|
on Experimental Economics |
| By: | Matthias Sutter (Max Planck Institute for Research on Collective Goods, Bonn, University of Cologne, University of Innsbruck, IZA Bonn, and CESifo Munich); Michael Weyland (Ludwigsburg University of Education); Anna Untertrifaller (University of Cologne); Manuel Froitzheim (University of Siegen); Sebastian O. Schneider (Max Planck Institute for Research on Collective Goods, Bonn) |
| Abstract: | We present the results of a randomized intervention to study how teaching financial literacy to 16-year old high-school students affects their behavior in risk and time preference tasks. Compared to two different control treatments, we find that teaching financial literacy makes subjects behave more patiently, more time-consistent, and more risk-averse. These effects persist for up to almost 5 years after our intervention. Behavior in the risk and time preference tasks is related to financial behavior outside the lab, in particular spending patterns. This shows that teaching financial literacy affects economic decision-making which in turn is important for field behavior. |
| Keywords: | Financial literacy, randomized intervention, risk preferences, time preferences, financial behavior, field experiment |
| JEL: | C93 D14 I21 |
| Date: | 2023–05 |
| URL: | http://d.repec.org/n?u=RePEc:ajk:ajkdps:229&r=exp |
| By: | Athey, Susan (Stanford U); Cersosimo, Matias (Stanford U); Koutout, Kristine (Stanford U); Li, Zelin (MIT) |
| Abstract: | Two leading hypotheses for why individuals unintentionally share misinformation are that 1) they are unable to recognize that a post contains misinformation, and 2) they make impulsive, emotional sharing decisions without thinking about whether a post contains misinformation. Much of the literature on interventions to counter misinformation focuses on the first hypothesis and tests interventions that educate social media users about reasoning-based techniques employed in social media posts to mislead them; however, other work shows that emotions are important for the spread of and belief in misinformation, supporting a focus on the second hypothesis and emotion-based techniques. To learn whether interventions to counter reasoning- or emotion-based techniques are more effective, or whether the approaches are complementary, we evaluate three distinct versions of a five-day, low-cost, and scalable text message educational course in a field experiment with approximately 9, 000 participants in Kenya. We assess the impact of the courses through a pre-post survey design eliciting intentions to share, finding that all treatment courses work, decreasing misinformation sharing 28% on average relative to no text message course. Treating the emotional drivers of misinformation sharing in the "Emotions" course is more effective than teaching about reasoning-based techniques either alone in the "Reasoning" course or in combination with emotion-based techniques in the "Combo" course. Moreover, the Emotions course performs best on misinformation posts that use emotion-based techniques, and does no worse than the Reasoning or Combo courses on misinformation posts that use reasoning-based techniques. In a follow-up experiment approximately two months later, 88% of the treatment effect of the three courses on misinformation sharing persists. |
| Date: | 2022–11 |
| URL: | http://d.repec.org/n?u=RePEc:ecl:stabus:4073&r=exp |
| By: | Tanja Hoertnagl; Rudolf Kerschbamer; Regine Oexl; Rudi Stracke; Uwe Sunde |
| Abstract: | We investigate how heterogeneity in contestants' investment costs affects competition expenditures in a dynamic elimination contest with different seeding variants of contestants. Theory predicts that expenditures in dynamic contests are lower when competitors are heterogeneous than when they are homogeneous. Cost heterogeneity influences expenditures directly - by inducing weak and strong competitors to reduce their expenditures - and indirectly - through their influence on continuation values. We present evidence from lab experiments that is qualitatively in line with the theoretical prediction for contestants with low investment costs: they incorporate the heterogeneity and the differences in continuation values when competing in stage one and they decrease their expenditures when competing against a weak agent in stage two. For high-cost contestants, the theoretical predictions are not confirmed: expenditures in heterogeneous interactions are not lower and sometimes even higher. As a consequence, we find that total expenditures in heterogeneous dynamic contests are not necessarily lower than in homogeneous ones. |
| Keywords: | Multi-Stage Contest, Heterogeneity, Experiment |
| JEL: | C72 D72 |
| Date: | 2023–04 |
| URL: | http://d.repec.org/n?u=RePEc:inn:wpaper:2023-04&r=exp |
| By: | Patricia Gil; Justin Holz; John List; Andrew Simon; Alejandro Zentner |
| Abstract: | In modern economies, when debt and trust issues arise, a partial forgiveness policy is often the solution to induce payment and increase disclosure. For their part, governments around the globe continue to use tax amnesties as a strategy to allow debtors to make amends for past misdeeds in exchange for partial debt forgiveness. While ubiquitous, much remains unknown about the basic facts of how well amnesties work, for whom, and why. We present a simple theoretical construct that provides both economic clarity into tax amnesties as well as insights into the necessary behavioral parameters that one must estimate to understand the consequences of tax amnesties. We partner with the Dominican Republic Tax Authorities to design a natural field experiment that is linked to the theory to estimate key causal mechanisms. Empirical results from our field experiment, which covers 125, 452 taxpayers who collectively owe $5.2 billion (5.5% of GDP) in known debt, highlight the import of deterrence laws, beliefs about future amnesties, and tax morale for debt payment and increased disclosure. Importantly, we find large short run effects: our most effective treatment (deterrence) increased payments of known debt by 25% and hidden debt by 48%. Further, we find no evidence of our intervention backfiring on subsequent tax payments. |
| Date: | 2023 |
| URL: | http://d.repec.org/n?u=RePEc:feb:natura:00772&r=exp |
| By: | Kapoor, Anuj (Indian Institute of Management, Ahmedabad); Narayanan, Sridhar (Stanford U); Sharma, Amitt (VDO.AI) |
| Abstract: | Modern digital advertising platforms allow ads to be targeted in a variety of ways, and generally aim to match the ad being shown with either the user or the content being shown. In this study, we examine the effect of matching in emotional content of ads and the video on which the ad is shown on consumers' engagement with the ad. On the one hand, ads that are emotionally matched with the content video could lead to greater engagement with the ad because of the desire for consumers to extend the emotion they are experiencing. On the other hand, emotional mismatch between the ad and content video can lead to greater perceptual contrast, thereby drawing more attention to the ad. Consumers may also be satiated in the emotion they feel in the content video. Both of these mechanisms would lead to greater engagement with the ad if it were mismatched with the content video. Thus, whether emotional matching is more effective in driving ad engagement, and thereby potentially consumers' evaluation of, affect towards, and purchase of the advertised good is an empirical question. We study this question through a field experiment run in collaboration with VDO.AI, a video and ad serving platform. In this experiment, we manipulate the video/ad combination consumers see, with variation across the experimental conditions in the emotional content of ads and videos. We find that in our setting incongruence, where emotional content of videos and ads are mismatched, leads to greater ad engagement. We find suggestive evidence for attention being the mechanism through which incongruence leads to better outcomes. Our findings contribute to the literature on advertising, and provide an important targeting variable for firms in the AdTech space. |
| Date: | 2022–10 |
| URL: | http://d.repec.org/n?u=RePEc:ecl:stabus:4041&r=exp |
| By: | Prest, Brian C. (Resources for the Future); Wichman, Casey (Resources for the Future); Palmer, Karen (Resources for the Future) |
| Abstract: | We investigate how well machine learning counterfactual prediction tools can estimate causal treatment effects. We use three prediction algorithms—XGBoost, random forests, and LASSO—to estimate treatment effects using observational data. We compare those results to causal effects from a randomized experiment for electricity customers who faced critical-peak pricing and information treatments. Our results show that each algorithm replicates the true treatment effects, even when using data from treated households only. Additionally, when using both treatment households and nonexperimental comparison households, simpler difference-in-differences methods replicate the experimental benchmark, suggesting little benefit from ML approaches over standard program evaluation methods.Click "Download" above to read the full paper. |
| Date: | 2021–09–29 |
| URL: | http://d.repec.org/n?u=RePEc:rff:dpaper:dp-21-30&r=exp |
| By: | Timo Goeschl; Marcel Oestreich; Alice Soldà (-) |
| Abstract: | How to design audit mechanisms that harness the benefits of self-reporting for achieving compliance with regulatory targets while limiting misreporting is a pressing question in many regulatory contexts, from climate policies to public health. Contrasting random audit and competitive audit mechanisms, this paper theoretically and experimentally studies their performance in regulating socially undesirable emissions when peer information about others’ emissions is present or absent. Our focus is on the compliance of emission levels with regulatory targets, going beyond existing results on truthfulness of reporting. Confirming theoretical predictions, the experiment shows that in contrast to the random audit mechanism, the competitive audit mechanism can leverage peer information for compliance: emission levels are closer to the social optimum. Yet, emission levels fall somewhat short of full compliance. The results highlight the considerable potential of competitive audit mechanisms for achieving not only more truthfulness, but also more compliance. |
| Keywords: | Regulation; compliance; tournament theory; online experiment |
| JEL: | D62 H41 H83 L51 Q58 |
| Date: | 2023–05 |
| URL: | http://d.repec.org/n?u=RePEc:rug:rugwps:23/1069&r=exp |
| By: | Monica Martinez-Bravo (CEMFI); Carlos Sanz (Banco de España) |
| Abstract: | The COVID-19 pandemic took place against the backdrop of growing political polarization and distrust in political institutions in many countries. Did deficiencies in government performance further erode trust in public institutions? Did citizens’ ideology interfere with the way they processed information on government performance? To investigate these two questions, we conducted a pre-registered online experiment in Spain in November 2020. Respondents in the treatment group were provided information on the number of contact tracers in their region, a key policy variable under the control of regional governments. We find that individuals greatly over-estimate the number of contact tracers in their region. When we provide the actual number of contact tracers, we find a decline in trust in governments, a reduction in willingness to fund public institutions and a decrease in COVID-19 vaccine acceptance. We also find that individuals endogenously change their attribution of responsibilities when receiving the treatment. In regions where the regional and central governments are controlled by different parties, sympathizers of the regional incumbent react to the negative news on performance by attributing greater responsibility for it to the central government. We call this the blame shifting effect. In those regions, the negative information does not translate into lower voting intentions for the regional incumbent government. These results suggest that the exercise of political accountability may be particularly difficult in settings with high political polarization and areas of responsibility that are not clearly delineated. |
| Keywords: | trust, accountability, polarization, COVID-19 |
| JEL: | P00 D72 H1 H7 |
| Date: | 2023–01 |
| URL: | http://d.repec.org/n?u=RePEc:bde:wpaper:2306&r=exp |
| By: | Miller, Klaus M. (HEC Paris); Sahni, Navdeep S. (Stanford U); Strulov-Shlain, Avner (U of Chicago) |
| Abstract: | Consumer inertia, the tendency to remain inactive, is a robust and well-documented phenomenon. However, if consumers are aware of their future inertia they can act to mitigate its effects on their outcomes. Using a large-scale randomized field experiment with a leading European newspaper we investigate consumer response to inertia-inducing subscription contracts and study, in the same setting, both the actual inertia, and the inertia consumers anticipate before it actually takes place. We vary the promotional subscription price, the duration, and whether the contract automatically renews by default, or not, after the promotional period. Indeed, we find strong inertia. Roughly half of auto- renewal contract takers continue to a full pay subscription after the promotional period, relative to the auto-cancel contract takers who rarely renew. Those added auto-renewal subscribers do not use their subscription to access the newspaper. However, consumers preempt inertia; 24%-36% of potential subscribers avoid subscribing on the first weeks after being offered an auto-renewal contract. Further, the share of subscribers, at all, for two years after the promo is 10% lower due to being offered the auto-renewal contract. Overall, even though auto-renewal generates a higher revenue in the short term, auto-renewal and auto-cancel are revenue equivalent after one year, but with fewer subscribers in auto- renewal. Using a simple mixed-type model we quantify inertia, the share of inert readers, and the share of sophisticated readers who are aware of it. Our estimates suggest that half of the readers are inert. At most 41% of these inert individuals are unaware of their future inertia, equivalent to a 72% monthly chance of not cancelling an unwanted subscription. Finally, we show that targeting contract types to maximize revenue or subscriptions does not pick up, ex post, sophistication. Our results highlight the often-ignored effects of potentially exploitative inertia-inducing contracts: lower take up in the short- and long-run driven by sophisticated consumers. |
| Date: | 2023–01 |
| URL: | http://d.repec.org/n?u=RePEc:ecl:stabus:4077&r=exp |
| By: | Léon Bartosch (Max Planck Institute for Research on Collective Goods, Bonn); Dorothee Mischkowski (Max Planck Institute for Research on Collective Goods, Bonn) |
| Abstract: | Several models of social preferences have been developed at the intersection of social psychology and behavioral economics, such as social value orientation (SVO) and conditional cooperation. Whereas SVO is well researched in its dispositional and situational correlates, we aim to locate conditional cooperation within the HEXACO personality model, particularly expecting a relation to reactive vs. active prosociality (i.e., Agreeableness vs. Honesty-Humility). Contrary to our expectations, however, in two preregistered, incentivized studies (n total = 521) conditional cooperation was neither related to Agreeableness nor to Honesty-Humility. When investigating the relation between SVO and conditional cooperation, we conceptually replicate a positive relation between both (pro-)social preferences. Surprisingly, while prosocials coincide with conditional cooperators, even most individualists who maximize their outcome in unilateral giving turn to conditionally cooperative behavior in strategic interactions. This underlines the importance of shaping situations as reciprocal acts to elicit cooperative behavior from originally self-interested individuals. |
| Keywords: | Conditional Cooperation, Social Value Orientation, Basic Personality Traits, HEXACO, Reciprocity |
| JEL: | C93 D01 D91 I12 |
| Date: | 2023–01 |
| URL: | http://d.repec.org/n?u=RePEc:mpg:wpaper:2023_01&r=exp |
| By: | Freundt, Jana (University of Fribourg, Switzerland); Herz, Holger (University of Fribourg, Switzerland); KOPP, leander (University of Fribourg, Switzerland) |
| Abstract: | Personal autonomy has been argued to be fundamental to well-being and is often discussed as an important driver of economic and political behavior. Yet, preferences for autonomy are not well understood, because their identification requires the separation of instrumental value attached to autonomous choice. We propose a novel elicitation method that solves this identification challenge. We establish the existence of intrinsic preferences for choice autonomy and show substantial heterogeneity in a large online sample. We further study their antecedents by relating them to existing personality scales and socioeconomic characteristics. Finally, we test their association with other preferences, attitudes and beliefs. |
| Keywords: | autonomy, delegation, experiment design, choice consistency |
| JEL: | C91 D01 D90 |
| Date: | 2023–04–07 |
| URL: | http://d.repec.org/n?u=RePEc:fri:fribow:fribow00530&r=exp |
| By: | Maximilian Andres |
| Abstract: | So far, the theory of equilibrium selection in the infinitely repeated prisoner's dilemma is insensitive to communication possibilities. To address this issue, we incorporate the assumption that communication reduces -- but does not entirely eliminate -- an agent's uncertainty that the other agent follows a cooperative strategy into the theory. Because of this, agents still worry about the payoff from cooperating when the other one defects, i.e. the sucker's payoff S, and, games with communication are more conducive to cooperation than games without communication. This theory is supported by data from laboratory experiments, and by machine learning based evaluation of the communication content. |
| Date: | 2023–04 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2304.12297&r=exp |
| By: | Aziz, Sonia; Boyle, Kevin; Akanda, Ali S.; Hanifi, M.A.; Pakhtigian, Emily L. |
| Abstract: | In Bangladesh, cholera poses a significant health risk. Yet, information about the nature and severity of cholera risk is limited as risk varies over time and by location and changing weather patterns have made historical cholera risk predictions less reliable. In this paper, we examine how households use geographically and temporally personalized cholera risk predictions to inform their water use behaviors. Using data from an eight month field experiment, we estimate how access to a smartphone application containing monthly cholera risk predictions unique to a user’s home location affects households’ knowledge about their cholera risk as well as their water use practices. We find that households with access to this application feel more equipped to respond to environmental and health risks they may face and reduce their reliance on surface water for bathing and washing—a common cholera transmission pathway. We do not find that households invest additional resources into drinking water treatment, nor do we find reductions in self-reported cholera incidence. Access to dynamic risk information can help households make safer water choices; tailoring information provision to those at highest risk could reduce cholera transmission in endemic areas. |
| Date: | 2022–10–19 |
| URL: | http://d.repec.org/n?u=RePEc:rff:dpaper:dp-22-24&r=exp |
| By: | Emanuele Ciani; Louis Fréget; Thomas Manfredi |
| Abstract: | A growing body of literature studies the effect of providing information about inequality to respondents of surveys on their preferences for redistribution. We provide a meta-analysis combining the results from 84 information treatments coming from 36 studies in Economics, Political Science, Psychology and Sociology. This meta-analysis complements and informs a broader project on perceptions of inequality and preferences for redistribution (Does Inequality Matter? How People Perceive Economic Disparities and Social Mobility, OECD publishing, Paris, 2021). In the meta-analysis, we focus on in-survey experiments where a randomly selected group of respondents receive either information about the overall extent of inequalities, or about their position in the income distribution. The results show that providing information on inequality has a sizeable impact on people’s perceptions and concerns about inequality, but a rather small effect on their demand for redistribution. Inspecting the heterogeneity across treatments and outcomes helps explaining the small average effect on demand for redistribution, but the evidence is not yet conclusive about the potential explanations. We further show that correcting respondents’ misperceptions about their own position in the income distribution increases the preferences for redistribution for those who previously overestimated their position and decreases it for those who underestimated, although the effects are, on average, small. |
| Keywords: | Meta-analysis, Perceptions of inequality, Preferences for redistribution |
| JEL: | D71 H2 |
| Date: | 2021–11–18 |
| URL: | http://d.repec.org/n?u=RePEc:oec:wiseaa:02-en&r=exp |
| By: | In\'acio B\'o; Li Chen; Rustamdjan Hakimov |
| Abstract: | We consider situations where consumers are aware that a statistical model determines the price of a product based on their observed behavior. Using a novel experiment varying the context similarity between participant data and a product, we find that participants manipulate their responses to a survey about personal characteristics, and manipulation is more successful when the contexts are similar. Moreover, participants demand less privacy, and make less optimal privacy choices when the contexts are less similar. Our findings highlight the importance of data privacy policies in the age of big data, where behavior in seemingly unrelated contexts might affect prices. |
| Date: | 2023–04 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2304.11415&r=exp |
| By: | Ngoc Thao NOET (GRANEM - Groupe de Recherche Angevin en Economie et Management - UA - Université d'Angers - Institut Agro Rennes Angers - Institut Agro - Institut national d'enseignement supérieur pour l'agriculture, l'alimentation et l'environnement); Serge Blondel (LIRAES (URP_ 4470) - Laboratoire Interdisciplinaire de Recherche Appliquée en Economie de la Santé - UPCité - Université Paris Cité, GRANEM - Groupe de Recherche Angevin en Economie et Management - UA - Université d'Angers - Institut Agro Rennes Angers - Institut Agro - Institut national d'enseignement supérieur pour l'agriculture, l'alimentation et l'environnement) |
| Abstract: | Using a contingent valuation approach over 114 subjects, we assess the degree of cooperation and explain the factors underlying ethical choices. Implementing the prisoners' dilemma and public good games reveals a correlation between the amount of the ethical premium and the degree of cooperation. We identify factors that increase this cooperation, such as the frequency of interactions with individuals. The more cooperative individuals are, the higher the ethical premium. In addition, individuals naturally seek to cooperate. The more the game is repeated, the higher the degree of cooperation. According to the items, the degree of contribution is lower when dilution of responsibility occurs. |
| Keywords: | cooperation, ethical values, public good game, prisoners' dilemma |
| Date: | 2023–01–01 |
| URL: | http://d.repec.org/n?u=RePEc:hal:wpaper:hal-04075048&r=exp |
| By: | Offer-Westort, Molly (U of Chicago); Rosenzweig, Leah R. (U of Chicago); Athey, Susan (Stanford U) |
| Abstract: | During a global pandemic, how can we best prompt social media users to demonstrate discernment in sharing information online? We ran a contextual adaptive experiment on Facebook Messenger with users in Kenya and Nigeria and tested 40 combinations of interventions aimed at decreasing intentions to share misinformation while maintaining intentions to share factual posts related to COVID-19. We estimate precise null effects of showing users warning flags or suggesting related articles alongside misleading posts, tactics used by social media platforms. Instead, users share more discerningly when they are given tips for spotting misinformation or are nudged to consider information’s accuracy, reducing misinformation sharing by 7.5% and 4.5% relative to control, respectively. We find significant heterogeneity in response to these treatments across users, indicating tips and the accuracy nudge affect outcomes through separate mechanisms. These low-cost, scalable interventions have the potential to improve the quality of information circulating online. |
| Date: | 2023–01 |
| URL: | http://d.repec.org/n?u=RePEc:ecl:stabus:4072&r=exp |
| By: | Robert M. Gillenkirch (University of Osnabrueck); Julia Ortner (Johannes Gutenberg University Mainz); Sebastian Robert (Rosenheim University of Applied Sciences); Louis Velthuis (Johannes Gutenberg University Mainz) |
| Abstract: | In a sequence of experiments, this study investigates how the design of an advisor’s performance-dependent pay affects a decision-maker’s reliance on advice. In all experiments, the decision-maker forms an initial judgment, receives advice and then makes a final judgment. The advisor’s compensation is manipulated to be fixed, based on individual performance, or based on group performance. We find that performancedependent pay does not affect the decision-maker’s reliance on advice unless performance measurement relates to group performance. Path model analyses show that the advisor’s performance measurement affects the decision-maker’s perceptions of responsibility and power, and that responsibility is the main driver of the decision-maker’s cooperativeness, which mediates the relationship between performance measurement and reliance on advice. In contrast, a decision-maker’s beliefs in the incentive effects of financial compensation on the quality of advice do not drive the results. |
| Keywords: | performance measurement, incentive design, reliance on advice; goal relatedness, incentives beliefs |
| JEL: | D83 D91 M52 |
| Date: | 2023–05–03 |
| URL: | http://d.repec.org/n?u=RePEc:jgu:wpaper:2304&r=exp |
| By: | Barbara Buljat (Université Côte d'Azur, France; GREDEG CNRS) |
| Abstract: | We often read about environmental issues, but we rarely personally witness them. Because of this lack of direct experience, people often perceive environmental threats as events distant in space and time and therefore underestimate their risks. Although direct contact with environmental threats might enhance people's risk perception and engagement, in reality such experiences may be dangerous, costly, and complicated to implement. One strategy to bridge this gap could be communication through virtual immersive experiences. This meta-research identifies the key psychological factors that lead to voluntary pro-environmental behavior (PEB) that have been investigated in previous studies using virtual experiments. After a systematic review of the existing literature, we conclude that immersive virtual experiences can influence some of the psychological factors that are important predictors of pro-environmental behavior, namely, concern, risk perception, connectedness to nature, intrinsic motivation for pro-environmental behavior, psychological distance, and presence. This work makes an academic and a practical contribution. It provides a foundation for policy makers and environmental communicators who want to enhance their campaigns with immersive storytelling, and for researchers who want to enrich and contextualize laboratory experiments that test environmental behaviors and risk preferences. |
| Keywords: | Pro-environmental behavior (PEB), immersive technology, virtual experiments, meta research |
| Date: | 2022–05 |
| URL: | http://d.repec.org/n?u=RePEc:gre:wpaper:2022-15&r=exp |
| By: | Narang, Unnati (U of Illinois at Urbana-Champaign); Shankar, Venkatesh (Texas A&M U); Narayanan, Sridhar (Stanford U) |
| Abstract: | How does a failure in a retailer's mobile app impact shoppers' purchases in its online (website and app) and offline (brick-and-mortar store) channels? Our main hypothesis is that an app failure has a negative effect on offline purchases and no effect on online purchases--after an app failure, shoppers who use the app primarily to search for information but make purchases mostly in the offline channels (search users) engage in search abandonment and forego or reduce their offline purchases; shoppers who use the app mainly to make purchases (checkout users) and online shoppers who use the app for search but purchase on the retailer's website as well engage in channel switching, i.e., switch their intended action to the website as it is just a click away. An alternative hypothesis is that an app failure has no effect on purchases because shoppers engage in temporal substitution, i.e., simply delay their purchase and in channel switching for both search and purchase. It is challenging to empirically examine these hypotheses and estimate the causal effects of an app failure because field experiments inducing such failures are infeasible. We use a natural experiment, an exogenous two-hour failure in a large offline-dominant omnichannel retailer's mobile app, to examine the effects of app failure on purchases in each channel. The results are consistent with our main hypothesis. An app failure has a significant overall negative effect on shoppers' frequency, quantity, and monetary value of purchases across channels. This decrease in purchases after an app failure is driven by lower purchases in offline stores and not in the online channel, consistent with search abandonment and channel switching. Finally, an app failure has a smaller (larger) negative effect on the purchases of shoppers who spent more in the past (purchased more recently, searched for products on the app in the past) with the retailer. |
| Date: | 2022–10 |
| URL: | http://d.repec.org/n?u=RePEc:ecl:stabus:4053&r=exp |
| By: | Alexander De Juan; Paul Hofman; Carlo Koos |
| Abstract: | Aid beneficiaries know very little about development interventions in their own communities. This lack of transparency and information is likely to reduce beneficiaries' ability and willingness to become active in local development. It may also dampen intended aid effects on beneficiaries' political and social attitudes. Can targeted information campaigns strengthen beneficiaries' understanding of aid projects? We test the effects of two types of interventions: the provision of information only and the combination of information and feedback opportunities. |
| Keywords: | Development aid, Information, Fragile states, Randomized controlled trial |
| Date: | 2023 |
| URL: | http://d.repec.org/n?u=RePEc:unu:wpaper:wp-2023-57&r=exp |
| By: | Carranza, Aldo Gael (Stanford U); Krishnamurthy, Sanath Kumar (Stanford U); Athey, Susan (Stanford U) |
| Abstract: | Contextual bandit algorithms often estimate reward models to inform decision-making. However, true rewards can contain action- independent redundancies that are not relevant for decision-making. We show it is more data- efficient to estimate any function that explains the reward differences between actions, that is, the treatment effects. Motivated by this obser- vation, building on recent work on oracle-based bandit algorithms, we provide the first reduction of contextual bandits to general-purpose hetero- geneous treatment effect estimation, and we de- sign a simple and computationally efficient algo- rithm based on this reduction. Our theoretical and experimental results demonstrate that hetero- geneous treatment effect estimation in contextual bandits offers practical advantages over reward estimation, including more efficient model esti- mation and greater flexibility to model misspeci- fication. |
| Date: | 2023–02 |
| URL: | http://d.repec.org/n?u=RePEc:ecl:stabus:4081&r=exp |
| By: | Lihua Lei; Roshni Sahoo; Stefan Wager |
| Abstract: | Practitioners often use data from a randomized controlled trial to learn a treatment assignment policy that can be deployed on a target population. A recurring concern in doing so is that, even if the randomized trial was well-executed (i.e., internal validity holds), the study participants may not represent a random sample of the target population (i.e., external validity fails)--and this may lead to policies that perform suboptimally on the target population. We consider a model where observable attributes can impact sample selection probabilities arbitrarily but the effect of unobservable attributes is bounded by a constant, and we aim to learn policies with the best possible performance guarantees that hold under any sampling bias of this type. In particular, we derive the partial identification result for the worst-case welfare in the presence of sampling bias and show that the optimal max-min, max-min gain, and minimax regret policies depend on both the conditional average treatment effect (CATE) and the conditional value-at-risk (CVaR) of potential outcomes given covariates. To avoid finite-sample inefficiencies of plug-in estimates, we further provide an end-to-end procedure for learning the optimal max-min and max-min gain policies that does not require the separate estimation of nuisance parameters. |
| Date: | 2023–04 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2304.11735&r=exp |