The World Bank Economic Review, 37(2), 2023, 205–220 https://doi.org10.1093/wber/lhac026 Article Do Women Contribute More Effort than Men to a Downloaded from https://academic.oup.com/wber/article/37/2/205/6997901 by Joint Bank-Fund library user on 04 September 2023 Real Public Good? Ingela Alger, Laura Juarez, Miriam Juarez-Torres, and Josepa Miquel-Florensa Abstract This study presents evidence from a lab-in-the-field experiment, conducted in eight small, rural villages in Mexico, in which subjects choose to exert real effort to fund real health centers in their own and other localities. The results show that women are more willing than men to exert effort to fund the health center in another locality, relative to the one in their locality. This gender gap is mostly due to women who have some trust in the way the government spends taxes, and to those who benefit from a government program that tar- gets women and fosters healthcare use. These results also suggest that women might be aware of their higher willingness to exert effort for a public good that does not benefit them directly, compared to men, because they seem to reduce their individual effort the more female their environment is. JEL classification: H41, C91, O12 Keywords: public goods, gender, lab-in-the-field experiment, real effort, in- versus out-group transfers 1. Introduction A society’s ability to provide public goods, which are essential for economic development, depends on raising potentially distortionary taxes. For instance, labor taxes might disincentivize work, particularly if they do not benefit the worker directly.1 The non-experimental literature has found that labor-force participation and work hours respond to taxes, and such response varies by gender and education (Meghir and Phillips 2010). Taxes could also affect work effort, a relevant margin not measured in standard labor surveys. Ingela Alger is a CNRS Senior scientist at Toulouse School of Economics, University of Toulouse Capitole, and the Institute for Advanced Study in Toulouse, France; her email address is ingela.alger@tse-fr.eu. Laura Juarez is a professor at Centro de Estudios Economicos, El Colegio de México, Mexico City; her email address is laura.juarez@colmex.mx. Miriam Juarez- Torres is Deputy Manager of Statistical Analysis at the General Directorate of Economic Research, Banco de Mexico, Mexico City; her email address is mjuarez@banxico.org.mx. Josepa Miquel-Florensa is an associate professor at Toulouse School of Economics, University of Toulouse Capitole, Toulouse, France; her email address is: pepita.miquel@tse-fr.eu. The authors thank Mariana Garcia and Ildrim Valley for outstanding research assistance, both in the field and in the office; Ingela Alger and Josepa Miquel-Florensa thank Banco de Mexico for its generous hosting. The views expressed in this article are solely those of the authors and do not necessarily reflect those of Banco de Mexico. A supplementary online appendix is available with this article at The World Bank Economic Review website. 1 For instance, Summers (1989) shows, in a simple model, that labor taxes tied to mandated benefits that workers value have a lower disincentive effect on work compared to pure taxes, which do not benefit the worker directly. © The Author(s) 2023. Published by Oxford University Press on behalf of the International Bank for Reconstruction and Development / THE WORLD BANK. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com 206 Alger et al. This study tests whether women are more inclined than men to contribute effort towards a public good that may not benefit them directly. If confirmed, this gender gap would imply that policies that promote female labor-market participation and attachment might enhance tax revenues with fewer distortions.2 The data come from a lab-in-the-field experiment conducted in rural villages, inhabited mostly by low- income farmers, in the Mexican state of Yucatan.3 Gender differences in effort responses to taxes are unlikely to stem from those in exposure to real taxes in this population because of its isolation from the Downloaded from https://academic.oup.com/wber/article/37/2/205/6997901 by Joint Bank-Fund library user on 04 September 2023 formal tax system. In our experimental design, a subject’s real effort in a simple but boring task determined their prob- ability of generating earnings above and beyond the initial endowment; any such extra earnings were allocated to a public good. In the Local Public Good (LPG) game, they went to the public health cen- ter of the subject’s village; in the Non-local Public Good (NLPG) game, they went to the health center of another village, which may not benefit the subject directly. Thus, in this experiment, the tax rate was 100 percent, imposed by the research team, and not varied across treatments. An advantage of this design, compared to the standard public good experiment (in which a subject’s contribution also reduces their net payoff), is that it eliminates the income effect. It also isolates the effort choice from the transfer choice, and the former can be interpreted as the subject’s willingness to contribute effort to the corresponding public good. In each village, participants were divided randomly in two groups: Donors and Recipients. In the Donor treatment, any extra earnings generated by the subject increased the resources given to the public good; in the Recipient treatment, failure to generate such earnings reduced those resources, because the subject was compensated from them. Hence, this investigation also examines whether the gender differences in the willingness to exert effort for a public good vary by whether the subject gives/takes resources to/from the public good. The empirical analysis consists in estimating the effort difference between the NLPG and LPG games, and testing whether it varies significantly with gender and role (Donor versus Recipient). Arguably, this gender gap is a clean measure of the discrepancy between men’s and women’s willingness to contribute real effort to a public good that does not benefit them directly, because it controls for the effort they contribute to one that does, keeping the tax rate constant. To control for any individual characteristics that might affect effort, and that remained constant throughout the experimental session, all estimations include player fixed effects. The results show that, on average, female Donors exert about 7 percent more effort than male Donors in the NLPG, relative to the LPG game, and this estimate is significant. Thus, female Donors are more willing than their male counterparts to contribute effort for the health center of another locality, relative to that in their own. Although there is no income reporting in this study’s experiment, that finding is consistent with the results of Bruner, D’Attoma, and Steinmo (2017), who show that the gender gap in tax compliance is largest when the tax revenues do not benefit participants directly. Thus, their work also implies that women are more willing to pay taxes, by reporting their income truthfully, than men in this case. For Recipients, no significant gender differences in effort between the NLPG and LPG games are found. All these findings are robust to controlling for the order of games in the experimental session. This study further examines whether the estimated gender gap varies with measures of informal giving and receiving, community involvement, trust, strength of the participant’s network in the matched village and their own, and other individual characteristics. The findings show that most of the positive gender gap in relative effort for Donors is due to the behavior of women who have at least some trust in the way 2 This study is in line with the growing interest in the potential effects that women’s empowerment has on economic performance, both in developed (Bertrand 2011; Adams and Funk 2012), and in developing countries (Beaman et al. 2012; Dupas and Robinson 2013; Field et al. 2016). 3 This study also adds to the literature on behaviors and network structures in small villages in Mexico (Angelucci et al. 2009, 2010). The World Bank Economic Review 207 the government spends taxes, and to those who participate in the Oportunidades program. This program pays a cash transfer to poor, rural women, conditional on complying with family health checks and other requirements.4 Therefore, Oportunidades beneficiaries, women who use health services relatively often, are those who are more willing to exert effort than men for the clinic at the matched locality, relative to that in their own. Most of the literature on anti-poverty and social insurance programs has focused on the negative effects Downloaded from https://academic.oup.com/wber/article/37/2/205/6997901 by Joint Bank-Fund library user on 04 September 2023 of government programs on private interventions. In contrast, the Oportunidades result suggests that some government programs could actually foster social capital and a willingness to contribute to public goods, instead of crowding them out, as other recent studies find.5 Adato (2000) reports mixed effects of Progresa on social capital. On the one hand, the formation of a “club” of program beneficiaries reduces their interaction with the rest of their community. On the other hand, the interactive program events, such as health meetings for mothers or group community tasks, strengthen the sense of community. Attanasio, Polania-Reyes, and Pellerano (2015) also document the latter in Colombia: they show that the Familias en Acción program increased public good contributions in a lab-in-the-field public good game.6 Finally, the evidence shows that the positive gender gap in relative effort between the NLPG and LPG games for Donors decreases with the share of female participants in the experimental session. This result suggests that women might be aware of their higher willingness to exert effort for a public good that does not benefit them directly, compared to men. Consequently, women reduce their individual effort the more female their environment is. Two caveats regarding the external validity of these results are (Levitt and List 2007) as follows: (a) in our experiment subjects undertake effort for a short time, while in reality productive efforts take place over extended periods; (b) since the population we sampled resides within the village-based com- munal land system known as ejido, which involves mostly men, men’s pro-sociality towards their own village may be greater than in other population groups.7 This work contributes to the experimental literature on gender differences in economic experiments8 and the disincentive effects of taxes on effort. Several authors find a negative relationship between tax rates and effort exerted in real effort experiments (Swenson 1988; Sillamaa 1999; Sutter and Weck-Hannemann 2003; Lévy-Garboua, Masclet, and Montmarquette 2009; Keser, Masclet, and Montmarquette 2019). Among them, only Keser, Masclet, and Montmarquette (2019) look at the impact that the use of tax revenue has on effort choices. Most of these studies do not look at gender differences in the effort responses to taxes, and none of them studies the interaction between gender and the destination of the tax revenue collected.9 Thus, to our knowledge, this experimental study is the first to examine gender difference in real-effort decisions that impact a real public good, with variation on whether subjects can benefit directly 4 Oportunidades, which started as Progresa in 1997, is one of the best known conditional cash transfer programs in the world. The transfer from the program is also conditioned on children’s enrollment and regular attendance to school. The literature evaluating different aspects of this program is extensive. For recent surveys on the long-term effects of this program, see Millán et al. (2019), and on impact on education see Gertler, Patrinos, and Rubio-Codina (2012). 5 This finding is also consistent with the theory of reciprocity between citizens and the elite proposed by Besley (2020). 6 Other studies on the relationship between cash transfer programs, social capital, and trust in the government are Nguyen and Rieger (2017) and Evans, Holtemeyer, and Kosec (2019) for Morocco and Tanzania, respectively. 7 Relatedly, Carpenter and Seki (2011) find that exposure to teamwork had a positive impact on contributions in public good games among fishermen. 8 Please refer to the surveys by Croson and Gneezy (2009), and Niederle (2017). 9 In contrast, the literature on tax compliance shows that there is heterogeneity in tax compliance behaviors (Slemrod 2007; Mohdali, Isa, and Yusoff 2014; Mascagni 2018). In particular, this literature finds that women report a much higher fraction of their earned income than men (Bruner, D’Attoma, and Steinmo 2017; D’Attoma, Volintiru, and Malezieux 2020) and that this gap is particularly large when individuals do not benefit directly from the tax revenues collected (Bruner, D’Attoma, and Steinmo 2017). 208 Alger et al. from the public good or not.10 In addition, all previous experimental studies have been carried out in developed countries. Evidence for developing countries is scarce, but relevant given that raising taxes to finance public goods appears to be a challenge particularly in these countries (Besley and Persson 2014). This work also adds to the growing literature on how social distance impacts giving in developing countries (Alt et al. 2018; Candelo, Eckel, and Johnson 2018; Adena, Hakimov, and Huck 2020). The research closest to our study is the lab-in-the-field experiment run in rural Mexico by Candelo, Eckel, and Downloaded from https://academic.oup.com/wber/article/37/2/205/6997901 by Joint Bank-Fund library user on 04 September 2023 Johnson (2018). Their design uses the dictator game and varies the type of recipient across treatments. While they detect a more pronounced generosity towards family than non-family members, subjects are on average equally generous towards a random individual in another village or their own. Alt et al. (2018) also used the standard dictator game in an experiment with university students in Indonesia. Social distance was instilled artificially through the minimal group paradigm, and their design included both a donor and a recipient treatment (or a “give” and a “take” treatment). In line with our findings, they find that donations were higher in the give than in the take treatment. However, none of these studies use real-effort decisions or examine gender differences. 2. The Effort-and-Transfer Experiment Participants played three games, all with the same initial endowment and real-effort task, and completed a post-experiment questionnaire.11 The effort task consisted of using a pencil to mark the outline of pre- printed squares on a sheet of paper.12 A subject’s effort determined the probability of generating additional income. The sheet had 10 lines of empty squares, each corresponding to one “level” ( = 1, …, 10), which required squares to be marked, to make the marginal cost of effort increase. For each completed level, the subject increased the probability of generating additional income by 1/11, the probability being 0 if no level was completed. In the Autarky game, any additional income was simply added to the subject’s endowment. In the LPG game, it went to the health center in the subject’s village, and in the NLPG game, to the health center of another village. The empirical analysis considers only these two public good games. Details of the payoffs, the experimental protocol, and the procedures are presented in supplementary online appendices S1 and S4. Our experimental design, which builds upon that in Alger et al. (2020), has several advantages. First, the subject’s effort choice in both public good games affects the resources allocated to the corresponding public good but not their own monetary payoff. Thus, this design avoids the income effect typically present in experiments using the classical public good or dictator games, where a subject’s contribution increases the monetary payoff for those who benefit from it and reduces their own. Second, because 100 percent of the extra earnings were “taxed away,” a subject’s choice could be interpreted as their willingness to contribute real effort for the public good. Third, imposing the tax rate isolates the effort choice from the transfer choice, whereas in other experiments both are simultaneous. Finally, any potential social 10 The large body of experimental work on monetary contributions made by subjects in artificial public good games in developed countries has resulted in no clear gender differences (DellaVigna et al. 2013; Niederle 2017). For example, in experiments conducted on students with standard linear public good games, Brown-Kruse and Hummels (1993) find that women make lower contributions than men on average, while Nowell and Tinkler (1994) find the opposite. 11 In each session one of the games was randomly drawn to determine the payoffs of all the subjects in the session. A fourth game (the One-to-One game), was included in the experimental sessions but is not used in this paper. 12 A variety of real effort tasks have been used in the literature, such as solving anagrams (Charness and Villeval 2009), stuffing envelopes (Carpenter, Matthews, and Schirm 2010), counting zeros (Abeler et al. 2011), moving sliders (Gill and Prowse 2012), and threading nuts onto bolts (Alger et al. 2020). This is a manual task that does not require the use of computers. Given that the effort task is performed repeatedly, learning or fatigue may also arise in this experiment. Supplementary online appendix S2 shows the robustness of the main results to the ordering of games. The World Bank Economic Review 209 pressure to exert effort was eliminated by ensuring that individual decisions were unobservable by other participants (Perez-Truglia and Troiano 2018). In each experimental session, the subject pool was divided randomly into two groups of equal size, usually around 20 participants. One group was allocated to the Donor treatment, in which the subject’s additional income, if materialized, was added to the total amount to be given to the health center. The other group was allocated to the Recipient treatment: if the subject failed to generate additional income, Downloaded from https://academic.oup.com/wber/article/37/2/205/6997901 by Joint Bank-Fund library user on 04 September 2023 they received the equivalent amount from the resources to be given to the health center. The average payment was over 100 pesos, close to the daily minimum wage of an agricultural machinery operator.13 Supplementary online appendix S3 presents a simple theoretical model of effort choice, in which in- dividuals may be motivated to contribute to the public good because they may directly benefit from it, because of altruism towards other beneficiaries, and/or because of warm glow. This model shows that (a) effort in the NLPG versus the LPG game depends on the strength of the subject’s social ties in their own village versus the matched one (if such social ties are positive in the sense that they entail altruism towards other beneficiaries), and (b) effort is expected to be lower in the Recipient treatment than in the Donor one. The latter is due to the payoff structure, which implies a lower marginal benefit of effort in the Recipient treatment. To study a population of subsistence farmers with low exposure to taxes, the experiment was con- ducted in eight small rural localities (fewer than 2,000 inhabitants) in Southern Yucatán, Mexico, where such localities are common. Agriculture is highly heterogeneous across the state, but most of the pro- ducers, particularly in this region, operate at a small scale. Each experimental locality was matched with another similar locality for the NLPG. The characteristics of experimental subjects are comparable to those reported in the 2010 Mexican Population Census for their respective localities, with a few excep- tions. Our pool seems to be relatively more female, less single, and more bilingual (Maya and Spanish), compared to the census. This information, the match between the experimental localities and those that received the resources generated in the NLPG game, and other descriptive statistics are in supplementary online appendix S1. 3. Descriptive Evidence We present graphical evidence on the distributions of effort differences between the LPG and NLPG games. Effort is measured as completed levels. For the whole sample, the average effort difference between the two games is negligible (7.27 versus 7.41 levels completed; Wilcoxon test p = 0.3496).14 Figure 1 explores whether the distribution of the effort difference between those games varies system- atically by role in the experimental session (Donor versus Recipient). As shown in the left panel, such distribution is practically identical for both (Wilcoxon test p = 0.9943). The absence of a significant dif- ference is compatible with a constant marginal benefit of donations, and/or systematic differences between the preferences of subjects in the Recipient and Donor treatments. This is because in the Recipient treat- ment the smallest possible donation to the health center is 100n (where n is the total number of subjects), whereas in the Donor treatment it is the largest possible donation that equals 100n. Hence, if for subjects the marginal benefit of the donation is decreasing, Recipients would have been expected to exert a lower effort on average than Donors. 13 In Mexico, there is no legal minimum wage for agricultural workers, but the minimum wage for an operator of agricul- tural machinery was 93.6 pesos per day at the time of the experiment (http://www.conasami.gob.mx). 14 The higher average effort in the NLPG game than in the LPG game can be driven by learning (the NLPG game was always played last) and/or preferences (please see the theoretical section in supplementary online appendix S3). The interest of this study is, however, in the gender NLPG–LPG effort gap rather than in the absolute effort levels. As discussed later, the main results on this gap are robust to the inclusion of a learning proxy (the game ordering in the session). 210 Alger et al. Figure 1. Distribution of the Effort Difference by Role (Left) and by Gender (Right) .8 Difference: NLPG−LPG by role Difference: NLPG − LPG by gender .8 .6 .6 Density Density Downloaded from https://academic.oup.com/wber/article/37/2/205/6997901 by Joint Bank-Fund library user on 04 September 2023 .4 .4 .2 .2 0 0 −5 0 5 10 −5 0 5 10 Difference (NLPG − LPG) Effort for a given participant Difference (NLPG − LPG) Effort for a given participant Recipient, mean=0.15 Donor, mean=0.14 Male, mean=0.02 Female, mean=0.23 Source: Data for the study come from the lab-in-the-field experiment conducted by the authors in eight rural villages in the southern part of Yucatan, Mexico, in August 2014. Note: The figure shows the distribution of the individual effort difference between the NLPG and LPG games. Figure 2. Distributions of Effort Difference by Role and Gender Donors: NLPG − LPG by gender Recipients: NLPG − LPG by gender .8 .8 .6 .6 Density Density .4 .4 .2 .2 0 0 −5 0 5 10 −5 0 5 10 Difference (NLPG − LPG) Effort for a given participant Difference (NLPG − LPG) Effort for a given participant Donor Female, mean=0.34 Donor Male, mean=−0.16 Recipient Female, mean=0.11 Recipient Male, mean=0.19 Source: Data for the study come from the lab-in-the-field experiment conducted by the authors in eight rural villages in the southern part of Yucatan, Mexico, in August 2014. Note: The figure shows the distribution of the individual effort difference between the NLPG and LPG games. The right-hand panel of fig. 1 shows the distributions of the effort difference between games for men and women. The mean effort difference is 0.23 levels for women, whereas it is close to zero for men (0.02), so the gender gap in effort difference is positive and statistically significant (0.23 − 0.02 = 0.21 levels, Wilcoxon test p = 0.0365). This result suggests that women are more willing than men to exert effort in the NLPG game, relative to the LPG one. Figure 2 shows the distributions of the effort difference, separately by the interaction of role and gender. Female subjects seem to be more willing to exert effort than men in the NLPG versus the LPG game in both roles, as discussed before, but this gender gap is positive and significant, for Donors (0.5 levels, Wilcoxon test p = 0.0019), whereas it is close to zero and not significant for Recipients (−0.08 levels, Wilcoxon test p = 0.9034). In sum, women seem to be more willing than men to exert effort in the NLPG game, which benefits the health center of the matched locality, compared to the LPG game, which benefits the health center in their own locality. Furthermore, this gender gap in effort between those two games seems to be larger in the Donor treatment, but does not arise in comparisons by role alone.15 15 Table S2.1 in the supplementary online appendix shows the mean and standard deviation of effort choices at each one of the games and these same patterns. The World Bank Economic Review 211 4. Main Results Empirical Specification To examine how the effort difference between the LPG and NLPG games varies with gender, the following specification was estimated: Effortig = α + β · Non-localg + δ · (Non-localg ∗ Femalei ) + λi + ig , (1) Downloaded from https://academic.oup.com/wber/article/37/2/205/6997901 by Joint Bank-Fund library user on 04 September 2023 where Effortig denotes the effort of subject i in game g; Non-localg is a dummy equal to 1 for the transfer game that benefits the health center of the matched locality, and 0 for the LPG game; the next term is the interaction of the Non-local dummy with a Female dummy; λi is a subject fixed effect, which controls for individual characteristics affecting effort that remained constant during the session; and ig is an error term. The coefficient β captures the mean difference in effort between the NLPG and LPG games for men, whereas β + δ captures the corresponding one for women. Thus, δ represents the gender gap in effort difference. Equation (1) is estimated separately for Donors and Recipients. In all estimations, the standard errors are clustered at the session level to account for any within-session correlation. Given the small number of clusters (8), p-values obtained using a wild cluster bootstrap (WCB) procedure, with Webb’s (2013) recommended weights, are also reported. Regression Results Table 1 presents the main results. The Non-local dummy alone is negative for Donors in column 1, and positive for Recipients in column 2, but none of them are statistically significant. Thus, male Donors and Recipients do not seem to put forth a significantly different effort in the two public good games. In contrast, the interaction of the Non-local dummy with the Female dummy is positive and significant at 1 percent for Donors in column 1, whereas it is small and not statistically significant for Recipients in column 2. On average, female Donors complete about half a level (0.508) more than male Donors in the NLPG, relative to the LPG game. This gender gap amounts to 7 percent of the mean effort in the NLPG game (7.41 levels). Table 1. Effort Exerted in NLPG and LPG Games (1) (2) Sample: Donors Recipients Non-local −0. 1667 0.1923 (0.097) (0.263) Non-local * Female 0.5088*** −0. 0747 (0.074) (0.301) Constant 7.2016*** 7.3417*** (0.048) (0.059) Observations 248 240 R2 0.109 0.014 Number of players 124 120 H0 : Non-local = 0 (WCB p-value) 0.050 0.549 H0 : Non-local * Female = 0 (WCB p-value) 0.001 0.856 Source: Data for the study come from the lab-in-the-field experiment conducted by the authors in eight rural villages in the southern part of Yucatan, Mexico, in August 2014. Note: Unit of observation is a player’s effort choice in a game. Reference: LPG game. Subject fixed effects included in all specifications. The p-values reported at the bottom of the table were obtained using a wild cluster bootstrap (WCB) with Webb’s weights. Robust standard errors in parentheses, cluster session. ***p < 0.01, **p < 0.05, *p < 0.1. 212 Alger et al. Thus, female Donors appear to be more willing than their male counterparts to exert effort when the transfer benefits the other village instead of their own. No comparable gender gap in relative effort is found for Recipients. As shown at the bottom of the table, the statistical significance of all the estimates in table 1 remains the same when using the WCB procedure to account for the small number of clusters, except for the Non-local dummy in column 1, which becomes significant at 5 percent. Table S2.2 shows that the main results are similar when replacing individual fixed effects with a combi- Downloaded from https://academic.oup.com/wber/article/37/2/205/6997901 by Joint Bank-Fund library user on 04 September 2023 nation of session fixed effects and individual controls. In all sessions, the last game was always the NLPG game, whereas the LPG game was either the second or the third game, depending on the session. To check whether the results are driven by learning, interactions with a dummy for the game order in the session were added to equation (1). Table S2.2 confirms that the main results hold when controlling for the or- der of the games. If anything, the positive gender gap in effort for Donors is a bit smaller in magnitude, suggesting ordering might explain some of it but not all. 5. What Explains the Gender Difference in Effort between the NLPG and LPG Games? This section analyzes whether the positive gender gap in the effort difference between games for Donors varies, with measures of informal giving and receiving, community involvement and trust, participation in government programs, strength of the participant’s network in their own and in the matched village, and other sociodemographic characteristics at the individual level. In particular, the interest is to see whether women’s relative willingness to exert effort in the NLPG versus the LPG game is related to their informal giving and receiving behavior. For this, dummy variables for whether the participant’s household gave or received any help to or from other households in the previous year are considered. Women could also be more willing to exert effort for others because they are more socially involved in the community and have higher trust in institutions, compared to men. For community involvement and trust, separate PCA indices were constructed. The community involvement index includes dummies for whether the respondent believes the people in their locality cooperate to solve their problems most of the time, and the number of organizations and community celebrations in which they participated in the last 12 months. The trust index includes dummies for whether the respondent believes (a) they can trust most people, (b) most people would treat them fairly, (c) effort is rewarded with higher income most of the time, (d) the government spends tax revenues adequately. Given the forced nature of the transfer in our experiment, we also consider variable (d) separately. The main results could also be driven by prosocial preferences, such as altruism (Becker 1974). If so, a subject’s effort towards the health center in their own locality, or the matched one, could be impacted by the strength of their network in each of them. We consider the following variables as proxies for the participant’s network in their own locality: dummies for whether the participant knows almost all or all of the other participants in the session, whether half or more of the session participants belong to their family, whether they share daily activities with half or more of them. The proxies for the strength of the participant’s network in the matched locality are dummies for whether they know the other locality, whether they have visited the other locality, and whether they have actually been in the health clinic of the matched locality. This part of the analysis also considers other standard sociodemographic characteristics of subjects, like age, dummies for having at least secondary education, for being married, for being born in the mu- nicipality where the session took place, number of siblings, and number of children. Participation in government programs could also affect women’s willingness to exert effort for the other village, either because of a crowding out or crowding in effect. Thus, we use dummies for whether someone in the par- ticipant’s household receives Procampo or Oportunidades (the largest cash transfer programs targeting The World Bank Economic Review 213 rural localities), and the number of government programs in which their household participates.16 Finally, the analysis explores whether the gap varies by household wealth by using dummies for having an asset index below the 25th percentile or above the 75th percentile. For each of these variables, the empirical specification is the following: Effortig = α + β ∗ Non-localg + γ1 ∗ Non-localg ∗ Femalei Downloaded from https://academic.oup.com/wber/article/37/2/205/6997901 by Joint Bank-Fund library user on 04 September 2023 + γ2 ∗ Non-localg ∗ Variablei + γ3 ∗ Non-localg ∗ Femalei ∗ Variablei + δi + ig , (2) where most terms are defined as above. The coefficient γ 1 measures the part of the gender gap in the relative effort exerted in the LPG versus the NLPG games that is common to all women, whereas γ 3 measures the part that varies with the value of Variablei , the particular characteristic examined in each specification. Table S2.3 shows the unconditional mean differences by gender for all the variables used in the triple interactions of equation (2). Most of these differences are not statistically significant. Informal giving and receiving patterns (Panel A), the participant’s trust in government (Panel B) and strength of their social networks in both localities (Panel C) do not seem to differ by gender. Conversely, on average, women seem to be significantly more socially involved in their community, but also display less trust towards other people, compared to men (Panel B). Most of the unconditional mean differences in standard sociodemographic characteristics by gender are not statistically significant (Panel D), except for the probability of having at least some secondary education, the probability of being married, and the participant’s number of children. About 48 percent of men have at least some secondary education, compared to only 26 percent of women, and this difference is significant at 1 percent. Women are 10 percentage points more likely to be married than men, and this difference is marginally significant. On average, female participants have more children (3.8) than male ones (3.0), and this difference is significant at 5 percent. Note that if women care more about their own children’s welfare than men, they could be more willing to exert effort in the LPG game. However, given that the additional earnings in the NLPG game do not benefit the subjects’ own children directly, this discrepancy in parental altruism could hardly explain the gender gap in relative effort for Donors. Tables 2 and 3 show the estimation results for equation (2) using some of the above variables. For brevity, the tables report results only for informal giving and receiving, community involvement, trust, participation in government programs, and household wealth. Results for other social network and sociodemographic variables are in supplementary online appendix S2. Overall, no significant patterns explain the main results in these excluded variables. In table 2, most of the main results remain after including measures of informal giving and receiv- ing, and the community involvement and trust indices (columns 1–4). In those columns, the estimate for Non-localg ∗ Female for Donors is positive after including each of these variables, and mostly statistically significant at conventional levels (Panel A). For Donors, the estimates of the triple interactions are not significant in columns 1–4, suggesting that the gender gap in the effort difference between the NLPG and LPG games does not vary with these indicators. The only exception is the triple interaction with the dummy for having some trust in government spending taxes adequately (column 5), which is positive but marginally significant, whereas the estimate of γ 1 in that column turns insignificant. This evidence suggests that, for Donors, most of the positive gender gap in effort between the two public good games could be due to the behavior of women who have some trust in the government. In Panel B, most of the estimates for Non-localg ∗ Female are not significant for Recipients, as in the main results. Some triple interactions are marginally significant, suggesting that women in this role 16 In the sample, about 62 and 85 percent of subjects report that their household receives support from Procampo and Oportunidades, respectively. On average, the subjects’ households receive support from 2.5 government programs. 214 Alger et al. Table 2. Interaction with Informal Giving and Receiving, Trust, and Community Involvement Panel A: Donors (1) (2) (3) (4) (5) Variable interacted: HH giving HH received Community index Trust index Trust govt. Non-local 0.0000 0.0455 −0. 184* −0. 1400 0.0833 Downloaded from https://academic.oup.com/wber/article/37/2/205/6997901 by Joint Bank-Fund library user on 04 September 2023 (0.155) (0.213) (0.092) (0.107) (0.235) Non-local * Female 0.3571** 0.3260 0.538*** 0.486*** 0.1944 (0.143) (0.185) (0.080) (0.063) (0.162) Non-local * Variable −0. 0690 −0. 3916 −0. 083 −0. 117** −0. 3333 (0.196) (0.278) (0.171) (0.035) (0.196) Non-local * Female * Variable 0.0354 0.3202 0.039 0.087 0.4240* (0.170) (0.318) (0.178) (0.090) (0.220) Constant 7.1639*** 7.2033*** 7.202*** 7.203*** 7.2033*** (0.054) (0.049) (0.049) (0.049) (0.048) Observations 244 246 248 246 246 R2 0.133 0.125 0.115 0.122 0.122 Number of players 122 123 124 124 123 Panel B: Recipients (1) (2) (3) (4) (5) Variable interacted: HH giving HH received Community index Trust index Trust govt. Non-local 0.6923 0.4286 0.2200 0.1580 −0. 8000 (0.418) (0.542) (0.234) (0.317) (0.582) Non-local * Female −0. 6611 −0. 2394 −0. 1020 −0. 1260 1.1684* (0.577) (0.549) (0.274) (0.408) (0.597) Non-local * Variable −1. 0673** −0. 2562 0.1730 0.1140 1.2286* (0.434) (0.584) (0.127) (0.232) (0.639) Non-local * Female * Variable 1.2482* 0.0670 −0. 2030 -0.392* −1. 5766* (0.598) (0.431) (0.150) (0.191) (0.709) Constant 7.3304*** 7.2931*** 7.3420*** 7.3420*** 7.3417*** (0.049) (0.046) (0.056) (0.055) (0.044) Observations 230 232 240 240 240 R2 0.088 0.035 0.025 0.049 0.084 Number of players 115 116 120 120 120 Source: Data for the study come from the lab-in-the-field experiment conducted by the authors in eight rural villages in the southern part of Yucatan, Mexico, in August 2014. Note: Unit of observation is a player’s effort choice in a game. Reference: LPG game. Subject fixed effects included in all specifications. As measures of giving and receiving patterns, we use dummy variables for whether the participant’s household gave any help to other households in the previous year (column 1), whether help was given to someone in the respondent’s locality or in another locality (column 2). An index of community involvement measures based on cooperation, whether the respondent believes the people in their locality cooperate to solve their problems most of the time, the number of organizations in which the respondent participated in the last 12 months (social, political, religious), and the number of community celebrations in which they participate (column 3). An index of community trust that measures whether player can trust most people, whether most people would treat them fairly, and whether effort is rewarded with higher income most of the time (column 4). Dummies for whether they have some confidence that the government will adequately spend tax revenues, and for whether they have high confidence on the same issue (column 5). Robust standard errors in parentheses, cluster session. ***p < 0.01, **p < 0.05, *p < 0.1. provided relatively more effort in the NLPG versus the LPG game if, for instance, their household gave help, and less the higher their trust is overall and in the government in particular. The estimates for participation in government programs and household wealth are in table 3. For Donors, Panel A shows that the positive gender gap in effort between the two public good games is due to women who participate in the Oportunidades program, a conditional cash transfer program for poor rural households. In column 1, γ 1 is not significant, and it is smaller than the corresponding estimates in other columns, whereas γ 3 is positive and significant at 10 percent. Oportunidades, formerly known as Progresa, The World Bank Economic Review 215 Table 3. Interaction with Participation in Government Programs Panel A: Donors (1) (2) (3) (4) Variables interacted: Oportunidades Elderly pension Procampo Count programs Non-local −0.1000 −0.1628 −0.2000 −0.4590** Downloaded from https://academic.oup.com/wber/article/37/2/205/6997901 by Joint Bank-Fund library user on 04 September 2023 (0.173) (0.087) (0.149) (0.141) Non-local * Female 0.1000 0.5128*** 0.7312*** 0.9667*** (0.173) (0.100) (0.085) (0.168) Non-local * Variable −0.0842 −0.0372 0.0571 0.1362** (0.152) (0.135) (0.143) (0.041) Non-local * Female * Variable 0.4556* −0.0003 −0.3838** −0.2028** (0.224) (0.268) (0.115) (0.068) Constant 7.2016*** 7.2016*** 7.2016*** 7.2016*** (0.047) (0.047) (0.048) (0.047) Observations 248 248 248 248 R2 0.118 0.109 0.130 0.133 Number of players 124 124 124 124 Panel B: Recipients (1) (2) (3) (4) Variables interacted: Oportunidades Elderly pension Procampo Count programs Non-local 0.2500 0.3684 −0.0714 0.0639 (0.165) (0.385) (0.109) (0.356) Non-local * Female −0.2500 −0.2498 0.1914 0.0923 (0.163) (0.429) (0.365) (0.519) Non-local * Variable −0.0682 −0.6541 0.3609 0.0439 (0.195) (0.463) (0.257) (0.043) Non-local * Female * Variable 0.2085 0.6466 −0.3646 −0.0596 (0.352) (0.547) (0.321) (0.109) Constant 7.3417*** 7.3417*** 7.3417*** 7.3417*** (0.059) (0.062) (0.056) (0.057) Observations 240 240 240 240 R2 0.015 0.036 0.021 0.016 Number of players 120 120 120 120 Source: Data for the study come from the lab-in-the-field experiment conducted by the authors in eight rural villages in the southern part of Yucatan, Mexico, in August 2014. Note: Unit of observation is a player’s effort choice in a game. Reference: LPG game. Subject fixed effects included in all specifications. Dummies for whether someone in the participant’s household receives Oportunidades (column 1), Rural elderly program (column 2), or Procampo (column 3), and the number of government programs that the participant’s household receives benefits from (column 4) are also considered. With respect to income, dummies for having a household asset index in lower than the 25th percentile or higher than the 75th percentile are included (columns 5 and 6, respectively). Robust standard errors in parentheses, cluster session. ***p < 0.01, **p < 0.05, *p < 0.1. has two features that are consistent with these results. First, the transfer from the program is paid to the mother or most senior woman in the household. Second, part of the transfer is conditioned on complying with scheduled health checks for children and adults in beneficiary households. Thus, Oportunidades beneficiaries, women who use health services relatively often, are those who are more willing to exert effort for the clinic at the matched locality, compared to men. Oportunidades also has other activities that female beneficiaries do together, which could further strengthen their social ties. No similar patterns are observed for the receipt of noncontributory pensions, which are unconditional cash transfers paid to the rural elderly, regardless of their gender (column 8). Column 9 shows that the positive effort gender gap is smaller for Donors who participate in the Procampo program. These are probably women whose household owns land, because this program pays a transfer per hectare of land held and sown in a given agricultural cycle. 216 Alger et al. Column 10 shows that participation in more government programs decreases the relative willingness of women to exert effort for the other locality, compared to men. For Recipients, Panel B shows once again no significant estimates for γ 1 and γ 3 . Finally, the interactions with an indicator for low or high household wealth are not significant in both panels. In sum, the positive gender gap in effort between the two public good games does not vary with infor- mal giving and receiving patterns, measures of community involvement and trust, proxies for the strength Downloaded from https://academic.oup.com/wber/article/37/2/205/6997901 by Joint Bank-Fund library user on 04 September 2023 of the local and non-local networks of players, or other standard sociodemographic characteristics at the individual level, except for the trust of the participant in the way the government spends taxes, and the participation in government programs. Specifically, the Oportunidades result suggests that some govern- ment programs could actually foster social capital and a willingness to contribute to a public good, instead of crowding them out. Most of the literature on anti-poverty and social insurance programs has tended to focus on the substitutability between government and private interventions. However, as discussed in the introduction, this research adds to other studies that, instead, find some positive effects of public cash transfers on social capital, trust in the government, and public good contributions in developing countries. Are Women Aware of Their Higher Willingness to Exert Effort for Others? The final piece of analysis examines whether the gender of other participants affects the willingness of women to exert effort in the LPG versus NLPG game. This involves estimating equation (2), including a triple interaction of Non-local and Female with each of these two variables: the share of women in the experimental session and in the room. The first measure pools Donors and Recipients in a given session when calculating the share, whereas the second captures the gender composition within each of these treatments. Sessions had on average 59 percent of female participants, with a minimum of 43 percent and a maximum of 74 percent, and the distribution in the two treatments in each session is similar. Table 4 continues to show a positive and significant gender gap in effort for Donors, which decreases significantly with the share of women in the session, but not in the room. These findings suggest that women reduce their individual effort the more female their environment is. A possible explanation is that they know of their higher willingness to exert effort for a public good that does not benefit them directly, compared to men. In this case, women might react more to the gender composition of the whole session and not of the room because they were more aware of the former. Recall that participants hung out for a while outside the room, while the research team waited for participants to arrive and carried out the final preparations for the session, and so they had plenty of time to look at each other. After this, they were randomly divided into two groups and the first one immediately entered the room, where participants were all the time facing the front of the room and had dividers between them to ensure privacy. 6. Concluding Discussion Are women more inclined than men to contribute to public goods from which they do not benefit person- ally? The data from a lab-in-the-field experiment, in which subjects choose to exert real effort to increase the contributions to health centers existing in other localities and their own, suggest that this may be the case. This study finds a positive and statistically significant gender gap in the difference between effort in the LPG and NLPG games for Donors, whose effort could have the effect of adding money to the amount given to the health centers. This implies that female Donors are more willing than their male counterparts to exert effort when it benefits the health center of another locality, compared to when it benefits the one in their own locality. In contrast, the gender effort gap between those two games for Recipients, whose effort could have the effect of avoiding to withdraw money from the contributions to health centers, is close to zero and not statistically significant. Although this work does not formally establish the external validity of these results, the gender gap in relative effort for Donors is mostly due to women who report The World Bank Economic Review 217 Table 4. Interaction with Share of Women in the Room and Session (1) (2) (3) (4) Session composition Room composition Donors Recipients Donors Recipients Downloaded from https://academic.oup.com/wber/article/37/2/205/6997901 by Joint Bank-Fund library user on 04 September 2023 Non-local game −0.9531 1.2589 −0.7279 0.6721 (0.611) (1.674) (0.431) (0.551) Non-local game * Female 1.2443*** −0.6412 0.7856 −0.2909 (0.279) (2.046) (0.442) (0.921) Non-local * Session_share 1.3928 −1. 9378 — — (0.967) (2.839) Non-local * Female * Session_share −1. 3069** 1.1110 — — (0.432) (3.282) Non-local * Room_share — — 1.0013 −0.9850 (0.642) (1.017) Non-local * Female * Room_share — — −0.5528 0.5548 (0.606) (1.355) Constant 7.2016*** 7.3417*** 7.2016*** 7.3417*** (0.047) (0.055) (0.046) (0.055) Observations 248 240 248 240 R2 0.119 0.026 0.120 0.023 Number of players 124 120 124 120 Source: Data for the study come from the lab-in-the-field experiment conducted by the authors in eight rural villages in the southern part of Yucatan, Mexico, in August 2014. Note: Unit of observation is a player’s effort choice in a game. Reference: LPG game. Subject fixed effects included in all specifications. The variable Session_share denotes the share of female participants in the session. The variable Room_share denotes the share of females in the room, i.e. in the Donor or Recipient treatment. Given that allocation to treatments is random, both shares can differ in a given session. Robust standard errors in parentheses, cluster session. ***p < 0.01, **p < 0.05, *p < 0.1. having at least some trust in the way the government spends taxes, and to women who benefit from a government program that targets women and fosters healthcare use (see Section 5), which lends credibility to them. As mentioned in the introduction, the evidence on gender differences regarding monetary contributions in the classic public good game in laboratory experiments is mixed. By contrast, studies that examine gender differences in the willingness to pay taxes show that women’s willingness exceeds that of men more consistently. This study complements these two strands of the literature by using a real-effort experiment in which proceeds went to actual public goods. As a caveat, the motivation of an individual for complying with the tax code might be different from the one for voluntarily exerting effort to contribute to a public good they do not benefit from personally. The fear of being caught and punished might play a role in the former, whereas not in the latter, especially when the effort choice is unobservable to others, as in this study’s experiment. Thus, from the results, it cannot generally be concluded that women are more willing than men to pay taxes in the studied population. Nevertheless, the evidence suggests that it would be interesting to disentangle the relative contribution of these two motivations to the gender difference in tax compliance found in the literature. Furthermore, while discrepancies in the behavior of women and men may be driven by social norms and culture (Fernández 2013), evolutionary theory suggests that they may also stem from deeper differences (Eswaran 2014). These findings are in line with the evolutionary theory which predicts that in- versus out-group attitudes are expected to differ between men and women, with men being expected to favor in-group members to a larger extent than women (Manson et al. 1991; Low 1993; Micheletti, Ruxton, and Gardner 2018). 218 Alger et al. However, the finding that the gender effort gap is mostly due to women who trust the government and/or benefit from a government program that targets women and fosters healthcare use, is also in line with a recent theory according to which reciprocity can drive willingness to pay taxes (Besley 2020). Further research on the competing explanations for gender differences in behavior is necessary, including those concerning their willingness to contribute towards real public goods (as Donors) or benefit from them (as Recipients), as suggested by the main results in this work. Downloaded from https://academic.oup.com/wber/article/37/2/205/6997901 by Joint Bank-Fund library user on 04 September 2023 Data Availability The data underlying this article are available in the article and in its online supplementary material. Funding This work was supported by the Agence Nationale de la Recherche (Chaire d’Excellence ANR-12-CHEX- 0012-01 to Ingela Alger; IAST grant ANR-17-EURE-0010 IAST, Investissements d’Avenir program). References Abeler, J., A. Falk, L. Goette, and D. Huffman. 2011. “Reference Points and Effort Provision.” American Economic Review 101 (2): 470–92. Adams, R. B., and P. Funk. 2012. “Beyond the Glass Ceiling: Does Gender Matter?” Management Science 58 (2): 219–35. Adato, M. 2000. “The Impact of Progresa on Community Social Relationships.” Technical report. Adena, M., R. Hakimov, and S. Huck. 2020. “Charitable Giving by the Poor: A Field Experiment in Kyrgyzstan.” WZB Discussion Paper SP II 2019-305r, Berlin. Alger, I., and D. Cox. 2013. “ The Evolution of Altruistic Preferences: Mothers versus Fathers. ” Review of Economics of the Household 11 (3): 421–46. Alger, I., L. Juarez, M. Juarez-Torres, and J. Miquel-Florensa. 2020. “Do Informal Transfers Induce Lower Efforts? Evidence from Lab-in-the-Field Experiments in Rural Mexico.” Economic Development and Cultural Change 69 (1): 107–71. Alt, M., C. Gallier, A. Schlüter, K. Nelson, and E. Anggraini. 2018. “Giving To versus Taking From In- and Out-Group Members.” Games 9 (3): 57. Andreoni, J. 1990. “Impure Altruism and Donations to Public Goods: A Theory of Warm-Glow Giving. ” Economic Journal 100 (401): 464–77. Angelucci, M., G. De Giorgi, M. Rangel, and I. Rasul. 2009. “Village Economies and the Structure of Extended Family Networks.” The BE Journal of Economic Analysis & Policy 9 (1). ———. 2010. “Family Networks and School Enrolment: Evidence from a Randomized Social Experiment.” Journal of Public Economics 94 (3–4): 197–221. Attanasio, O., S. Polania-Reyes, and L. Pellerano. 2015. “Building Social Capital: Conditional Cash Transfers and Cooperation.” Journal of Economic Behavior & Organization 118: 22–39. Beaman, L., E. Duflo, R. Pande, and P. Topalova. 2012. “Female Leadership Raises Aspirations and Educational Attainment for Girls: A Policy Experiment in India.” Science 335 (6068): 582–86. Becker, G. S. 1974. “A Theory of Social Interactions.” Journal of Political Economy 82 (6): 1063–93. Bertrand, M. 2011. “New Perspectives on Gender.” In Handbook of Labor Economics, 1543–1590. Elsevier. Besley, T. 2020. “State Capacity, Reciprocity, and the Social Contract.” Econometrica 88 (4): 1307–35. Besley, T., and T. Persson. 2014. “Why Do Developing Countries Tax So Little?” Journal of Economic Perspectives 28 (4): 99–120. Bhalotra, S., and I. Clots-Figueras. 2014. “Health and the Political Agency of Women.” American Economic Journal: Economic Policy 6 (2): 164–97. The World Bank Economic Review 219 Brown-Kruse, J., and D. Hummels. 1993. “Gender Effects in Laboratory Public Goods Contribution: Do Indi- viduals Put Their Money Where Their Mouth Is?” Journal of Economic Behavior & Organization 22 (3): 255–67. Brüggen, A., and M. Strobel. 2007. “Real Effort versus Chosen Effort in Experiments.” Economics Letters 96 (2): 232–36. Bruner, D. M., J. D’Attoma, and S. Steinmo. 2017. “The Role of Gender in the Provision of Public Goods through Tax Downloaded from https://academic.oup.com/wber/article/37/2/205/6997901 by Joint Bank-Fund library user on 04 September 2023 Compliance.” Journal of Behavioral and Experimental Economics 71: 45–55. Candelo, N., C. Eckel, and C. Johnson. 2018, Oct. “Social Distance Matters in Dictator Games: Evidence from 11 Mexican Villages.” Games 9 (4): 77. Carpenter, J., P. H. Matthews, and J. Schirm. 2010. “Tournaments and Office Politics: Evidence from a Real Effort Experiment.” American Economic Review 100 (1): 504–17. Carpenter, J., and E. Seki. 2011. “Do Social Preferences Increase Productivity? Field Experimental Evidence from Fishermen in Toyama Bay.” Economic Inquiry 49 (2): 612–30. Charness, G., and M.-C. Villeval. 2009. “Cooperation and Competition in Intergenerational Experiments in the Field and the Laboratory.” American Economic Review 99 (3): 956–78. Croson, R., and U. Gneezy. 2009. “Gender Differences in Preferences.” Journal of Economic Literature 47 (2): 448–74. D’Attoma, J. W., C. Volintiru, and A. Malezieux. 2020. “Gender, Social Value Orientation, and Tax Compliance.” CESifo Economic Studies 66 (3): 265–84. DellaVigna, S., J. A. List, U. Malmendier, and G. Rao. 2013. “The Importance of Being Marginal: Gender Differences in Generosity.” American Economic Review 103 (3): 586–90. Dupas, P., and J. Robinson. 2013. “Savings Constraints and Microenterprise Development: Evidence from a Field Experiment in Kenya.” American Economic Journal: Applied Economics 5 (1): 163–92. Eswaran, M. 2014. Why Gender Matters in Economics. Princeton, NJ: Princeton University Press. Evans, D. K., B. Holtemeyer, and K. Kosec. 2019. “Cash Transfers Increase Trust in Local Government.” World Development 114: 138–155. Fernández, R. 2013. “Cultural Change as Learning: The Evolution of Female Labor Force Participation over a Cen- tury.” American Economic Review 103 (1): 472–500. Field, E., A. Holland, and R. Pande et al. 2016. “Microcredit: Points of Promise.” Technical report. Gertler, P. J., H. A. Patrinos, and M. Rubio-Codina. 2012. “Empowering Parents to Improve Education: Evidence from Rural Mexico.” Journal of Development Economics 99 (1): 68–79. Gill, D., and V. Prowse. 2012. “A Structural Analysis of Disappointment Aversion in a Real Effort Competition.” American Economic Review 102 (1): 469–503. Keser, C., D. Masclet, and C. Montmarquette. 2020. “Labor Supply, Taxation and the Use of the Tax Revenues: A Real-Effort Experiment in Canada, France, and Germany.” Public Finance Review 48 (6): 714–50. Levitt, S. D., and J. A. List. 2007. “What Do Laboratory Experiments Measuring Social Preferences Reveal about the Real World?” Journal of Economic Perspectives 21 (2): 153–74. Lévy-Garboua, L., D. Masclet, and C. Montmarquette. 2009. “A Behavioral Laffer Curve: Emergence of a Social Norm of Fairness in a Real Effort Experiment.” Journal of Economic Psychology 30 (2): 147–61. Low, B. S. 1993. “An Evolutionary Perspective on War.” In Behavior, Culture, and Conflict in World Politics, edited by H. Jacobson and W. Zimmerman, 13–56. Ann Arbor, MI: University of Michigan Press. Lundberg, S. J., R. A. Pollak, and T. J. Wales. 1997. “Do Husbands and Wives Pool Their Resources? Evidence from the United Kingdom Child Benefit.” Journal of Human Resources 32 (3): 463–80. Manson, J. H., R. W. Wrangham, J. L. Boone, B. Chapais, R. Dunbar, C. R. Ember, and W. Irons. 1991. “Intergroup Aggression in Chimpanzees and Humans [and Comments and Replies].” Current Anthropology 32 (4): 369–90. Mascagni, G. 2018. “From the Lab to the Field: A Review of Tax Experiments.” Journal of Economic Surveys 32 (2): 273–301. Meghir, C., and D. Phillips. 2010. “Labour Supply and Taxes.” Dimensions of Tax Design: The Mirrlees Review 202–74. Micheletti, A. J., G. D. Ruxton, and A. Gardner. 2018. “Why War Is a Man’s Game.” Proceedings of the Royal Society B: Biological Sciences 285 (1884): 20180975. Millán, T. M., T. Barham, K. Macours, J. A. Maluccio, and M. Stampini. 2019. “Long-Term Impacts of Conditional Cash Transfers: Review of the Evidence.” World Bank Research Observer 34 (1): 119–59. 220 Alger et al. Mohdali, R., K. Isa, and S. H. Yusoff. 2014. “The Impact of Threat of Punishment on Tax Compliance and Non- compliance Attitudes in Malaysia.” Procedia-Social and Behavioral Sciences 164: 291–97. Nguyen, T. C., and M. Rieger. 2017. “Community-Driven Development and Social Capital: Evidence from Morocco.” World Development 91: 28–52. Niederle, M. 2017. “A Gender Agenda: A Progress Report on Competitiveness.” American Economic Review 107 (5): 115–19. Downloaded from https://academic.oup.com/wber/article/37/2/205/6997901 by Joint Bank-Fund library user on 04 September 2023 Nowell, C., and S. Tinkler. 1994. “The Influence of Gender on the Provision of a Public Good.” Journal of Economic Behavior & Organization 25 (1): 25–36. Perez-Truglia, R., and U. Troiano. 2018. “Shaming Tax Delinquents: Evidence from a Field Experiment in the United States.” Journal of Public Economics 167: 120–37. Schady, N., and J. Rosero. 2008. “Are Cash Transfers Made to Women Spent Like Other Sources Of Income?” Eco- nomics Letters 101 (3): 246–48. Sear, R., and R. Mace. 2008. “Who Keeps Children Alive? a Review of the Effects of Kin on Child Survival.” Evolution and Human Behavior 29 (1):1–18. Sillamaa, M. 1999. “How Work Effort Responds to Wage Taxation: An Experimental Test of a Zero Top Marginal Tax Rate.” Journal of Public Economics 73 (1): 125–34. Slemrod, J. 2007. “Cheating Ourselves: The Economics of Tax Evasion.” Journal of Economic Perspectives 21 (1): 25–48. Summers, L. H. 1989. “Some Simple Economics of Mandated Benefits.” American Economic Review 79 (2): 177–83. Sutter, M., and H. Weck-Hannemann. 2003. “Taxation and the Veil of Ignorance: A Real Effort Experiment on the Laffer Curve.” Public Choice 115 (1/2): 217–40. Swenson, C. W. 1988. “Taxpayer Behavior in Response to Taxation. An Experimental Analysis.” Journal of Accounting and Public Policy 7 (1): 1–28. Webb, Matthew D. 2013. “Reworking Wild Bootstrap Based Inference for Clustered Errors.” Queen’s Economics Department Working Paper.