |
| Date: | 2025–12 |
| URL: | https://d.repec.org/n?u=RePEc:exc:wpaper:2025-03 |
| By: | Joshi, Mukta; Teh, Wen Wen; Vargas, Ariadna; Dadzie, Christabel E.; Datta, Saugato |
| Abstract: | Abstract Cash plus programs, which combine cash transfers with complementary services and interventions, have become an increasingly popular tool for providing livelihood support and poverty alleviation in low- and middle-income countries. While there is robust evidence to indicate that cash programs provide poverty relief in the short term, the impact of cash programs on productive investment behaviors and activities is less understood. This study presents the results of a cluster randomized trial that evaluates the effects of light-touch behavioral interventions to encourage saving and entrepreneurial behaviors for low-income Ghanaians participating in a multi-faceted cash plus program focused on economic inclusion. Participants received business skills training, coaching and mentoring, and a cash grant to support the initiation and expansion of their businesses. The study incorporated a suite of behavioral interventions designed to help recipients set business-related savings goals, create plans for achieving those goals, and follow through on saving towards those goals. In addition, the design included a pamphlet outlining key steps for growing or expanding a business, accompanied by a tracker to help recipients remember these steps and track progress. Drawing from a sample of 3, 109 participants, the study found that behavioral interventions significantly improved goal-setting and plan-making behaviors related to savings and consequently, increased the incidences of saving among study participants. However, the study did not find a statistically significant impact of the behavioral interventions on improving business skills. Using a cluster-randomized trial (N=3, 109), this study evaluated the effects of light-touch behavioral interventions on recipients of a multi-faceted cash plus program for economic inclusion, which included business skills training, coaching and mentoring, and a cash grant to support the initiation and expansion of businesses. Results show that the behavioral interventions, featuring goal-setting and plan-making activities, savings tools, and business practice reminders, improved goal-setting and plan-making behaviors related to savings and consequently, increased the incidence of saving. However, the interventions did not significantly improve business practices. Findings suggest that simple behavioral tools can strengthen savings behaviors and financial resilience among poo r households, complementing cash and training programs, though further research is needed on long-term effects. |
| Date: | 2025–11–30 |
| URL: | https://d.repec.org/n?u=RePEc:wbk:hdnspu:207029 |
| By: | Muehlheusser, Gerd; Roider, Andreas |
| Abstract: | The main aim of the project was to contribute to a better understanding of group decisions. In a series of randomized experiments, we study the effects of important group characteristics such as group size, group gender composition, and other potential determinants of group behavior. Due to the frictions caused by the Covid-19 pandemic, the initial project plan had to be adapted. Developing the video chat tool chaTree necessary to allow for face-to-face communication in online experiments took considerable time, which lead us to focus on the domains of unethical behavior and solving complex tasks as fields of application. Our results also have interesting implications outside academia, in particular for the design of work teams and committees. For example, a diffusion of responsibility seems to be a highly relevant phenomenon in the context of unethical decisions. Moreover, larger groups behave more unethically than smaller ones, and all-male groups seem to be particularly prone to unethical behavior and should hence be avoided. Moreover, we also find that group diversity seems to enhance group performance, particularly in smaller groups. The experiments were originally planned to be conducted in the lab, allowing group members to interact and communicate with each other. However, this was made impossible by the pandemic, which hit shortly after the project started. This forced us to resort to online experiments. Thereby, we were facing the challenge that at that point, in online experiments there was no possibility of having face-to-face communication in groups, and the only feasible mode of communication was the exchange of (written) chat messages. For the purpose of our project, this was not satisfactory. As a response, one (unforeseen) methodological contribution of the project was the development of the tool chaTree. We are convinced that it will also be helpful to other researchers. |
| Abstract: | Das Hauptziel des Projekts war es, zu einem besseren Verständnis von Gruppenentscheidungen beizutragen. In einer Reihe ökonomischer Experimente untersuchen wir die Auswirkungen wichtiger Gruppenmerkmale, wie der Gruppengröße, der Zusammensetzung der Gruppe nach Geschlecht und anderen potenzielle Determinanten des Gruppenverhaltens. Aufgrund der durch die Covid-19-Pandemie verursachten Friktionen waren im Verlauf Anpassungen des Projektplans nötig. Die Entwicklung des Video-Chat-Tools chaTree, die notwendig war, um die Kommunikation in Bild und Ton in Online-Experimenten zu ermöglichen, hat substanzielle Zeit in Anspruch genommen. Dies hat uns veranlasst, uns auf unethisches Verhalten und das Lösen komplexer Aufgaben als Anwendungsgebiete zu konzentrieren. Unsere Ergebnisse haben auch außerhalb der Forschung Relevanz, insbesondere hinsichtlich der Ausgestaltung von Arbeitsteams oder Gremien. Beispielsweise scheint die Diffusion von Verantwortung ein höchst relevantes Phänomen im Zusammenhang mit unethischen Entscheidungen zu sein. Außerdem scheinen sich größere Gruppen unethischer zu verhalten als kleinere, und rein männliche Gruppen sind offenbar besonders anfällig für unethisches Verhalten und aus dieser Perspektive potenziell problematisch. Darüber hinaus scheint es, dass Heterogenität in der Zusammensetzung insbesondere für die Leistung kleinerer Gruppen förderlich ist. Ursprünglich sollten die Experimente im Labor durchgeführt werden, um den Gruppenmitgliedern die Möglichkeit zu geben, miteinander zu kommunizieren. Dies wurde durch die Pandemie, die kurz nach Beginn des Projekts ausbrach, jedoch unmöglich gemacht. Dadurch waren wir gezwungen, auf Online-Experimente auszuweichen. Dies war jedoch mit der Herausforderung verbunden, dass es zu diesem Zeitpunkt in Online-Experimenten keine Möglichkeit gab, in Bild und Ton zu kommunizieren. Die einzig praktikable Art der Kommunikation war vielmehr der Austausch von schriftlichen Chat-Nachrichten. Für die Zwecke unseres Projekts war dies nicht zufriedenstellend. Ein (unvorhergesehener) methodischer Beitrag des Projekts war daher die Entwicklung des Tools chaTree. Wir sind überzeugt, dass es auch für andere Forschende hilfreich sein wird. |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:esrepo:335700 |
| By: | Maria Chaykina (Department of Economics (University of Verona)) |
| Abstract: | Notional Defined Contribution (NDC) pension schemes convert accumulated pension wealth into an annuity, based on an average life expectancy at retirement. When longevity differs across social groups, a single conversion factor implies systematic transfers from shorter-lived to longer-lived individuals. This motivates proposals to differentiate benefits by socio-demographic characteristics related to life expectancy. We study whether such differentiation is perceived as fair using a survey experiment involving 3, 004 Italian residents aged 18-66. Respondents completed an incentivised allocation task used to elicit their fairness views and then evaluated six reform scenarios that adjust pension benefits based on gender, region, income, household wealth, workplace fatigue, and health status. The results show that the fatigue-based and wealth-based scenarios receive the highest support, whereas the gender-based and region-based scenarios are strongly opposed. Self-interest predicts approval, with higher support among those who stand to gain from a reform. Respondents with libertarian views are consistently less supportive of changes in benefits, while egalitarians and, to a lesser extent, liberal egalitarians are more favourable. Our results inform policymakers on the importance of citizens’ fairness perceptions for the implementation and communication of financially sustainable pension reforms. |
| Keywords: | Pensions, Fairness, Life Expectancy, Survey Experiment, Redistribution, Italy |
| JEL: | H55 D63 C91 C38 |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:ver:wpaper:01/2026 |
| By: | Hoyos, Manuela; Perez-Vincent, Santiago M.; Tobón, Santiago; Souza, Pedro CL; Vanegas-Arias, Martín |
| Abstract: | We study whether structured, evidence-based information can shape local policymakers beliefs and stated intentions to adopt violence prevention interventions. Using a randomized experiment embedded in Colombia's national policy planning cycle, we exposed municipal officials to varying types of information about programs aimed at reducing violence against women. The information differed along two key dimensions: the strength of the underlying evidence (effective or inconclusive) and the inclusion of practical implementation guidance. We find that receiving information increased the expected effectiveness of interventions by 3.7 percentage points and raised willingness to implement by 0.15 points on a 14 scale. These effects were larger and more precisely estimated when the intervention had strong empirical support and when the information included concrete implementation guidance. Our findings suggest that actionable, credible information can meaningfully shift beliefs and policy intentions, especially when it helps officials evaluate both the relevance and feasibility of applying a program in their own context. The results contribute to growing evidence on information frictions in public administration, showing that policymakers are responsive to research when it is presented in a structured and practically useful format. Tailoring dissemination strategies to address local implementation needs may be key to encouraging evidence-informed policymaking. |
| JEL: | D73 J16 H76 |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:idb:brikps:14484 |
| By: | Ting-Chih Hung; Yu-Chang Chen |
| Abstract: | We study the identification and estimation of long-term treatment effects under unobserved confounding by combining an experimental sample, where the long-term outcome is missing, with an observational sample, where the treatment assignment is unobserved. While standard surrogate index methods fail when unobserved confounders exist, we establish novel identification results by leveraging proxy variables for the unobserved confounders. We further develop multiply robust estimation and inference procedures based on these results. Applying our method to the Job Corps program, we demonstrate its ability to recover experimental benchmarks even when unobserved confounders bias standard surrogate index estimates. |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2601.17712 |
| By: | Sheheryar Banuri (School of Economics, University of East Anglia and Hughes Hall, University of Cambridge (UK)); Christa Brunnschweiler (Department of Economics, Norwegian University of Science and Technology and CESifo); Deanna Karapetyan (Financial Conduct Authority (UK)) |
| Abstract: | This paper investigates why firms engage in costly environmental and ethical practices, focusing on whether consumer responses depend on firms’ intentions or outcomes. Existing literature links ESG practices to positive performance and stakeholder rewards, but most evidence is observational and cannot disentangle intentionality from outcomes. Using a controlled experiment, we examine consumer reactions when firms choose between a “clean†technology (avoiding harm at a cost) and a “dirty†technology (higher returns with negative externalities). Two treatments isolate intentionality: Random Choice versus Willful Choice. After observing the firm’s choice and the resulting externality, consumers can respond by transferring (taking away) resources to the firm in a give-or-take Dictator Game. We find a pronounced asymmetry in how intentions matter. Consumers punish firms whenever a negative externality is incurred, regardless of intentionality, indicating that punitive responses are largely outcome-driven. By contrast, when harm is avoided, intentions play a central role: firms that deliberately choose to prevent a negative externality are treated with significantly greater leniency than firms for which absence of harm arises randomly, reflected in positive transfers on average. These findings highlight that intentionality affects punitive responses and helps explain why firms may voluntarily adopt costly ethical practices when choices are observable. |
| Keywords: | Intentionality; Harm Avoidance; Consumer Responses; ESG; Environmental Externalities |
| JEL: | D03 D64 L21 C91 |
| Date: | 2026–01–29 |
| URL: | https://d.repec.org/n?u=RePEc:nst:samfok:20526 |
| By: | Yu Wang; Xiangchen Liu |
| Abstract: | As LLMs increasingly function as economic agents, the specific mechanisms LLMs use to update their belief with heterogeneous signals remain opaque. We design experiments and develop a Behavioral Kalman Filter framework to quantify how LLM-based agents update expectations, acting as households or firm CEOs, update expectations when presented with individual and aggregate signals. The results from experiments and model estimation reveal four consistent patterns: (1) agents' weighting of priors and signals deviates from unity; (2) both household and firm CEO agents place substantially larger weights on individual signals compared to aggregate signals; (3) we identify a significant and negative interaction between concurrent signals, implying that the presence of multiple information sources diminishes the marginal weight assigned to each individual signal; and (4) expectation formation patterns differ significantly between household and firm CEO agents. Finally, we demonstrate that LoRA fine-tuning mitigates, but does not fully eliminate, behavioral biases in LLM expectation formation. |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2601.17527 |
| By: | Tsai, Chia-Jung; Lechner, Clemens; Behr, Dorothée; Nkong, Ulrike Efu; Radinger, Anke |
| Abstract: | Comparable questionnaire translation is essential for drawing valid conclusions in cross-cultural survey research. Sound translation methodology, including the use of adequate personnel, is seen as crucial for reaching this goal (Harkness 2003). Recommended methodology should be empirically backed and stay tuned to latest developments, such as machine translation. Against this backdrop, to investigate the potential effect of varied translators’ backgrounds and machine translation on the statistical properties of surveys, we conducted an experiment in which an English questionnaire was translated into German by 16 professional translators and 16 social scientists; translations were subsequently fielded in web surveys. We introduced two translation conditions: translation from scratch and post-editing (machine translation corrected by a human translator). To investigate the quality of the survey data from these 32 translation versions (approx. 250 responses each), we use standardized mean distance and Cohen’s d with the official translation as a benchmark. We have four key findings: First, the resulting statistical means of the survey items vary, sometimes substantially, across translations. Second, post-editing is associated with a reduced gap between the survey data from the experimental questionnaires and the official translation, and it also lowers the variability among different translations. Third, when translating from scratch, social scientists are more likely to produce translations leading to statistical outliers of survey data. Fourth, post-editing can lead to systematic bias for both social scientists and professional translators if translation errors made by the machine are not identified and corrected. This study highlights to what extent decisions concerning the choice of translators and the integration of machine translation can impact the statistical properties of survey data. We offer evidence to implement recommendations for good practices in translation protocols to enhance data comparability in cross-cultural studies. |
| Date: | 2026–01–23 |
| URL: | https://d.repec.org/n?u=RePEc:osf:socarx:rw6f2_v1 |
| By: | Chonghuan Wang |
| Abstract: | Matching mechanisms play a central role in operations management across diverse fields including education, healthcare, and online platforms. However, experimentally comparing a new matching algorithm against a status quo presents some fundamental challenges due to matching interference, where assigning a unit in one matching may preclude its assignment in the other. In this work, we take a design-based perspective to study the design of randomized experiments to compare two predetermined matching plans on a finite population, without imposing outcome or behavioral models. We introduce the notation of a disagreement set, which captures the difference between the two matching plans, and show that it admits a unique decomposition into disjoint alternating paths and cycles with useful structural properties. Based on these properties, we propose the Alternating Path Randomized Design, which sequentially randomizes along these paths and cycles to effectively manage interference. Within a minimax framework, we optimize the conditional randomization probability and show that, for long paths, the optimal choice converges to $\sqrt{2}-1$, minimizing worst-case variance. We establish the unbiasedness of the Horvitz-Thompson estimator and derive a finite-population Central Limit Theorem that accommodates complex and unstable path and cycle structures as the population grows. Furthermore, we extend the design to many-to-one matchings, where capacity constraints fundamentally alter the structure of the disagreement set. Using graph-theoretic tools, including finding augmenting paths and Euler-tour decomposition on an auxiliary unbalanced directed graph, we construct feasible alternating path and cycle decompositions that allow the design and inference results to carry over. |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2601.21036 |
| By: | Zhiqi Zhang; Zhiyu Zeng; Ruohan Zhan; Dennis Zhang |
| Abstract: | Randomized Controlled Trials (RCTs), or A/B testing, have become the gold standard for optimizing various operational policies on online platforms. However, RCTs on these platforms typically cover a limited number of discrete treatment levels, while the platforms increasingly face complex operational challenges involving optimizing continuous variables, such as pricing and incentive programs. The current industry practice involves discretizing these continuous decision variables into several treatment levels and selecting the optimal discrete treatment level. This approach, however, often leads to suboptimal decisions as it cannot accurately extrapolate performance for untested treatment levels and fails to account for heterogeneity in treatment effects across user characteristics. This study addresses these limitations by developing a theoretically solid and empirically verified framework to learn personalized continuous policies based on high-dimensional user characteristics, using observations from an RCT with only a discrete set of treatment levels. Specifically, we introduce a deep learning for policy targeting (DLPT) framework that includes both personalized policy value estimation and personalized policy learning. We prove that our policy value estimators are asymptotically unbiased and consistent, and the learned policy achieves a root-n-regret bound. We empirically validate our methods in collaboration with a leading social media platform to optimize incentive levels for content creation. Results demonstrate that our DLPT framework significantly outperforms existing benchmarks, achieving substantial improvements in both evaluating the value of policies for each user group and identifying the optimal personalized policy. |
| Date: | 2026–02 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2602.05099 |
| By: | Shrestha, Maheshwor; Okunogbe, Oyebola M.; Kalra, Naira; Fashogbon, Ayodele Emmanuel; Bossuroy, Thomas; Audy, Robin; Ajayi, Kehinde |
| Abstract: | A rigorous randomized controlled trial in Nigeria shows that integrated safety nets produce sustained improvements in nutrition and food security for poor households relative to a basic safety nets program, with effects lasting three years. Stunting rates among young children fall by 18 percent only when all three critical features are present: (i) targeting families during early childhood; (ii) substantial grant on top of a regular safety net support; and (iii) behavior change training. Scaling up these proven interventions is urgent to address Nigeria’s high burden of stunting and food insecurity and unlock future human capital. |
| Date: | 2025–12–31 |
| URL: | https://d.repec.org/n?u=RePEc:wbk:hdnspu:207679 |
| By: | Pedro Picchetti |
| Abstract: | This paper develops a finite population framework for analyzing causal effects in settings with imperfect compliance where multiple treatments affect the outcome of interest. Two prominent examples are factorial designs and panel experiments with imperfect compliance. I define finite population causal effects that capture the relative effectiveness of alternative treatment sequences. I provide nonparametric estimators for a rich class of factorial and dynamic causal effects and derive their finite population distributions as the sample size increases. Monte Carlo simulations illustrate the desirable properties of the estimators. Finally, I use the estimator for causal effects in factorial designs to revisit a famous voter mobilization experiment that analyzes the effects of voting encouragement through phone calls on turnout. |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2601.16749 |
| By: | Ruicheng Ao; Hongyu Chen; Haoyang Liu; David Simchi-Levi; Will Wei Sun |
| Abstract: | We study semi-supervised stochastic optimization when labeled data is scarce but predictions from pre-trained models are available. PPI and SVRG both reduce variance through control variates -- PPI uses predictions, SVRG uses reference gradients. We show they are mathematically equivalent and develop PPI-SVRG, which combines both. Our convergence bound decomposes into the standard SVRG rate plus an error floor from prediction uncertainty. The rate depends only on loss geometry; predictions affect only the neighborhood size. When predictions are perfect, we recover SVRG exactly. When predictions degrade, convergence remains stable but reaches a larger neighborhood. Experiments confirm the theory: PPI-SVRG reduces MSE by 43--52\% under label scarcity on mean estimation benchmarks and improves test accuracy by 2.7--2.9 percentage points on MNIST with only 10\% labeled data. |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2601.21470 |
| By: | Constance Frohly; Roberto Galbiati; Emeric Henry; Nicolas Jacquemet (PSE - Paris School of Economics - UP1 - Université Paris 1 Panthéon-Sorbonne - ENS-PSL - École normale supérieure - Paris - PSL - Université Paris Sciences et Lettres - EHESS - École des hautes études en sciences sociales - ENPC - École nationale des ponts et chaussées - CNRS - Centre National de la Recherche Scientifique - INRAE - Institut National de Recherche pour l’Agriculture, l’Alimentation et l’Environnement, CES - Centre d'économie de la Sorbonne - UP1 - Université Paris 1 Panthéon-Sorbonne - CNRS - Centre National de la Recherche Scientifique, UP1 - Université Paris 1 Panthéon-Sorbonne) |
| Abstract: | Moral reminders, also referred to as moral appeals or moral nudges, are widely used by governments, companies, and NGOs to promote pro-social behavior. These appeals function by either increasing the salience of moral concerns or the cost of diverting attention away from relevant information on payoffs or social norms. Drawing on over 400 studies across psychology, sociology, management and economics, we present a meta-analysis of their effects. Our findings reveal that, on average, moral reminders are effective, with an effect size (Hedge's g) of 0.24 in a random-effects model, but with significant backfiring occurring in 12% of studies. We identify sources of heterogeneity based on disciplinary focus and design choices. Crucially, we introduce a taxonomy of moral reminders: we distinguish those that provide information on consequences, those that highlight descriptive or injunctive norms, and those that prime moral awareness. Our analysis shows that all of these instruments are effective, particularly those providing information on consequences, whereas information on injunctive norms is more likely to backfire. |
| Keywords: | Meta-analysis, Behavioral ethics, Moral reminders, Pro-social behavior |
| Date: | 2026–01–01 |
| URL: | https://d.repec.org/n?u=RePEc:hal:cesptp:halshs-05456784 |