Overcoming a Commitment Problem

Esther Duflo et al. (2008, 2011) documented the low take-up of fertilizer in Kenya, which they estimated had an average return of just under 50%. Since fertilizer had long been used in the survey area, how could the low take-up be reconciled with the large increases in yield from fertilizer? Just as biases limit Westerners’ investment in attractive pension plans, biases also limit profitable investment in fertilizer by farmers in Kenya. Take-up was increased by 47% to 70% by giving farmers at the time of harvest, when they have ready cash, the opportunity to commit to fertilizer purchase. The commitment helped farmers overcome procrastination problems, while minimally distorting the investment decisions of farmers who did not suffer from present bias. Small, time-limited reductions in the cost of buying fertilizer at the time of harvest induced increases in fertilizer use comparable to those induced by much larger price reductions later in the season.

Overcoming Present Bias

Abhijit Banerjee et al. (2010) document the underresponsiveness of parents in India to reliable, free vaccination programs for children. The researchers ran an experiment to increase take-up. Some parents of children under age 2 were given small nonfinancial rewards after each of the four visits to the clinic that were needed to complete the vaccination series that immunizes a child against multiple diseases. In the sparsely populated rural area of India where the experiment was conducted, the small rewards raised the percentage of children who completed the series of vaccines from 18% to 39%. Since a large part of the cost of immunization is a fixed cost of setting up the immunization camp, the intervention greatly reduced the average cost of fully immunizing a child: This cost fell from $56 to $28. Because the 39% proportion of children who were immunized still falls far short of what is needed to achieve herd immunity, Jishnu Das (2010) views the study as “a wake-up call for more experimentation and evaluation on the fundamental design of vaccination programmes” (p. 1258).

Increasing Willpower

Pascaline Dupas and Jonathan Robinson (2013) document that many households in Kenya do not have adequate savings to purchase bed nets and cover emergency medical costs. They find that the mere provision of a lockbox, together with a passbook on which a woman writes down what she wants to purchase with her savings, increases the amount she saves for health needs by 66%. The mechanism appears to be a kind of “mental accounting”: Having mentally committed to save a certain amount enables women to resist temptation and requests for help from friends. Thus, they save much more.

The book entitled Nudge (Thaler & Sunstein, 2008) was the spearhead of the first wave of behavioral economics. A nudge is a policy that changes the way options are presented or designed in order to make it easier for individuals to make the rational choice. One kind of nudge is a commitment device. Since commitment devices may help people who are present biased to save, there might be a demand for commitment devices to increase savings. Nava Ashraf et al. (2006) offered a commitment savings account to a random subset of participants through a bank in the Philippines. Under the commitment, the individuals had restricted access to their savings until they net their savings goal or until a target date. Twenty-eight percent of respondents took up this offer. A year later, the individuals who had been offered commitment savings had 82% higher savings balances than those in the control group.

Another mechanism that has been used to “nudge” people to make rational decisions to save is to make saving for retirement be the default option for employees at a firm. (The default option is the outcome that people get if they make no explicit decision over a set of options.) An intervention of this kind in the United States that gained a lot of attention is Save More Tomorrow. It asks employees to commit to save for retirement a small portion of their earnings, and it links planned increases in savings rates to future pay raises. At the first company that implemented Save More Tomorrow, most of the employees joined, and they ended up quadrupling their saving rate in 4 years (Benartzi & Thaler, 2013). As of this writing, over 15 million employees in the United States participate in Save More Tomorrow.

An Influence on Female Labor Force Participation: The Plough in Preindustrial Agriculture

Preindustrial agriculture relied primarily on either shifting cultivation or plough cultivation. Unlike the hoe or digging stick used to prepare the soil in shifting cultivation, the plough requires significant upper body strength either to pull it or control the animal that pulls it. In areas that were topographically well suited to use of the plough (i.e., well suited to grow wheat, barley, and rye rather than maize, sorghum, and millet), the plough created gendered occupations: Men specialized in agriculture and women specialized in domestic activities.

Alesina et al. (2013) find evidence that this historical social pattern creates in the modern world a very specific aspect of gender inequality: It lowers female participation in the labor market, entrepreneurship, and national political positions. In regions that are topographically well suited to use of the plough, compared to regions that are not, female labor force participation in the year 2000 was more than 10 percentage points lower. There was also lower participation of women in firm ownership and in national politics. The qualitative results extend to second-generation immigrants in the United States and Europe from historically plough-using regions. This is strong evidence that cultural beliefs and values, based on historically gendered occupations, influence attitudes today of the “natural” or “appropriate” role of women. A possible interpretation is that the historical experience of gendered occupations changed the mental models of gender—the “mini-theory” of the kinds of work that it is appropriate for men and women to do.

An Influence on Mistrust: Slave Exports in the Atlantic Slave Trade

For nearly 500 years, there was a slave trade in Africa. To protect themselves from enslavement, Africans bought guns. To purchase the guns, many kidnapped or tricked others and sold them to slave traders. It is an “historical fact that by the end of the slave trade, it was not uncommon for individuals to be sold into slavery by neighbors, friends, and family members” (Nunn & Wantchekon, 2011, p. 3222). Nathan Nunn and Leonard Wantchekon show that heavy targeting of some ethnic groups by the slave trade in the past 500 years adversely affects the modern-day level of trust in these the ethnic groups in their relatives, neighbors, and local government. This is measured by responses to specific questions in the 2005 Afrobarometer survey. The average level of trust is very low. For example, 18% of all respondents report that they trust their relatives only a little, and only 30% report that they trust their elected local government council a lot (Nunn and Wantchekon, online Appendix Table 12). The heavy targeting of some ethnic groups has also lowered per capita incomes in their countries today (Nunn, 2008). The effect of the slave trade on mistrust arises much more from the level of slave exports from an individual’s own ethnic group than from the level of slave exports in the location in which they currently live (p. 3249). Both channels influence trust, but the slave exports from the individual’s own ethnic groups affect their level of trust at least twice as much. This is supportive evidence that individuals are shaped by the beliefs of their cultural groups. The slave exports from each ethnic group damaged the group’s social fabric and shaped its cultural mental models, and the mental models were passed down across generations. Embedded in the mental models are norms of mistrust.

An Influence on Civic Culture: The Experience of Self-Government in the Middle Ages

In 1970, a national law in Italy created from scratch 15 regional governments with identical formal rules. Many of the new governments in north Italy became strong, responsive, and effective. Most in the south did not. There is evidence that the divergence can be explained by regional differences in the strength of civic-spirited behavior that date back to the late Middle Ages (Guiso et al., 2016; Putnam, 1993). Around the turn of the last millennium, southern regions of Italy were brought under the Norman Empire. In contrast, in north Italy, many cities formed self-governing communes when the Holy Roman Empire weakened. People in municipalities with a history as self-governing communes appear to have a different psychology. Luigi Guiso and colleagues found differences between municipalities with and without a history as city-states on the basis of responses to survey questions posed to students in Italy.

Here is one of the questions: “Your performance is so bad that you have to do the assignment over. How do you explain this?” The set of possible answers includes the following: No one helped me; I was unlucky; it was difficult; I lack ability; and I did not put effort on it. The answer that indicates high self-efficacy is the last one. Individuals with stronger self-efficacy expect their own effort to determine the outcome.

Consistent with the higher self-efficacy scores of pupils in municipalities with a history as self-governing communes, such cities have a greater number of nonprofit organizations, a greater probability of having an organ donation program (a direct measure of individuals’ internalization of the common good and ability to act on it), lower frequency of cheating in a national mathematics exam that fifth graders take, more resistance to the Nazis during World War II, and higher participation of citizens in referenda. The authors suggest that the institutional shock of being a medieval city-state influenced the regions’ psyches (p. 1434) in a way that made it possible for the citizens today to form effective regional governments.

An Influence on Discrimination by Judges: Terrorism in the Neighborhood of the Courts

Moses Shayo and Asaf Zussman (2011, 2017) find a persistent effect of a period of intense Arab-Jewish conflict in Israel in 2000–2004 on discrimination in the aftermath of the violence, 2007–2010. Most of the ethnic violence took the form of suicide bombings by Arabs, rather than violence by the local population. Shayo and Zussman created a measure of ethnic bias from decisions made by judges in Israel’s small claims court in cases that involved either an Arab plaintiff suing a Jewish defendant or a Jewish plaintiff suing an Arab defendant.¹

Judges in Israel’s small claims court are assigned to cases in an effectively random process. The litigants present their case, which are typically “fender-bender” accidents. The judge alone makes the decision. This is a situation where an effectively randomly assigned judge allocates money between two individuals, one of whom belongs to their ethnic group, while the other one does not.

Shayo and Zussman find little ingroup bias in small claims court before the outbreak of political violence in the neighborhood of the court. But during and after a period of intense conflict in the neighborhood of the court, judges from each ethnic group (Jewish and Arab) were more likely to make judgments in a way that took into account the welfare of members of their ethnic group. Jewish judges were 15 to 20 percentage points less likely than Arab judges to accept claims filed by Arab plaintiffs. The persistence of discrimination was not due to the judges’ personal exposure to violence but rather to violence within the area of the court as many as several years before. Thus, what elicited the discrimination and likely made it salient was the common knowledge of the events and the culture of the court.

Coaching

In many cases, boys growing up in violent neighborhoods learn to “act tough” when others assert authority in order to develop a reputation that deters potential predators. Having learned this strategy, males may automatically respond with aggression to all assertions of authority. In places where social control is strong, such a response is not adaptive: in particular, it leads to high rates of suspensions from high school, high rates of avoidable violence, and high risk of incarceration.

Interventions with at-risk, disadvantaged young men in high schools and prisons in the Chicago area helped participants reflect on whether aggression was the best response in a wide variety of situations (Heller et al., 2017). Coaches presented participants with many situations and, in each case, asked participants to consider a variety of interpretations of the situation and to practice ways to interact. In one exercise, called “the Fist,” participants were divided into pairs. One member of each pair was given a ball, and the other was told he had 30 seconds to get the ball from his partner. Many boys took the ball by force from their partners. The counselor then asked questions to show that their partners would have been willing to give up the ball if they had been asked calmly to do that. A typical response was, “I would have given it: it’s just a stupid ball” (p. 3). In this and many other exercises, the participants practiced overcoming automatic “fast” thinking with “slow” reflection. They were never told that aggression was wrong but rather that situations varied and aggression was not always an appropriate way to interact. The coaching programs substantially decreased suspensions from high school, increased high school graduation rates, and reduced violent-crime arrests. For the participants in prison, the programs decreased recidivism.

Sara Heller and colleagues have limited information on the mechanisms driving the change in behavior. Their preferred interpretation is that the participants learned to “think slow.” This interpretation falls within the first wave of behavioral economics. But as Kahneman (2011) emphasizes, to think slow is hard to maintain in the uncountably many social interactions people have in their everyday lives. A more plausible interpretation may be that through practice, the participants learned a new mental model, and it was activated in situations where social control was strong. Participants learned to make a distinction between dangerous situations, where aggressive responses were appropriate, and safe situations, where aggression was not appropriate. To use the language that describes mental models, participants gained from the activities that they did in the intervention an epistemological resource, and it mediated their experience of the world. It led them automatically to respond cooperatively in settings that they saw were probably safe. Instead of learning to think “slow,” they learned a new way to think “fast.”

A related intervention, evaluated by Blattman et al. (2017, 2022), coached criminally engaged men in Liberia in ways that led them to reduce reliance on harmful, automatic patterns of thinking linked to their criminal identity. They developed a different representation of themselves (a different self-schema). The intervention reduced the participants’ criminal and violent activities. Based on self-reports, the effect persisted 10 years later.

Collaborative Intergroup Contact

In India, individuals in the lowest castes were historically called “untouchables.” They were represented as innately polluted. At independence, India abolished untouchability, but the “pollution barrier” between former untouchable castes and all others remains strong in North India today. This barrier depresses the performance of the low caste in school (Hoff & Pandey, 2006, 2014). This barrier also causes gains that could be realized from trade between high-caste and low-caste individuals to be foregone: Anderson (2011) finds that low-caste farmers have agricultural yields 45% higher if they reside in a village where the majority of groundwater sellers are of low caste rather than of high caste, since the latter generally refuse to sell water to low-caste farmers. The obstacles to low castes to develop a positive group identity may explain the much lower willingness of low-caste than high-caste individuals to punish norm violators who hurt members of their own subcaste; this makes cooperation within a low compared to a high subcaste more difficult to sustain (Hoff et al., 2011). Social punishment is believed to play a large role in the enforcement of contract and property rights and, thus, in economic development.

Gordon Allport (1954, p. 276, cited in Lowe, 2021) hypothesized that intergroup collaborative contacts, in which people do things together for a common goal, changes attitudes that the groups had about themselves and the other groups. He argued that “the principle is clearly illustrated in the multi-ethnic athletic team . . . It is the cooperative striving for the goal that engenders solidarity.” Matt Lowe (2021) tested the hypothesis in a field experiment in North India; he recruited volunteers from young men in mixed-caste villages to play cricket. He created cricket leagues of teams that were either mixed low and high caste or only high caste or only low caste. The mixed-caste teams were the treatment group (with intergroupcollaboration) and the single-caste-status teams were the control group (without intergroup collaboration). Each team played eight games over the course of the month-long experiment. Lowe shows that collaborative contact (playing on a team with members from both high and low castes) reduces caste prejudice in young men, while competitive contact (playing against others of a different caste status) increases caste prejudice. Having all other-caste teammates instead of none increased by one the number of friendships between high- and low-caste men in the cricket leagues. Having most rather than few opponents in cricket matches be of the other caste status decreases, on average, by 3.5 the number of other-caste-status friends that a participant has in the league.

After the cricket matches ended, Lowe implemented a second field experiment with the same participants. He gave each person two left flip-flops or two right flip-flops and two left-handed gloves or two right-handed gloves. There was a monetary incentive to engage in trade with other participants in order to get correctly matched flip-flops and gloves. Being on a team with both low- and high-caste individuals increased by up to 21% the participation in trade between low- and high-caste individuals. Thus, Lowe’s experiments strongly bear out Alport’s hypothesis that intergroup collaborative contact reduces prejudice.

Supporting evidence are the results of the natural experiment in New Delhi evaluated by Gautam Rao (2019). He shows that the integration of rich and poor children in schools from the age of 4 makes the rich children more inequity averse and less discriminatory toward the poor. The change in rich students’ preferences occurs only if they have regular contact with poor students in a study group, to which students were as-good-as-randomly assigned. No effect on rich students’ prejudice against the poor arises merely from being in a classroom of which 20% of the students are poor.

Affirmative Action and Other Strategies for Changing the Status of Groups

Despite formalequality across men and women in political rights in most countries, women are underrepresented in positions of political leadership. To address this, many countries have adopted affirmative action quotas for women in certain elected positions. In one case—India—reservations in the elected position of village chief for women were randomly assigned to local governments. This makes it possible to causally assess the impact.

A study in the state of West Bengal, India, found that 7 years of experience of living under a woman village chief erased male villagers’ prejudice against women leaders by many measures—an Implicit Association Test, the evaluation of political speeches in the Goldberg paradigm, and the assessment of the quality of actual female village chiefs (Beaman et al., 2009). In villages that had been exposed to women village chiefs for at least 7 years, parents’ aspirations for their teenage daughters and the daughters’ aspirations for themselves were higher, and girls went to school somewhat longer and did somewhat fewer hours of housework (Beaman et al., 2012). Affirmative action for women in the local village councils made female victims of crime more willing to report the crimes to the police and made the police more willing to record the reports, even though the local village council has no formal authority over the police (Iyer et al., 2012).

All these changes in attitudes and behavior suggest that the creation of exemplars of women in high-status political positions changed widely shared mental models of women. Supportive evidence comes from a field experiment where the employment, for the first time, of women in call center jobs (an average of just three new employees per village) increased the care that parents gave to their daughters, as measured by the daughters’ and sons’ body mass index (Jensen, 2012).

But success in using affirmative action to change attitudes towards a disadvantaged group depends on how the dominant group responds. In contrast to the positive effects in India of affirmative action for women in village government, there is no evidence of such effects in the case of affirmative action for the formerly untouchable castes. These reservations did not reduce the practice of untouchability after the period of reservation for the village chief had ended (Girard, 2018). In many districts of India, affirmative action for the low castes induced a backlash. In the public schools, on which low-caste children disproportionately depend, there was increased absenteeism of high-caste teachers. The high caste teachers reduced their teaching effort, and there was greater embezzlement of scholarships intended for low-caste students (Pandey & Pandey, 2022). The process of debiasing perceptions and improving interactions across a social cleavage does not follow automatically from affirmative action. There can be a powerful backlash.