Swiss Political Sci. Rev. 2023;00:1–21. | 1wileyonlinelibrary.com/journal/spsr Received: 14 October 2022 | Revised: 1 October 2023 | Accepted: 16 November 2023 DOI: 10.1111/spsr.12583 O R I G I N A L A R T I C L E Why do people give to their governments? Lab- in- the- field evidence on the role of norms, social information, and political support Raúl López- Pérez1 | Aldo Ramirez- Zamudio2 | Gibrán Cruz- Martínez3 This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. © 2023 The Authors. Swiss Political Science Review published by John Wiley & Sons Ltd on behalf of Swiss Political Science Association. 1Institute of Public Goods and Policies (IPP), Spanish National Research Council (CSIC) 2Center for Economics, Banking and Finance Studies, Department of Economics, Universidad de Lima 3Complutense University of Madrid Correspondence Gibrán Cruz- Martínez, Complutense University of Madrid, Department of Political Science and Public Administration. Campus de Somosaguas s/n, Pozuelo de Alarcón, Madrid, Spain, 28223. Email: gicruz@ucm.es Funding information Instituto de Investigacion Cientifica IDIC, University of Lima Peru; Ministerio de Economía y Competitividad, Grant/Award Number: ECO2017- 82449- P Abstract Although factors leading to selfless acts, such as chari- table donations, have been a central concern in political sciences, voluntary donations are among the most atypi- cal and less well- known public revenue- raising sources. In this article, we explore which factors influence people's donations to their government. We conduct an artefac- tual field experiment in Peru where subjects anonymously decide how much of an endowment they freely donate to the government. We run six sessions with a sample that is representative of the taxpayer population of Metropolitan Lima regarding age, gender, and socioeconomic condi- tions. Our results suggest that donations depend on the subject's support to the government, the average donation by other subjects (social information) and their beliefs about the average donation of others (perceived social norms). K E Y W O R D S Altruism, Beliefs, Donations, Latin America, Public Goods Zusammenfassung Obwohl die Faktoren, die zu selbstlosen Handlungen führen, in der Politikwissenschaft ein zentrales Thema sind, gehören freiwillige Spenden zu den untypischsten und weniger bekannten Quellen öffentlicher Einnahmen. In diesem Artikel untersuchen wir, welche Faktoren die Spenden der Bürger an ihre Regierung beeinflussen. Wir haben ein artefaktisches Feldexperiment in www.wileyonlinelibrary.com/journal/spsr mailto:raul.lopez-perez@cchs.csic.es https://orcid.org/0000-0002-8952-8914 mailto: https://orcid.org/0000-0001-7509-3975 mailto: https://orcid.org/0000-0002-4583-2914 http://creativecommons.org/licenses/by-nc-nd/4.0/ mailto:gicruz@ucm.es http://crossmark.crossref.org/dialog/?doi=10.1111%2Fspsr.12583&domain=pdf&date_stamp=2023-12-20 2 | WHY DO PEOPLE GIVE TO THEIR GOVERNMENTS? Peru durchgeführt, bei dem die Individuum anonym entscheiden, wie viel sie freiwillig an die Regierung spenden. Die Stichprobe ist in Bezug auf Alter, Geschlecht und sozioökonomische Bedingungen repräsentativ für die steuerzahlende Bevölkerung der Metropole Lima. Unsere Ergebnisse deuten darauf hin, dass die Spenden von der Unterstützung der Testpersonen für die Regierung, von der durchschnittlichen Spende anderer Testpersonen (soziale Informationen) und von ihren Überzeugungen über die durchschnittliche Spende anderer (wahrgenommene soziale Normen) abhängen. Résumé Bien que les facteurs conduisant à des actes altruistes, tels que les dons caritatifs, aient été une préoccupation centrale en sciences politiques, les dons volontaires figurent parmi les sources de revenus publics les plus atypiques et moins connues. Dans cet article, nous explorons les facteurs qui influent sur les dons des individus à leur gouvernement. Nous menons une expérience de terrain artéfactuelle au Pérou, où les sujets décident anonymement de la quantité de leur dotation qu'ils veulent librement donner au gouvernement. Nous réalisons six sessions avec un échantillon représentatif de la population contribuable de la région métropolitaine de Lima en ce qui concerne l'âge, le sexe et les conditions socio- économiques. Nos résultats suggèrent que les dons dépendent du soutien du sujet au gouvernement, de la donation moyenne des autres sujets (informations sociales) et de leurs croyances quant à la donation moyenne des autres (normes sociales perçues). Riassunto Sebbene i fattori che portano ad atti altruistici, come le donazioni a scopi di beneficenza, siano stati al centro delle preoccupazioni della scienza politica, le donazioni volontarie sono tuttora tra le fonti di raccolta di entrate pubbliche più atipiche e meno conosciute. In questo articolo esploriamo i fattori che influenzano la decisione dei cittadini e delle cittadine di fare donazioni al loro proprio governo. L'analisi empirica si basa su un esperimento di campo artefatto condotto in Perù, in cui i soggetti potevano decidere liberamente e in forma anonima la quota di un lascito da donare al governo. Abbiamo condotto sei sessioni con un campione della popolazione dei contribuenti di Lima metropolitana, rappresentativo in termini di età, genere e condizioni socieconomiche. I nostri risultati rivelano che quanto una persona decide di donare dipende dal suo suo sostegno al governo, dalla donazione media delle altre persone (informazione sociale) nonché dalla percezione circa la donazione media delle altre persone (norme sociali percepite). 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense | 3LÓPEZ- PÉREZ et al. INTRODUCTION Political and social science literature has examined voluntary donations to political parties (Phillips, 2012; Gilens, 2015), charities (Kriesi & Baglioni, 2003; Margolis & Sances, 2017), and public goods provision (Chaudhuri, 2011). However, voluntary donations to the government have largely been neglected in the political science literature and are possibly among the most atypical and less well- known revenue- raising sources (Li et al., 2015). Further, their determi- nants are still unclear. This research explores whether people donate to their government and which factors might influence their donations. In particular, we pay attention to the individual perceptions about (i) the average behavior of others and (ii) the efficiency and competency of the government. We are interested in how perceived social norms (beliefs about the average donation of others), descriptive social information (the real information provided regarding the average donation of their reference group), and political support and quality of government affect giving to the government. Our research is informed by literature on political behavior, norms, and ex- periments on donations to charities and government (Cialdini et al., 1990; Bicchieri, 2005; Li et al., 2011, 2015; Hellmüller et al., 2020; Ozdemir & Jacob, 2022). Voluntary giving to the government refers to an intentional donation that someone gives to a government via the National Treasury or income taxes checkboxes. These donations can be used to fund the functioning of the government or specific actions and policies of a public organization. In this aspect, they differ from donations to a political party, which are a way of supporting that party's policies, campaigns, and candidates. Donations to the government exploit the willingness of certain individuals to pay more taxes or contribute more than they currently do. But does anyone voluntarily give money to the government? In short, the answer is yes. Historically prominent examples appear in times of war or national emergencies, such as nat- ural disasters or terrorist attacks, when citizens' donations to their governments are arguably a way of capitalizing on patriotism (Slemrod & Kuchumova, 2023). For a perhaps more cur- rently appropriate example, citizens in the United States (US) can donate to their government via contributions to an account called “Gifts to the United States.” Similarly, in many US states, income tax forms have “check- off” programs enabling taxpayers to contribute to spe- cific causes (see Slemrod, 2003; Li et al., 2011). In Spain, taxpayers can renounce their total income tax refund by checking the corresponding box in the form. In 2016, for instance, around 52,000 taxpayers did so, and the Spanish tax agency increased its revenues by one million Euros (out of total receipts of 186,249 million).1 In short, donations to the government exist. At the same time, it is crucial to recognize that they are not a panacea for fiscal space creation and, most relevantly, that they are at the present a minuscule part of the public budget. For example, Slemrod and Kuchumova (2023) calculated that for 2019 gifts totaled 4.99 million US dollars, i.e., 0.0006% of federal income tax collections – compare this with giving to private charities, which totaled $295 billion in 2006, or 2.2% of the US Gross Domestic Product (Li et al., 2011). Our conjecture however, is that donations to the government could be potentially increased (if necessary), although that requires a better understanding of their potential drivers. Why would voters donate to the government, given that they already pay taxes? From our point of view, the most likely explanation (indeed consistent with our experimental find- ings) resorts to non- selfish, altruistic motivations. In short, some people think they should 1See Ayuso (2018). The motivation behind these donations is uncertain. While we believe that altruism is a major motive, mistakes in checking the appropriate box might also operate, although the fact that most individual donations range between 13 and 18 Euros suggests some intentionality. Alternatively, some taxpayers might (wrongly) believe that renouncing to the refund prevents any posterior audit. 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense 4 | WHY DO PEOPLE GIVE TO THEIR GOVERNMENTS? contribute further to (some of) the goods and services offered by their government. This is, however, a too generic statement, and our research aims to shed further light on those motivations. Specifically, we hypothesize that social norms (Cialdini et al., 1990; Cialdini & Trost, 1998; Bicchieri, 2005; López- Pérez, 2008) are a key driver of donations. We consider both injunctive and descriptive norms, the latter being defined as a behavior that is (perceived to be) “normal” or modal (Cialdini et al., 1990). In more detail and based on López- Pérez (2008), we implic- itly propose two stages of a normative decision. In the first one (injunctive part), people infer how they ought to act in a particular context based on distributive justice arguments. For instance, a person might find it advisable to choose an act that promotes equity, i.e., reducing ‘undeserved’ or unfair differences, or/and social efficiency, i.e., the size of the cake; people can differ in the weights they give to each component. Having determined the normative action X, second, people decide whether to choose X or the payoff- maximizing choice (which can be different from X). For this, they consider both (a) the eventual loss in material payoff from choosing X (i.e., the ‘price’ of following the norm) and (b) the expectation of how many others in the reference group will do X as well –intuitively, if most others do not “do their part,” the person feels less remorse for not choosing X. To test several hypotheses derived from the theoretical framework just described (see more detail in Section 2), we run a Tobit regression analysis with data collected from a sim- ple artefactual field experiment in Peru (Druckman & Green,  2021). Each subject is en- dowed with 30 Soles (around $10 US dollars) and can voluntarily and anonymously donate some of this endowment to the Peruvian government, specifically the Peruvian Public Treasury.2 The first set of sessions, called T1 henceforth, was run when former Peruvian President Kuczynski enjoyed relatively high levels of popularity. In contrast, his popularity was substantially lower one year later when the second group of sessions, called T2, was run.3 At the same time T2 sessions were being held, we ran additional sessions with an INFO treatment (T2- INFO henceforth) in which subjects were informed ex- ante about the average donation in T1. The experiment incorporates at least two innovations to the ones we find in the literature on donations to governments. First, we run the experiment in Peru, a middle- income country. Second, subjects were not undergraduate students but a represen- tative sample of the taxpayer population of Metropolitan Lima regarding quotas of age, gender, and socioeconomic conditions (see Online Appendix 2).4 A substantial share of our subjects gives something (around ¾ of them; the average dona- tion across treatments amounts to 14% of the endowment). Hence, our results show that the act of giving voluntary donations applies beyond the sample of college students who have been the traditional segment of the population tested in prior experimental studies. We find that giving is correlated with several variables. First, a regression analysis shows that a subject's belief about the average donation by her peers is significantly correlated with her donation. Second, the INFO treatment impacts the distribution of donations to the government, suggesting the causality of subjects' beliefs about peers' average behavior and the role of social information. 2Our design resembles that of Eckel and Grossman (1996), although with a different focus: they showed that using a well- known charity as a recipient rather that an anonymous subject increases donation in a dictator game. 3His support at the time of T1 in November 2016 was 51% but decreased to 27% by the time of T2 in November 2017 (IPSOS, 2017). By March 21st, 2018 President Kuczynski resigned his presidency after being involved in a vote- buying scandal. 4This arguably makes our experiment an artefactual field experiment as defined by Harrison and List (2004, p. 1014), i.e., “the same as a conventional lab experiment but with a nonstandard subject pool”. Note also that subjects in our experiment make donations with real consequences, not hypothetical ones as in a conventional survey asking about citizens' government donations. This is relevant because hypothetical and real donation levels may differ substantially, for multiple reasons, e.g., survey respondents might tend to overestimate what they would actually give. In this sense, therefore, our design adds realism and external validity. 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense | 5LÓPEZ- PÉREZ et al. In addition, donations co- move with the subject's stated level of support for the current presi- dent (although the variable becomes marginally significant when we add having children as a control). As we explain later, these findings align well with our theoretical framework. The rest of the paper proceeds as follows. The following section provides some theoretical expectations on donations to the government, from which hypotheses are derived. The third section describes the experimental design and procedures. The hypotheses are tested in the fourth section, which reports the experimental data. The last section discusses our main con- tributions and potential policy implications. EXPLA IN ING TH E ACT OF GIVING: TH E DEBATE IN TH E LITERATURE When and to what extent do people give to their governments? Recent literature finds that subjects donate with different intensities depending on the specific public recipient, its aim, and other factors. Using a design with a series of donation decisions, Li et al. (2011) compare giving to specific US government agencies and private charities with similar missions (e.g., Cancer Research or Parks and Wildlife). Subjects were undergraduate students and donated on average 22% of their endowment to government agencies, significantly lower than the 27% to private charities (see also Jones, 2017, and Luccasen & Thomas, 2020). Li et al. (2011) report that organizations are more likely to attract funds if they are perceived as serving an impor- tant cause, being trustworthy and efficient, and providing a high quality of service. In turn, Li et al. (2015) use a similar setting to compare giving to the US federal general revenue fund with directed giving to government organizations focused on (a) disaster relief and (b) cancer prevention and research. The average donation to the general fund is 8.4% of the endowment, whereas directed giving more than doubles the size of contributions, a significant difference. The novelties of our research in comparison to the existing experimental literature on giving to governments are that (i) the subjects in our experiment were not undergraduate students but a representative sample of the taxpayer population, and (ii) we explore the role of descriptive norms, social information, and political support in explaining people's donations. Concerning (ii), we consider five hypotheses detailed below. We first consider norms, defined here as “rules and standards that are understood by mem- bers of a group, and that guide and/or constrain social behavior without the force of law” (Cialdini & Trost,  1998, p. 152). The relevance of social interactions, normative standards, and trust in forming actions and behaviors has been stressed in the literature (Uslaner, 2000; Bicchieri,  2005; Rothstein & Uslaner,  2005; Hoffman,  2007; Fieldhouse & Cutts,  2021; Fieldhouse et al., 2022). In particular, giving is likely to be influenced by personal psychology but also by the behavior of the subject's reference group, a point related to the concept of de- scriptive norms, i.e., what is (perceived to be) “normal” or what most people do in a particular context (Cialdini et al., 1990). In this regard, abundant field evidence (Frey & Meier, 2004; Shang & Croson, 2009) and lab evidence (Hansen et al., 2014) confirms that social comparisons influence donations to chari- ties. Social effects have also been observed in experimental studies on the detrimental impact of a disaster (Becchetti et  al.,  2017), deception (see Gino et  al.,  2009; López- Pérez & Spiegelman, 2013), and bargaining games (see Bohnet & Zeckhauser, 2004) on the preferences of giving. Motivated by previously mentioned research and the theory sketched in the intro- duction (see López- Pérez, 2008, for more details), our first hypothesis predicts a positive cor- relation between donations and beliefs about average donation.5 Note that we assume that 5In addition, beliefs about average behavior could affect donations if the (expected) donations by others signal something about the quality of the public goods provided by the government (e.g., Vesterlund, 2003). This is compatible with the theory tested here, if one assumes incomplete information about the value of the public good. 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense 6 | WHY DO PEOPLE GIVE TO THEIR GOVERNMENTS? subjects compare with each other so that a subject's reference group contains all the other participants in the session. H1 (beliefs): The amount donated increases the higher the subject's belief of the reference group donation. Hypothesis 1 refers to expectations. At the same time, it follows that providing informa- tion about the average donation of peers may impact human behavior (although the direc- tion and extent of the effect should depend on whether individuals under- or over- estimate average donations). Experimental social scientists have measured the impact of giving in- formation on charity giving with mixed results. For instance, Croson et al. (2009) show that descriptive social information affects giving. In contrast, Van Teunenbroek et al. (2021) use online experiments in the United Kingdom and find no effect of descriptive social infor- mation on charitable giving. Our study contributes to the research on social information, adding a middle- income Latin American country and measuring giving to the govern- ment via donations rather than the more studied charities option. The next hypothesis is straightforward. H2 (social information): The distribution of donations to the government changes if subjects receive information about peers' average behavior. A final set of hypotheses follows from our assumption that norms derive from a con- sequentialist logic or at least have some outcome- oriented flavor (López- Pérez,  2008). Suppose, for instance, that people consider both equity and social efficiency in their norma- tive inferences, i.e., they think that actions ought to be chosen to maximize some social wel- fare function, including both variables. In that case, the normative donation is conditional on expectations of efficiency and deservingness. For instance, if a person believes that the government is very inefficient and corrupt, such a norm would prescribe no giving; in this scenario, the money is better kept in the subject's pocket. In other words, the same norm selects different acts depending on the subject's idiosyncratic perceptions about the efficacy and deservingness of the government. The subsequent three hypotheses follow from the argument just described. H3: The amount donated increases the higher the subject's political support of the government. H4: The amount donated increases the lower the subject's perception of government corruption. H5: The amount donated increases the higher the subject's perception that the government works for the public good. Although our regression analysis below will test hypotheses H3 to H5, we note as an aside that an extensive literature on tax compliance hints that citizens' trust in their governments, as well as their perceptions about the extent of public waste and corruption, affect the decision to evade taxes (Taliercio, 2004; Torgler, 2005; Bodea & LeBas, 2016; Prichard, 2018; Dularif & Rustiarini, 2021; Kouamé, 2021). This is perhaps relevant, as norms might be a relevant part of tax morale, in which case our research on giving to the government could also be relevant for tax compliance, at least from a behavioral perspective. EXPERIM ENTA L DESIGN A N D PROCEDU RES To test the determinants of giving to the government, we first collect data using several arte- factual field experiments with different treatment groups and then conduct Tobit regression analysis with the data. We consider a simple, one- shot decision problem where each subject is endowed with 30 Soles and can voluntarily donate some of this endowment to the Peruvian government.6 The 6We note that the public sector in Peru is relatively small, as government spending averaged around 16.5% of total output (GDP) over the years 2014 to 2017 (source: Central Bank of Peru). 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense | 7LÓPEZ- PÉREZ et al. donation is implemented by means of an actual bank deposit to an account of the Peruvian Public Treasury (Banco de la Nación account number 00000–299294), made anonymously by two of the experimenters after all participants have finished their choices (with some subjects acting as witnesses, as standard in donation experiments, e.g., Eckel & Grossman, 1996). Any subject's payoff equals the initial endowment minus the donation plus a 20 soles (around US$ 7) show- up fee. Each session was conducted as follows. Before it started, the instructions and a decision sheet were distributed in conveniently separated seats across a room to avoid communication between subjects. Then, every subject entered the room and chose one of those seats. They first read the instructions at their own pace; later, the experimenter read them aloud to promote common information.7 Questions were privately clarified. All decisions were taken with pencil and paper. Any subject was identified by a unique ID number included in her decision sheet. The experimenter noted verbally while reading aloud the instructions that the Public Treasury would use the subject's donation to finance similar expenditures and public projects as taxes do. Subjects were also informed in this manner about the Banco de la Nación account number mentioned above, writing that number as well on a blackboard. The instructions re- called subjects that the Peruvian government offers different public services. While informing subjects about the goals of the institutions they can donate to is far from unusual in the liter- ature (e.g., Li et al., 2011; Li et al., 2015), one must consider this aspect when drawing conclu- sions outside this context. For instance, a more negative statement mentioning, say, corruption or waste in the public sector could have motivated few giving. Our precise framing seems to induce no confounds concerning our research questions. Moreover, clearly stating that the donations are paid to the government's treasury increases the external validity of the findings, as participants are aware that their decision will have real- world consequences. When subjects had decided on their donation, decision sheets were collected, and an elicita- tion sheet was given. Here we elicited some estimates designed to test the five hypotheses men- tioned above. In particular, we asked each subject to estimate the average donation among all participants in the session.8 After all subjects had their estimates elicited, we collected the corresponding sheet. The experiment ended with the completion of a brief questionnaire on socio- demographics, frequency of use of public services, support for the current presidential team, concern for inequity, etc. Many of these questions appear in similar terms in the World Values Survey. Subjects were then paid in private by an assistant who knew only the subject's ID number and her final payoff – this means that the assistant was not informed about the details of the experiment.9 To further confidence in our procedures, subjects were told that they would be asked to vol- unteer as witnesses at the end of the experiment. If there were no volunteers, two participants would be randomly chosen. After all subjects had been paid, these witnesses checked the de- cision sheets and recorded the sum of all individual donations. Afterward, the experimenters and the witnesses went to the bank office in the mall in front of the University campus, where an anonymous deposit was made for the total amount donated. We run four sessions of the described T1 and T2 at the University of Lima. The first two had 30 participants each and took place respectively on October 29 and November 12, 2016. 7The translated instructions, decision forms and questionnaires can be found in the online appendix. 8These estimates were not incentivized. López- Pérez and Ramírez- Zamudio (2020) study two policies to increase donations. In the control treatment, subjects faced the same decision problem as in our T1 and T2 here but received a prize for the accuracy of their estimates. Comparing the relevant regressions in both studies, we observe that the significance and strength of the correlation between beliefs and giving is remarkably similar. We conclude therefore that the lack of incentives is not the cause of the correlation reported here. In any case, the INFO treatment provides further evidence of the role of social comparisons on giving. 9Following a usual practice in experiments on giving, we therefore chose a design in which subjects' perceived degree of scrutiny of their choices was most likely very low, if not nil. It is yet an open question whether the social influence effects that we observe are conditional on the degree of experimenter's scrutiny. 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense 8 | WHY DO PEOPLE GIVE TO THEIR GOVERNMENTS? In what follows, we pool this data (N = 60) under T1. Then we ran two sessions with 25 sub- jects each, and both took place on November 4, 2017; we call this T2 (N = 50) henceforth. The time gap between T1 and T2 helped us obtain additional insights into the subjects' motives for giving. Subjects were always between 23 and 59 years old and economically ac- tive. In T1, they were selected by IMASEN following precise instructions,10 so that the random sample was representative of the taxpayer population of Metropolitan Lima re- garding age, gender, and socioeconomic conditions. University of Lima's market research department selected the participants for T2 with a similar methodology. Recruiters never disclosed details about the experiment to the subjects, except that this was a “focus group” meeting about topics like government, institutions, and other social issues. Each session lasted approximately 90 minutes, including paying the subjects individually. The average payoff in T1 and T2 was 45.33 Soles and 46.86 Soles, respectively, including the show- up fee of 20 soles. Our design also includes one treatment called T2- INFO, a slight variation of T1 and T2, as we had in the decision form the rounded average donation made by the participants in T1. This T2- INFO treatment consisted of 54 subjects separated into two classrooms with 27 people each; each of these sessions was run in parallel to each of the two sessions of T2 but in different classrooms.11 Subjects were randomly assigned to the T1, T2, and T2- INFO following quotas by gender, age, and socioeconomic status to have a representative sample. The 54 participants in T2- INFO earned, on average, a total payoff of 45.30 Soles. No subject attended more than one session or treatment.12 It must be noted that providing information on others' prior choices is not an unusual manipulation in the literature. In the sequential dictator game of Cason and Mui  (1998), subjects act first as dictators, learn the dictator decision of another subject, and then make a second dictator decision. In ultimatum games, Bohnet and Zeckhauser (2004) report that the size of the offer and the probability of rejections vary when responders are told the average offer received by others. Bicchieri and Xiao  (2009) inform participants in their dictator games about the majoritarian choice in prior dictator games. As in any lab study, experimenter demand effects are a potential concern in these studies (Zizzo, 2010): if sub- jects guess that the experimenter wants them to alter their behavior in response to the infor- mation provided, they might behave accordingly, maybe out of altruism or conformity with authority. Still, such effects should be stronger in repeated games, and hence of a relatively reduced size when there is little perceived scrutiny and participants are extremely unlikely to have any future interactions with the experimenter, as in our sample pool. Theoretically, however, demand effects can also appear in one- shot games. Our instructions attempted to diminish them by stating that subjects should choose as they preferred. A potential moti- vation by any subject to behave to ‘please’ the experimenters arguably put no constraints on her choice. 10IMASEN is a Peruvian research- based consulting company, well- known for its market studies, surveys and polls: http:// www. imase nperu. com/ . Subjects had to be literate adults, could not be another participant's relative, and were asked not to come accompanied by other people. See the online appendix 2 for more details on the methodology as well as further argument of the representativeness of our sample. 11The questionnaire in T2- INFO and T2 was a small variation of the one we used in T1, as we elicited the subjects' beliefs about the eventual donations of members of some of their daily- life reference groups, e.g. family, co- workers, classmates, neighbors, close friends and even members of the same church if applicable. The full questionnaire is available upon request. 12We used the methodology of List et al. (2011) to determine the optimum sample size. The analysis indicated an optimal size of 51 for the treatment and 31 for control (results available upon request). Since we obtained enough resources to finance more subjects, we formed slightly bigger groups. 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense http://www.imasenperu.com/ http://www.imasenperu.com/ | 9LÓPEZ- PÉREZ et al. DATA A NA LYSIS This section starts with a summary and discussion of the subjects' decisions in each treatment. We later study, partly by means of a regression analysis, what variables affect donations to the government across the variety of treatment groups. Summary of Results Table 1 describes the distribution of donations in each group; we distinguish between T1 and T2 throughout the paper because of significant differences across these groups. We observe the highest average donations in T1 and T2- INFO, whereas the lowest is in T2. We also find differences across groups in the distribution of donations, e.g., the fraction of subjects who donate less than 5 Soles is above 60% in T2 but below 40% in T2- INFO and T1.13 For further detail, Figure 1 below depicts the distribution of donations in each experimental group – we introduce five ‘intervals’ for the donations; the left- hand bar in each graph indicates the frequency of nil donations. Perhaps the most noticeable difference is that the distribution in T2 has a clear mode on zero donations, i.e., around half of the participants decided to do- nate nothing to the Peruvian treasury in T2. There exist few differences in socio- demographic characteristics across groups. Table  2 below provides information about the average participant in each group and her responses to some questions (see the questionnaire in the online appendix for more detail). For instance, Corruption indicates the position where that subject places Peru in the Transparency International ranking out of the 168 countries analyzed in 2015. Clearly, the average subject in all groups believes that Peru is a rather corrupt country. On a different issue, when we perform all pairwise comparisons of the distributions of gender, political ideology, socioeconomic sta- tus, and religiosity across the three groups, we observe no significant differences except in the distribution of ideology across T1 and T2.14 Further, the distribution of status across T2- INFO and T2 is also different.15 While we cannot reject the possibility that these differences might have induced behavioral variations across groups (although our regression analysis below sug- gests otherwise), they do not seem to compromise our main findings. We provide more details later. Note also that the average subject in T2- INFO expects a rather high average donation by the other subjects in her group (a point we discuss further below); in regard to the median sub- ject, however, there is no significant difference between T2- INFO and T1 (median test; p- value = 0.715), and a significant difference between T2 and any of the other two groups (p- value ≤0.004 always). What affects giving: A regression analysis Testing Hypothesis 1: the role of beliefs and descriptive norms Model 1 in Table  3 pools the data from T1 and T2 and includes the variable related to Hypothesis 1 (i.e., beliefs about the average donation of others). As we observe, this hy- pothesis is largely vindicated by the data. In effect, this variable's estimated coefficient is positive and significant at the 1% level. Therefore, we have strong evidence that the belief 13A Kolmogorov–Smirnov test indeed finds a significant distributional difference when comparing T2 to any other experimental group (p ≤ 0.012 always). In contrast, the test does not report a significant difference in the comparison between T1 and INFO. 14Two- sample Mann–Whitney test; p = 0.043. 15p = 0.040; we note, though, that a Kolmogorov–Smirnov test does not report significant differences across conditions. 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense 10 | WHY DO PEOPLE GIVE TO THEIR GOVERNMENTS? T A B L E 1 D es cr ip ti ve s ta ti st ic s of e ac h tr ea tm en t an d se ss io n. T re at m en ts a nd s es si on s N um be r of su bj ec ts A gg re ga te a ve ra ge d on at io n (s ol es ) R el at iv e av er ag e do na ti on (% o f en do w m en t) S ta nd ar d de vi at io n S ub je ct s by th e in te rv al o f d on at io n (% in p ar en th es es ) 0 [1 ,4 ] [5 ,9 ] [1 0, 15 ] [1 6, 30 ] T 1 60 4. 67 15 .5 7 4. 67 10 20 18 10 2 (1 6. 7) (3 3. 3) (3 0. 0) (1 6. 7) (3 .3 ) T 2 50 3. 14 10 .4 7 5. 94 23 13 10 1 3 (4 6. 0) (2 6. 0) (2 0. 0) (2 .0 ) (6 .0 ) T 2- IN F O 54 4. 7 15 .6 7 4. 66 9 12 25 7 1 (1 6. 7) (2 2. 2) (4 6. 3) (1 3. 0) (1 .9 ) T 1 + T 2 + T 2- IN F O 16 4 4. 21 14 .0 3 5. 1 42 45 53 18 6 (2 5. 6) (2 7. 4) (3 2. 3) (1 1. 0) (3 .7 ) N ot e: T 1 = fi el d w av e nu m b er 1 (w it ho ut t re at m en t) b et w ee n 10 .2 01 6 an d 11 .2 01 6; T 2 = fi el d w av e nu m b er 2 (w it ho ut t re at m en t ot he r th an t im e d if fe re nc e) o n 11 .2 01 7; T 2- IN F O = fi el d w av e nu m b er 2 (w it h in fo t re at m en t of t he a ve ra ge d on at io n m ad e in T 1) o n 11 .2 01 7. 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense | 11LÓPEZ- PÉREZ et al. about what fellow citizens give to the government, a descriptive norm, positively affects their own giving behavior. Note that the significant effects of beliefs are robust to the in- clusion of socio- demographic controls, i.e., age, gender, socio- economic status, education level, political ideology, religiosity, having children, and car ownership (YES/NO); see more details below. While beliefs and donations positively correlate in all sessions and treatments consid- ered, the reader can also perceive that such a correlation is far from perfect: donations are often smaller than beliefs. In other words, subjects tend to give to the government less than they believe others give. Indeed, the corresponding coefficient in Model 1 indicates that an increase in the beliefs in one Sol leads to an increase in the donation of around 0.51 Soles. Although not part of Hypothesis 1, we must also note that the effect of beliefs on giving is less strong in the T2- INFO, as confirmed by a regression analysis focused on this treatment (results available upon request; also compare the coefficients of Variable 1 in Models 1 and 2, Table 3). Result 1 (beliefs and donations): Donations co- move with beliefs about the average donation in the reference group, and the relation is highly significant. Therefore, H1 is supported. A problem in the previous analysis is that the correlation between estimates and donations can be spurious. A potential reason is the so- called false consensus effect, which captures the tendency of an individual to think that others are similar to her (Ross et al., 1977). Donations might not be affected by the subject's beliefs/estimates yet be co- linear with them, just because people tend to think that others are like themselves and hence donate similar amounts. We believe our results are not driven by the false consensus (at least entirely) for several reasons. To start, we repeat that donations are systematically lower than the estimates: Subjects tend to believe that others give more (see Table 3). More substantially, though, the results from T2- INFO, which we conducted to explore how social information affects giving, are at odds with the idea that beliefs are irrelevant for choice. F I G U R E 1 Distribution of donations in each experimental group. 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense 12 | WHY DO PEOPLE GIVE TO THEIR GOVERNMENTS? Testing the social information hypothesis: The effect of the T2- INFO treatment Recall that subjects in T2- INFO were informed in the donation sheet – that is, before choosing – about the (rounded) average donation to the government in T1 (i.e., 5 Soles; the actual aver- age was 4.67 Soles). This treatment allows us to test H2. That is, the distribution of donations in T2- INFO and T2 should be statistically different if beliefs affect behavior, other things equal.16 In this respect, recall from the previous section that a two- sample Kolmogorov– Smirnov test indicates that the two donation distributions are statistically different (p ≈ 0.001). This is, therefore, first evidence that social information affects giving to the government. Note also that despite the dramatic decrease in donations in T2 compared to T1, donations in T2- INFO (4.67 Soles) are comparable to those of T1. It seems to suggest that the positive impact of social information offsets the drop in donations due to the declining political support for the president (see below). Model 2 in Table 3 above pools T2 and T2- INFO data to further explore the social infor- mation effect. On one hand, we can see that the treatment dummy is (marginally) significant (p- value = 0.07). Note also that some of the effects might operate via higher expectations about average giving (Variable 1). Indeed, as we saw above, the mean expectation in T2- INFO is substantially higher than that in T2 (the median belief, in turn, equals 5 and 8.75 Soles in T2 and T2- INFO, respectively). Incidentally, the difference in beliefs across T2 and 16The last proviso indicates that the proper comparison is between T2 and T2- INFO session, as both were run at the same time. In contrast, T1 and T2- INFO session were run with a year of difference, and a significant variable like the support for the president changed during that time. TA B L E 2 Descriptive statistics of the subject pool in each experimental group. T2- INFO T1 T2 Mean Mean Mean 1. Corruption (1: least - 168: most) 125.56 124.53 136.56 2. Estimated average donation of others 11.18*** 5.85 6.49** 3. political ideology 5.00*** 6.07* 5.40 4. Support the current president 3.85*** 6.25** 4.92** 5. Level of religiosity 5.20 5.42 5.68 6. Children 0.78 0.73 0.82 7. Education level 3.83 3.60 3.74 8. Socioeconomic level 2.94 3.00* 3.32** 9. Gender 0.44 0.48 0.52 10. Age (years) 42.68** 38.36 40.20 11. Car ownership 0.05** 0.20 0.32*** N 54 60 50 Note: Political ideology is measured from 0 (extreme left) to 10 (extreme right); support to president from 1: not at all to 10: entirely; religiosity from 1: least to 10: most; Children is a dummy, i.e., 0 = no children; Education level from 1: Incomplete primary school to 5: University higher education; Socio- economic level from 1: lowest to 5; highest; Gender: 0 = male; car ownership is a binary variable (NO = 0). The number of respondents is slightly different across cells, as some subjects left empty some questions. Symbols ***, **, and * indicate a significant difference at 1%, 5%, and 10% levels, respectively, of a test of equality of means of each variable across treatments. The stars in columns (i) T2- INFO, (ii) T1, and (iii) T2, specifically, compare treatments (i) T2- INFO and T1, (ii) T1 and T2, and (iii) T2- INFO and T2, respectively. T1 = field wave number 1 (without treatment) between 10.2016 and 11.2016; T2 = field wave number 2 (without treatment other than time difference) on 11.2017; T2- INFO = field wave number 2 (with info treatment of the average donation made in T1) on 11.2017. 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense | 13LÓPEZ- PÉREZ et al. T A B L E 3 R eg re ss io n an al ys is o f de te rm in an ts o f do na ti on t o th e go ve rn m en t. D ep en de nt v ar ia bl e: D on at io n ce ns or ed f ro m b el ow a t z er o In de pe nd en t v ar ia bl e M od el 1 ( T 1 + T 2) M od el 2 ( T 2 + T 2- IN F O ) M od el 3 ( T 1 + T 2 + T 2- IN F O ) 1. B el ie fs a bo ut t he a ve ra ge d on at io n of o th er s 0. 51 33 ** * 0. 30 3* ** 0. 31 57 ** * (0 .1 19 ) (0 .0 84 ) (0 .0 67 ) 2. T re at m en t du m m y − 2. 98 7* * 2. 69 5* (1 .3 20 ) (1 .4 70 ) 3. S up po rt t o cu rr en t pr es id en t (0 : n ot a t al l, 10 : e nt ir el y) 0. 58 4* ** 0. 68 7* * 0. 57 8* * (0 .2 18 ) (0 .2 95 ) (0 .1 72 ) 4. C or ru pt io n (1 : l ea st - 1 68 : m os t) − 0. 01 63 − 0. 00 5 − 0. 01 17 (0 .0 14 ) (0 .0 16 7) (0 .0 11 ) 5. G ov er n m en t is c on tr ol le d by (0 : f ew in te re st s, 1 0: w or ks o n ly fo r th e p eo pl e) − 0. 11 6 − 0. 01 2 − 0. 01 16 (0 .2 54 ) (0 .3 27 ) (0 .2 02 ) 6. I nt er ce pt − 0. 12 9 − 4. 07 5 − 0. 64 7 (2 .7 10 ) (3 .4 18 ) (2 .0 77 ) Si gm a 6. 05 8 6. 53 6 5. 75 6 (0 .5 14 ) (0 .5 86 ) (0 .3 84 ) O bs . 10 9 10 4 16 3 P se ud o R - s qu ar ed 0. 05 86 0. 04 10 0. 03 56 N ot e: R ob u st s ta nd ar d er ro rs in p ar en th es es . M o de l 1 p oo ls t he d at a fr om T 1 an d T 2, M o de l 2 f ro m T 2 an d T 2- IN F O , a nd M o de l 3 f ro m T 1, T 2, a nd T 2- IN F O . T he d ep en de nt v ar ia bl e is a lw ay s a su bj ec t's d on at io n to t he G ov er n m en t, in S ol es ; g iv en t he c en so re d na tu re o f th is v ar ia bl e, w e ru n T ob it r eg re ss io n s. T he t re at m en t du m m y, i. e. , v ar ia bl e 2, t ak es v al u e 1 in M o de l 1 (2 ) if t he s ub je ct p ar ti ci p at ed in T 2 (T - 2 I N F O ), a nd z er o ot he rw is e, * ** , * *, a nd * in d ic at e si gn if ic an ce a t 1% , 5 % , a nd 1 0% le ve ls , r es p ec ti ve ly . T 1 = fi el d w av e nu m b er 1 (w it ho ut t re at m en t) b et w ee n 10 .2 01 6 an d 11 .2 01 6; T 2 = fi el d w av e nu m b er 2 (w it ho ut t re at m en t ot he r th an t im e d if fe re nc e) o n 11 .2 01 7; T 2- IN F O = fi el d w av e nu m b er 2 (w it h in fo t re at m en t of t he a ve ra ge d on at io n m ad e in T 1) o n 11 .2 01 7. 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense 14 | WHY DO PEOPLE GIVE TO THEIR GOVERNMENTS? T2- INFO suggests that subjects in T2- INFO anticipate the social information effect (and in fact overestimate it). Although our preferred interpretation of our data in T2- INFO alludes to descriptive norms and peer effects, we note that there is at least another possibility. That is, suppose that altruistic/reciprocal subjects in T2- INFO are uncertain about the quality of the public goods provided by the Peruvian government and hence about the efficacy of the prospective donation. In that case, subjects could use the information provided on prior average dona- tions as an indication of the government's efficiency – see Vesterlund (2003) for a similar argument. More generally, beliefs could affect donations if others' donations signal some- thing about quality. But do subjects in T2- INFO learn something about quality when they are told that others donated 5 Soles in a previous session? While a careful analysis of this question is out of the scope of this paper, we can offer some preliminary evidence. For this, we compare responses across T2 and T2- INFO to several questions that provide informa- tion about subjects' perceptions of the government's performance. These are the ‘waste’ and ‘Transparency International’ (TI) questions in the elicitation sheet (Part 2) as well as ques- tions 6, 7, 8, 9, 11, and 12 in the questionnaire (Part 3). If subjects in T2- INFO learned something from the average donation in T1, one could expect systematic differences be- tween T2 and T2- INFO. Out of the eight considered questions, however, a Wilkoxon signed- rank test only finds significant differences in the ‘waste’ question (p = 0.0279) and question 12, that is, subjects' support to current government (p = 0.0437).17 In short, the effect is far from systematic; in particular, subjects have similar perceptions across groups regarding the quality of public services like education, health, and security, i.e., questions 7, 8, and 9. In any case, the issue may warrant further research to clarify the relative importance of the informational and motivational channels. Result 2 (social information effect): The distribution of donations changes if subjects receive information about others' average behavior. Therefore, the effect of descriptive social informa- tion and H2 are supported. Testing hypotheses 3–5: the role of the subject's political support and perceptions about the government's honesty and its promotion of public good We considered three potential measures of the subject's perceptions concerning political sup- port and perceptions about the government's honesty and its promotion of public good (see the questionnaire in the online appendix for the exact wording of each question). Variable 3 (Table 3) portrays the subject's support for the current presidential team. In contrast, variable 4 presents the subject's estimated position of Peru in the corruption index of 2015 by TI. Variable 5 shows the subject's agreement with the following two statements: (i) The Peruvian govern- ment is controlled by a few interests that are only concerned with themselves, and (ii) the Peruvian government governs for the benefit of all. Answers were numerical, from 0 (complete agreement with the first statement) to 10 (complete agreement with the second one). Variable 3, 4, and 5 in Table 3 test H3- H5, respectively. We observe that the coefficients of variables 4 and 5 in Models 1, 2, and 3 are never sig- nificant. In contrast, variable 3, support for the current president, is statistically significant (p = 0.009, 0.026, and 0.022 in Models 1, 2, and 3, respectively). Hypothesis 3, therefore, seems vindicated. However, hypotheses 4 and 5 are rejected. The perceived degree of corruption and perceived promotion of public good do not seem to explain changes in the amount donated 17Subjects in T2- INFO gave less support to the president than participants in T2, see Table 2. 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense | 15LÓPEZ- PÉREZ et al. to the government, perhaps because most subjects have very negative perceptions in both re- spects and hence we lack enough variability. We have also considered an extension of regression Model 3 including several socio- demographic controls, i.e., age, gender, socio- economic status, education level, political ideology, religiosity, and car ownership (YES/NO) (see Appendix  3; our sample for this model is reduced to 145 subjects, as some did not answer all questions). Several results are worth mentioning. First of all, we find that gender, age and religiosity have no significant effect on the donation. Still, we note that donations increase as we move to the right- wing of the ideological spectrum (p- value = 0.099), which is perhaps natural given that president Kuczynski's political party (Peruanos Por el Kambio) has been described as either center- right or conservative (Meléndez, 2019). Second, support for the president remains signifi- cant at the 0.05 level (p = 0.029). We also tested for the socio- economic level with a variable constructed by the Peruvian Market Research Firms' Association (APEIM), which consid- ers the subject's income but also her neighborhood of residence, the number of vehicles that she owns, the education level, having a (private) health insurance, and other characteristics. This variable takes five possible values (A, B, C, D and E), A being the highest, and it seems a fairly good approximation to the level of wealth and income of the subject's household. We find it not to be correlated with the amount donated. Finally, we note that support for the president becomes marginally significant (p = 0.093) if we add in the regression a (signif- icant) dichotomic variable for having children (results available upon request). We leave a more focused analysis of this issue for further research. In any case, the ability of our simple model of norms to explain giving is arguably noteworthy. The following result summarizes our discussion so far. Result 3: Donations directly depend on the subject's support for the current government. In contrast, donations are not explained by the perceived government's honesty or pro- motion of public good. Socio- demographic variables do not influence donations, except having children. For a final comment, the significant reduction of 32.8% in the average donation (from 4.67 to 3.14 Soles) across T1 and T2, which were run approximately with one year of difference, is also evidence consistent with Hypothesis 3 (although admittedly of a less controlled nature). In effect, we have seen that support for the current president is a significant explanatory variable, and this variable has a significantly lower median value in T2,18 possibly reflecting the fall in popular support that President Kuczynski's government suffered during his first year of man- date.19 Still, the decrease in the average donation from T1 to T2 is perhaps not entirely due to this factor: In Model 1, we introduce a binary variable for T2 and find a significant negative effect (p- value = 0.026). DISCUSSION This paper makes contributions to the literature on public goods and donations (Zelmer, 2003; Chaudhuri, 2011). In the well- known VCM game, a selfish player has a domi- nant strategy not to contribute any of her tokens to a public project, even though the so- cially efficient outcome is obtained when everyone contributes her entire endowment. In contrast with this standard prediction, a robust result from this literature is conditional cooperation (CCO henceforth): Subjects contribute more if they expect high contributions 18Median support in T1 and T2 was 7 and 5, respectively; Mann–Whitney, p = 0.017. 19His support at the time of T1 in November 2016 was 51% but decreased to 27% by the time of T2 in November 2017 (source: IPSOS Market Research; see https:// www. ipsos. com/ es- pe/ encue sta- lider es- de- opini on- de- latin oamer ica- aprob acion - presi dencial). By March 21st, 2018 President Kuczynski resigned his presidency after being involved in a vote- buying scandal. 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense https://www.ipsos.com/es-pe/encuesta-lideres-de-opinion-de-latinoamerica-aprobacion-presidencial https://www.ipsos.com/es-pe/encuesta-lideres-de-opinion-de-latinoamerica-aprobacion-presidencial 16 | WHY DO PEOPLE GIVE TO THEIR GOVERNMENTS? from their co- players (Fischbacher et al., 2001). Our finding that beliefs about average do- nation correlate with giving have clear parallelisms with CCO, showing that real taxpayers are conditional even in a one- shot, anonymous interaction with their governments. The fact that donations under- respond to beliefs in our experiment is also coherent with the “self- serving bias” that characterizes CCO, i.e., when a conditional cooperator expects others to contribute x to the public good, she tends to contribute less than x (Fischbacher et al., 2001; Fischbacher & Gächter, 2010). Therefore, giving to the government in the lab exhibits many of the patterns observed in VCM experiments. Our paper is also related to the many experimental studies in the lab and the field that address motives for charitable giving; see Vesterlund (2016) for a survey.20 Three utility mod- els have received particular attention in this literature. I: pure altruism, as in Becker (1974), assumes that individuals benefit from the aggregate level of the public good. II: the warm glow is a private benefit that derives from the act of giving, irrespectively of the subsequent output or the recipients' well- being (Andreoni, 1989). III: impure altruism considers both I and II simultaneously (Andreoni, 1989). The evidence from T2- INFO showing that social information affects behavior speaks against pure altruism as applied to the government giving. The correlation between beliefs and giving is also at odds, as pure altruism predicts that the contributions by others should crowd out one's own giving under the assumption of diminishing marginal utility for the public good (Warr, 1982). If all givers in our experiments were pure altruists, that is, beliefs and donations should be negatively correlated, contrary to what we observe. Regarding warm- glow, our regression analysis indicates that individu- als care about the government's output. In effect, a subject's donation to the government is significantly correlated with her level of support of the current president, arguably a proxy for the government's perceived competency and efficiency. Overall, our results support im- pure altruism, in line with substantial literature on charitable donations (e.g., Andreoni, 1993; Bolton & Katok, 1998), although at a population level, the three motives cited above may be present to some extent. More specifically, our findings suggest the importance of social norms in both a descriptive and injunctive sense (Cialdini et  al.,  1990; Bicchieri,  2005; López- Pérez, 2008). We also feel that our evidence provides insights as well on people's non- selfish motives to pay their taxes. In our experiment, for granted, there was no formal obligation to give, and sanctions play most likely an essential role in accounting for actual tax compliance and evasion. However, our findings show that people have motives to contribute voluntarily to public projects, condi- tional on some factors. Such motives might not be fully crowded out or ‘deactivated’ when there is compulsion.21 In this regard, our results might potentially contribute to the abundant literature on “Tax Morale”, which analyzes the importance of psychological and cultural elements to explain taxpayers' behavior (Scholz & Witte  1989; Andreoni et  al.,  1998; Luttmer & Singhal,  2014; Alm, 2019).22 In a nutshell, we conjecture that, as subjects in our study, taxpayers might be willing to contribute (some) money to the government (e.g., voluntarily paying (part of) their taxes), par- ticularly if they have relatively positive perceptions about its performance and believe that other 20Relatedly, recent papers like Deb et al. (2014), Drouvelis and Marx (2021), and Fielding et al. (2019), show the importance of social comparisons in charitable giving. There are several differences between these studies and ours: (i) subject pool (undergraduate students vs. taxpayers); (ii) location (high vs. middle income country), and (iii) recipient of donation (private charity vs. government). In our experiment, further, subjects make a one- shot decision, and not a series of donation decisions conditional, say, on income, average giving by others, or the level of some tax or subsidy. This means that we cannot categorize subjects by giving type, a point that we leave for further research. 21In this vein, some neurological evidence suggests that both taxation and voluntary donations activate similar neural substrates: Harbaugh et al. (2007) show that both voluntary donations and mandatory transfers to a charity elicit activity in the same brain region associated with processing rewards. 22This literature indeed operates under the implicit assumption that non- standard motivations like peer effects, reciprocity, and social norms, to name a few, are not deactivated when tax evaders risk punishment. 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense | 17LÓPEZ- PÉREZ et al. taxpayers comply as well. This conjecture seems supported by previous survey evidence from Latin America (e.g., Torgler, 2005; Ortega et al., 2016).23 CONCLUSIONS A N D POLICY IM PLICATIONS This paper aimed to explore the determinants of giving to the government via donations. We tested several theory- based hypotheses on norms, beliefs, social information, and political sup- port. Further, we conducted an artefactual fiend experiment in Peru and tested our hypotheses with Tobit regressions. While giving was not directed to some specific public project or govern- ment agency, subjects still gave around 14% of the endowment. Further, we offer evidence that giving depends on descriptive social information (i.e., what others actually do), perceived social norms (i.e., beliefs about what others do), and the subject's support for the government. Our findings suggest ways to optimize donations as a revenue- raising source in the Global South and North. First, governments could create fiscal space if citizens are given the option to donate via direct deposits to a bank account, using income- tax forms to donate to specific public projects/agencies, or even renouncing their income tax refund.24 Second, voluntary donations are more propitious when people expect others to give as well, so that managing expectations in this respect matters (some anecdotal evidence in countries like Peru suggests that taxpayers have over- pessimistic and inaccurate beliefs). Relatedly, third, our INFO treatment results prove the revenue- raising potential of informing the population about norms. For real- world equivalents, direct messages about the average donations of fellow citizens in previous years, via mailing, social media ads, or in the income tax form, might positively affect the amount donated to the government (particularly if people have over- pessimistic perceptions in this regard). These infor- mational strategies can be pursued at a relatively small cost for the public administration, which often has limited resources to pursue a very strict control strategy (see Hallsworth et al., 2017). For example, real- world equivalents of our informational treatment on donation behaviour could be media advertising by the Treasury, or a statement in the personal income tax statement where citizens are informed on the average donations of fellow citizens in previous years. Fourth, since donations correlate with citizens' support for the government/president, giving can fluctu- ate highly through time with the president's or government's approval rate, as the comparison of average giving in T1 and T2 suggests. For granted, improving perceptions about the government can be challenging in some societies, particularly in polarized ones. But knowing that it might lead to higher contributions (e.g., donations or tax receipts) is not irrelevant.25 A natural question is whether our findings can be extrapolated to the whole population and have external validity. On the first point, and even though we have a larger sample size than the optimum (see List et al., 2011), it seems likely that only large effects can be identified with our sample. Regarding external validity, our design has several positive aspects. First, subjects' decisions in our artefactual field experiment were not hypothetical: These decisions 23Several Latinobarómetro reports also find that the payment of taxes has a high statistical relationship with the citizens' perception that governments work for the wellbeing of all. It must be noted that such perceptions tend to be negative in most Latin American countries, particularly in Argentina, Dominican Republic, and Peru. In this respect, while we find that Peruvian taxpayers differ in their willingness to give money to the government, our conjecture also hints that, in a cross- country comparison, aggregate tax evasion will be relatively high in countries where the average or modal perception is negative, like Peru. Similarly, the differences in the rates of tax evasion normally observed between developed and other economies could be partly due to differences in these perceptions and social comparison effects. 24Donations to governments are a source of raising public revenues that is generally not included in the literature nor used by practically any government (Cruz- Martinez, 2018; Ortiz et al., 2019). 25As a cautionary tale concerning policy implications, donations to the government might exacerbate corruption and mismanagement (irrespective of whether in the Global South or North). For instance, governments might seek to manipulate citizens' view of the efficiency of the state apparatus, and partisanship might lead to unconditional payments by partisans even when the government is failing to use the additional revenues fairly and efficiently. 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense 18 | WHY DO PEOPLE GIVE TO THEIR GOVERNMENTS? have real consequences, i.e., money is actually donated to the Peruvian Treasury, and partic- ipants are aware of this. This design feature arguably adds realism and increases the external validity of results. Second, subjects are a representative sample of the taxpayer population of Metropolitan Lima regarding quotas of age, gender, and socioeconomic conditions. Third, subjects were randomly assigned to treatment and experimental groups. Finally, and as we have noted in Sections 2 and 5, our results accord well with several robust findings in the ex- perimental literature on VCM experiments and donations, and also with several findings from the literature on Tax Morale. At the same time, we highly doubt that our exact numerical findings can be directly extrapo- lated out of the lab. For instance, around half of the participants in our experiment give more than 16% of their endowment to the government. Yet it seems absolutely unlikely that they would donate the same fraction of their income every fiscal year (or even any),26 for multiple reasons, like decreasing marginal utility of donations (i.e., some form of ‘satiation’); a distinction between money truly earned and effortlessly coming “from heaven”, as in our experiment; con- text and framing effects, etc. This can be explored in further experiments, but it seems to us that the key message of our findings is not in the exact frequency and average of donations, i.e., quantitative external validity is totally misplaced. On the contrary, our important finding is the identification of direction rather than the magnitude of the effects/hypotheses that we study (see Frechette & Schotter, 2015, for a similar argument). ACK NOW LEDGM EN TS We are grateful to Turkay Salim Nefes, Rocío Sánchez- Mangas, and participants at the IV Annual Congress for Economists for helpful comments and suggestions; we appreciate the help provided by Luciano Federico Radesca with the translations of the abstracts. Finally, we are grateful for the helpful research assistance provided by Deyvi Abanto, University of Lima. F U N DI NG I N FOR M AT ION This work was supported by the Instituto de Investigacion Cientifica IDIC at the University of Lima Peru and the Spanish Ministry of Economics and Competitiveness (ECO2017- 82449- P). DATA AVA I LA BI LI T Y STAT EM EN T The data that support the findings of this study are available on request from the correspond- ing author. The data are not publicly available due to privacy or ethical restrictions. ORCI D Raúl López- Pérez  https://orcid.org/0000-0002-8952-8914 Aldo Ramirez- Zamudio  https://orcid.org/0000-0001-7509-3975 Gibrán Cruz- Martínez  https://orcid.org/0000-0002-4583-2914 R E F ER E NC E S Alm, J. (2019). What motivates tax compliance? Journal of Economic Surveys, 33(2), 353–388. Andreoni, J. (1989). Giving with Impure Altruism: Applications to Charity and Ricardian Equivalence. The Journal of Political Economy, 97(6), 1447–1458. Andreoni, J., Erard, B., & Feinstein, J. (1998). Tax compliance. Journal of Economic Literature, 36, 818–860. Ayuso, M. (2018, May 30). ¿Renuncia a su devolución en favor del Tesoro Público? 52.000 personas sí. La Información. https:// www. lainf ormac ion. com/ manag ement/ renun cia- a- su- devol ucion - en- favor - del- tesor o- publi co- 52- 000- perso nas- si/ 63491 67/ 26We note incidentally that the average income tax rate in Peru was equal to 17.8% in 2017. For details, see (in Spanish): http:// renta. sunat. gob. pe/ 2017/ assets/ pdf/ caso_ pract ico_ Renta_ traba jo_ Fuente_ Extra njera_ 2017. pdf. 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense https://orcid.org/0000-0002-8952-8914 https://orcid.org/0000-0002-8952-8914 https://orcid.org/0000-0001-7509-3975 https://orcid.org/0000-0001-7509-3975 https://orcid.org/0000-0002-4583-2914 https://orcid.org/0000-0002-4583-2914 https://www.lainformacion.com/management/renuncia-a-su-devolucion-en-favor-del-tesoro-publico-52-000-personas-si/6349167/ https://www.lainformacion.com/management/renuncia-a-su-devolucion-en-favor-del-tesoro-publico-52-000-personas-si/6349167/ http://renta.sunat.gob.pe/2017/assets/pdf/caso_practico_Renta_trabajo_Fuente_Extranjera_2017.pdf http://renta.sunat.gob.pe/2017/assets/pdf/caso_practico_Renta_trabajo_Fuente_Extranjera_2017.pdf | 19LÓPEZ- PÉREZ et al. Andreoni, J. (1993). An Experimental Test of the Public- Goods Crowding- Out Hypothesis. The American Economic Review, 83(5), 1317–1327. Becchetti, L., Castriota, S., & Conzo, P. (2017). Disaster, aid, and preferences: The long- run impact of the tsunami on giving in Sri Lanka. World Development, 94, 157–173. Becker, G.S. (1974). A Theory of Social Interactions. Journal of Political Economy, 82(6), 1063–1093. Bicchieri, C. (2005). The grammar of society: The nature and dynamics of social norms. Cambridge University Press. Bicchieri, C., & Xiao, E. (2009). Do the right thing: but only if others do so. Journal of Behavioral Decision Making, 22, 191–208. Bodea, C., & LeBas, A. (2016). The Origins of Voluntary Compliance: Attitudes toward Taxation in Urban Nigeria. British Journal of Political Science, 46(1), 215–238. Bohnet, I., & Zeckhauser, R. (2004). Social comparisons in ultimatum bargaining. Scandinavian Journal of Economics, 106(3), 495–510. Bolton, G.E., & Katok, E. (1998). An Experimental Test of the Crowding- Out Hypothesis: The Nature of Beneficent Behavior. Journal of Economic Behavior & Organization, 37(3), 315–331. Cason, T. & Mui, V.L. (1998). Social Influence in the Sequential Dictator Game. Journal of Mathematical Psychology, 42, 248–265. Chaudhuri, A. (2011). Sustaining cooperation in laboratory public goods experiments: A selective survey of the liter- ature. Experimental Economics, 14, 47–83. Cialdini, R.B., Reno, R.R. & Kallgren, C.A. (1990). A focus theory of normative conduct: Recycling the concept of norms to reduce littering in public places. Journal of Personality and Social Psychology, 58, 1015–1026. Cialdini, R., & Trost, M.R. (1998). Social influence: Social norms, conformity, and compliance. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), The handbook of social psychology (pp. 151–192). Oxford University Press. Croson, R., Handy, F., & Shang, J. (2009). Keeping up with the Joneses: The relationship of perceived descriptive social norms, social information, and charitable giving. Nonprofit Management and Leadership, 19(4), 467–489. https:// doi. org/ 10. 1002/ nml. 232 Cruz- Martinez, G. (2018). Revenue- Generating Potential of Taxation for Older- Age Social Pensions. Ageing International, 43, 415–437. https:// doi- org. bucm. idm. oclc. org/ 10. 1007/ s1212 6- 017- 9298- 2 Deb, R., Gazzale, S., & Matthew, K. (2014). Testing motives for charitable giving: A revealed- preference method- ology with experimental evidence. Journal of Public Economics, 120, 181–192. https:// doi. org/ 10. 1016/j. jpube co. 2014. 09. 008 Drouvelis, M., & Marx, B. M. (2021). Dimensions of donation preferences: The structure of peer and income effects. Experimental Economics, 24, 274–302. https:// doi. org/ 10. 1007/ s1068 3- 020- 09667 - 0 Druckman, J. N., & Green, D. (2021). Advances in experimental political science. Cambridge University Press. Dularif, M., & Rustiarini, N. W. (2021). Tax compliance and non- deterrence approach: A systematic review. International Journal of Sociology and Social Policy. https:// doi. org/ 10. 1108/ IJSSP - 04- 2021- 0108 Eckel, C., & Grossman, P. (1996). Altruism in anonymous dictator games. Games and Economic Behavior, 16, 181– 191. https:// doi. org/ 10. 1006/ game. 1996. 0081 Fieldhouse, E., & Cutts, D. (2021). Do as I say or do as I do? How social relationships shape the impact of descriptive and injunctive norms of voting. British Journal of Political Science, 51(4), 1516–1528. https:// doi. org/ 10. 1017/ S0007 12342 0000058 Fieldhouse, E., Cutts, D., & Bailey, J. (2022). Who cares if you vote? Partisan pressure and social norms of voting. Political Behavior, 44, 1297–1316. https:// doi. org/ 10. 1007/ s1110 9- 020- 09646 - 3 Fielding, D., Knowles, S., & Robertson, K. (2019). Materialists and altruists in a charitable donation experiment. Oxford Economic Papers, 72(1), 216–234. https://doi.org/10.1093/oep/gpz027 Fischbacher, U., Gächter, S., & Fehr, E. (2001). Are people conditionally cooperative? Evidence from a public goods experiment. Economics Letters, 71, 397–404. https:// doi. org/ 10. 1016/ S0165 - 1765(01) 00394 - 9 Fischbacher, U., & Gächter, S. (2010). Social preferences, beliefs, and the dynamics of free riding in public goods experiments. American Economic Review, 100(1), 541–56. https:// doi. org/ 10. 1257/ aer. 100.1. 541 Frechette, G., & Schotter, A. (2015). Handbook of experimental economic methodology. Oxford University Press. Frey, B., & Meier, S. (2004). Social comparisons and pro- social behavior: Testing conditional cooperation in a field experiment. American Economic Review, 94(5), 1717–1722. https:// doi. org/ 10. 1257/ 00028 28043 052187 Gilens, M. (2015). Descriptive representation, money, and political inequality in the United States. Swiss Political Science Review, 21(2), 222–228. https:// doi. org/ 10. 1111/ spsr. 12164 Gino, F., Ayal, S., & Ariely, D. (2009). Contagion and differentiation in unethical behavior. Psychological Science, 20(3), 393–398. https:// doi. org/ 10. 1111/j. 1467- 9280. 2009. 02306. x Hallsworth, M. J., List, A., Metcalfe, R., & Vlaev, I. (2017). The behavioralist as tax collector: Using natural field experiments to enhance tax compliance. Journal of Public Economics, 148, 14–31. https:// doi. org/ 10. 1016/j. jpube co. 2017. 02. 003 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense https://doi.org/10.1002/nml.232 https://doi-org.bucm.idm.oclc.org/10.1007/s12126-017-9298-2 https://doi.org/10.1016/j.jpubeco.2014.09.008 https://doi.org/10.1016/j.jpubeco.2014.09.008 https://doi.org/10.1007/s10683-020-09667-0 https://doi.org/10.1108/IJSSP-04-2021-0108 https://doi.org/10.1006/game.1996.0081 https://doi.org/10.1017/S0007123420000058 https://doi.org/10.1017/S0007123420000058 https://doi.org/10.1007/s11109-020-09646-3 https://doi.org/10.1093/oep/gpz027 https://doi.org/10.1016/S0165-1765(01)00394-9 https://doi.org/10.1257/aer.100.1.541 https://doi.org/10.1257/0002828043052187 https://doi.org/10.1111/spsr.12164 https://doi.org/10.1111/j.1467-9280.2009.02306.x https://doi.org/10.1016/j.jpubeco.2017.02.003 https://doi.org/10.1016/j.jpubeco.2017.02.003 20 | WHY DO PEOPLE GIVE TO THEIR GOVERNMENTS? Hansen, P., Kergozou, N., Knowles, S., & Thorsnes, P. (2014). Developing countries in need: Which characteristics appeal most to people when donating money? The Journal of Development Studies, 50(11), 1494–1509. https:// doi. org/ 10. 1080/ 00220 388. 2014. 925542 Harbaugh, W. T., Mayr, U., & Burghart, D. R. (2007). Neural responses to taxation and voluntary giving reveal mo- tives for charitable donations. Science, 316, 1622–1625. https:// doi. org/ 10. 1126/ scien ce. 1140738 Harrison, G. W., & List, J. A. (2004). Field experiments. Journal of Economic Literature, 42(4), 1009–1055. https:// doi. org/ 10. 1257/ 00220 51043 004577 Hellmüller, S., Pring, J., & Richmond, O. P. (2020). How norms matter in mediation: An introduction. Swiss Political Science Review, 26(4), 345–363. https:// doi. org/ 10. 1111/ spsr. 12418 Hoffman, A. M. (2007). The structural causes of trusting relationships: Why rivals do not overcome suspicion step by step. Political Science Quarterly, 122(2), 287–312. https:// doi. org/ 10. 1002/j. 1538- 165X. 2007. tb006 17. x IPSOS. (2017, 8 February). Encuesta a líderes de opinión de Latinoamérica: Aprobación Presidencial. IPSOS Update. https:// www. ipsos. com/ es- pe/ encue sta- lider es- de- opini on- de- latin oamer ica- aprob acion - presi dencial Jones, K. (2017). Government or charity? Preferences for welfare provision by ethnicity. Journal of Behavioral and Experimental Economics, 66, 72–77. https:// doi. org/ 10. 1016/j. socec. 2016. 04. 011 Kouamé, W. A. K. (2021). Trust to pay? Tax morale and trust in Africa. The Journal of Development Studies, 57(7), 1086–1105. https:// doi. org/ 10. 1080/ 00220 388. 2021. 188149 Kriesi, H., & Baglioni, S. (2003). Putting local associations into their context. Preliminary results from a Swiss study of local associations. Swiss Political Science Review, 9(3), 1–34. https:// doi. org/ 10. 1002/j. 1662- 6370. 2003. tb003 40. x Li, X., Eckel, C., Grossman, P., & Brown, L. (2011). Giving to government: Voluntary taxation in the lab. Journal of Public Economics, 95(9–10), 1190–1201. https:// doi. org/ 10. 1016/j. jpube co. 2011. 03. 001 Li, X., Eckel, C., Grossman, P., & Brown, L. (2015). Directed giving enhances voluntary giving to government: Implications for tax policy. Economics Letters, 133, 51–54. https:// doi. org/ 10. 1016/j. econl et. 2015. 05. 021 List, J. A., Sadoff, S., & Wagner, M. (2011). So you want to run an experiment, now what? Some simple rules of thumb for optimal experimental design. Experimental Economics, 14, 439–457. https:// doi. org/ 10. 1007/ s1068 3- 011- 9275- 7 López- Pérez, R. (2008). Aversion to norm- breaking: A model. Games and Economic Behavior, 64, 237–267. https:// doi. org/ 10. 1016/j. geb. 2007. 10. 009 López- Pérez, R., & Spiegelman, E. (2013). Why do people tell the truth? Experimental evidence for pure lie aversion. Experimental Economics, 16(3), 233–247. https:// doi. org/ 10. 1007/ s1068 3- 012- 9333- 4 López- Pérez, R., & Ramírez- Zamudio, A. (2020). An experimental test of two policies to increase donations to public projects. International Review of Law and Economics, 62, 1–11. https:// doi. org/ 10. 1016/j. irle. 2020. 105899 Luccasen, R. A., & Thomas, M. K. (2020). Voluntary taxation and the arts. Journal of Cultural Economics, 44, 589–604. https:// doi. org/ 10. 1007/ s1082 4- 020- 09376 - 2 Luttmer, E., & Singhal, M. (2014). Tax morale. Journal of Economic Perspectives, 28(4), 149–168. https:// doi. org/ 10. 1257/ jep. 28.4. 149 Margolis, M. F., & Sances, M. W. (2017). Partisan differences in nonpartisan activity: The case of charitable giving. Political Behavior, 39, 839–864. https:// doi. org/ 10. 1007/ s1110 9- 016- 9379- 5 Meléndez, C. (2019). La derecha que se bifurca. Las vertientes populista- conservadora y tecnocrática- liberal en Perú post- 2000. Colombia Internacional, 99, 3–27. https:// doi. org/ 10. 7440/ colom biain t99. 2019. 01 Ortega, D., Ronconi, L., & Sanguinetti, P. (2016). Reciprocity and willingness to pay taxes: Evidence from a survey experiment in Latin America. Economía, 16(2), 55–87. https:// doi. org/ 10. 1353/ eco. 2016. 0010 Ortiz, I., Chowdhury, A., Durán- Valverde, F., Muzaffar, T. & Urban, S. (2019). Fiscal space for social protection: A handbook for assessing financing options. International Labour Organization. Ozdemir, U., & Jacob, M. (2022). Values, taboos, and votes: How basic human values affect populist electoral sup- port. Swiss Political Science Review, 28(3), 455–476. https:// doi. org/ 10. 1111/ spsr. 12449 Phillips, H. (2012). The funding of political parties. The Political Quarterly, 83(2), 318–324. https:// doi. org/ 10. 1111/j. 1467- 923X. 2012. 02323. x Prichard, W. (2018). Electoral competitiveness, tax bargaining and political incentives in developing countries: Evidence from political budget cycles affecting taxation. British Journal of Political Science, 48(2), 427–457. https:// doi. org/ 10. 1017/ S0007 12341 5000757 Ross, L., Greene, D., & House, P. (1977). The “false consensus effect”: An egocentric bias in social perception and attribution processes. Journal of Experimental Social Psychology, 13(3), 279–301. https:// doi. org/ 10. 1016/ 0022- 1031(77) 90049 - X Rothstein, B., & Uslaner, E. (2005). All for all. Equality, corruption and social trust. World Politics, 58(1), 41–73. https:// doi. org/ 10. 1353/ wp. 2006. 0022 Scholz, R., & Witte, V. (1989). Taxpayer compliance. University of Pennsylvania Press. Shang, J., & Croson, R. (2009). A field experiment in charitable contribution: The impact of social information on the voluntary provision of public goods. The Economic Journal, 119, 1422–1439. https:// doi. org/ 10. 1111/j. 1468- 0297. 2009. 02267. x 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense https://doi.org/10.1080/00220388.2014.925542 https://doi.org/10.1080/00220388.2014.925542 https://doi.org/10.1126/science.1140738 https://doi.org/10.1257/0022051043004577 https://doi.org/10.1257/0022051043004577 https://doi.org/10.1111/spsr.12418 https://doi.org/10.1002/j.1538-165X.2007.tb00617.x https://www.ipsos.com/es-pe/encuesta-lideres-de-opinion-de-latinoamerica-aprobacion-presidencial https://doi.org/10.1016/j.socec.2016.04.011 https://doi.org/10.1080/00220388.2021.188149 https://doi.org/10.1002/j.1662-6370.2003.tb00340.x https://doi.org/10.1016/j.jpubeco.2011.03.001 https://doi.org/10.1016/j.econlet.2015.05.021 https://doi.org/10.1007/s10683-011-9275-7 https://doi.org/10.1007/s10683-011-9275-7 https://doi.org/10.1016/j.geb.2007.10.009 https://doi.org/10.1016/j.geb.2007.10.009 https://doi.org/10.1007/s10683-012-9333-4 https://doi.org/10.1016/j.irle.2020.105899 https://doi.org/10.1016/j.irle.2020.105899 https://doi.org/10.1007/s10824-020-09376-2 https://doi.org/10.1257/jep.28.4.149 https://doi.org/10.1257/jep.28.4.149 https://doi.org/10.1007/s11109-016-9379-5 https://doi.org/10.7440/colombiaint99.2019.01 https://doi.org/10.1353/eco.2016.0010 https://doi.org/10.1111/spsr.12449 https://doi.org/10.1111/j.1467-923X.2012.02323.x https://doi.org/10.1111/j.1467-923X.2012.02323.x https://doi.org/10.1017/S0007123415000757 https://doi.org/10.1016/0022-1031(77)90049-X https://doi.org/10.1016/0022-1031(77)90049-X https://doi.org/10.1353/wp.2006.0022 https://doi.org/10.1111/j.1468-0297.2009.02267.x https://doi.org/10.1111/j.1468-0297.2009.02267.x | 21LÓPEZ- PÉREZ et al. Slemrod, J. (2003). Trust in public finance. In S. Cnossen & H.- W. Sinn (Eds.), Public finance and public policy in the new century (pp. 49–88). MIT Press. Slemrod, J., & Kuchumova, Y. (2023). Gifts to government. International Tax and Public Finance, 30, 453–492. https:// doi. org/ 10. 1007/ s1079 7- 022- 09675 - 8 Taliercio, R. (2004). Administrative reform as credible commitment: The impact of autonomy on revenue authority performance in Latin America. World Development, 32(2), 213–232. https:// doi. org/ 10. 1016/j. world dev. 2003. 07. 009 Torgler, B. (2005). Tax morale in Latin America. Public Choice, 122(1/2), 133–157. https:// doi. org/ 10. 1007/ s1112 7- 005- 3214- 3 Uslaner, E. M. (2000). Producing and consuming trust. Political Science Quarterly, 115(4), 569–590. https:// doi. org/ 10. 2307/ 2657610 van Teunenbroek, C., Bekkers, R., & Beersma, B. (2021). They ought to do it too: Understanding effects of social information on donation behavior and mood. International Review of Public and Nonprofit Marketing, 18, 229– 253. https:// doi. org/ 10. 1007/ s1220 8- 021- 00258 - 0 Vesterlund, L. (2013). The informational value of sequential fundraising. Journal of Public Economics, 87(3–4), 627– 657. https:// doi. org/ 10. 1016/ S0047 - 2727(02) 00132 - 9 Vesterlund, L. (2016). Using experimental methods to understand why and how we give to charity. In J. Kagel & A. Roth (Eds.), The handbook of experimental economics (Vol. 2, pp. 91–152). Princeton University Press. Warr, G. (1982). Pareto optimal redistribution and private charity. Journal of Public Economics, 19(1), 131–38. https:// doi. org/ 10. 1016/ 0047- 2727(82) 90056 - 1 Zelmer, J. (2003). Linear public goods experiments: A meta- analysis. Experimental Economics, 6, 299–310. https:// doi. org/ 10. 1023/A: 10262 77420119 Zizzo, D. J. (2010). Experimenter demand effects in economic experiments. Experimental Economics, 13, 75–98. https:// doi. org/ 10. 1007/ s1068 3- 009- 9230- z AU T HOR BIOGR A PH I E S Raúl López- Pérez, Research Fellow at the Institute of Public Goods and Policies of the Spanish National Research Council (CSIC). His main research interests lie primarily in areas like social norms, prosocial and antisocial behavior, the formation of beliefs and pref- erences, conflict, and collective action. Email: raul.lopez-perez@cchs.csic.es Aldo Ramirez- Zamudio, Director of the Center for Economics, Banking and Finance Studies, Department of Economics, Universidad de Lima. His research focuses on taxation using social experiments and game theory. Email: aframire@ulima.edu.pe Gibrán Cruz- Martínez, Assistant Professor and Coordinator of the master's in Political Analysis at the Department of Political Science and Public Administration, Complutense University of Madrid. His research focuses on comparative social policy, welfare regimes, determinants of welfare state development, multidimensional poverty and the multilevel governance of welfare states. Email: gicruz@ucm.es SU PPORT I NG I N FOR M AT ION Additional supporting information can be found online in the Supporting Information section at the end of this article. How to cite this article: López- Pérez, R., Ramirez- Zamudio, A. & Cruz- Martínez, G. (2023). Why do people give to their governments? Lab- in- the- field evidence on the role of norms, social information, and political support. Swiss Political Science Review, 00, 1–21. https://doi.org/10.1111/spsr.12583 16626370, 0, D ow nloaded from https://onlinelibrary.w iley.com /doi/10.1111/spsr.12583 by R eadcube (L abtiva Inc.), W iley O nline L ibrary on [20/12/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense https://doi.org/10.1007/s10797-022-09675-8 https://doi.org/10.1016/j.worlddev.2003.07.009 https://doi.org/10.1007/s11127-005-3214-3 https://doi.org/10.1007/s11127-005-3214-3 https://doi.org/10.2307/2657610 https://doi.org/10.2307/2657610 https://doi.org/10.1007/s12208-021-00258-0 https://doi.org/10.1016/S0047-2727(02)00132-9 https://doi.org/10.1016/0047-2727(82)90056-1 https://doi.org/10.1016/0047-2727(82)90056-1 https://doi.org/10.1023/A:1026277420119 https://doi.org/10.1023/A:1026277420119 https://doi.org/10.1007/s10683-009-9230-z mailto:raul.lopez-perez@cchs.csic.es mailto:aframire@ulima.edu.pe mailto:gicruz@ucm.es https://doi.org/10.1111/spsr.12583 Why do people give to their governments? Lab-­in-­the-­field evidence on the role of norms, social information, and political support Abstract Zusammenfassung Résumé Riassunto INTRODUCTION EXPLAINING THE ACT OF GIVING: THE DEBATE IN THE LITERATURE EXPERIMENTAL DESIGN AND PROCEDURES DATA ANALYSIS Summary of Results What affects giving: A regression analysis Testing Hypothesis 1: the role of beliefs and descriptive norms Testing the social information hypothesis: The effect of the T2-­INFO treatment Testing hypotheses 3–5: the role of the subject's political support and perceptions about the government's honesty and its promotion of public good DISCUSSION CONCLUSIONS AND POLICY IMPLICATIONS ACKNOWLEDGMENTS FUNDING INFORMATION DATA AVAILABILITY STATEMENT REFERENCES