WPS8317 Policy Research Working Paper 8317 The Making of Behavioral Development Economics Allison Demeritt Karla Hoff Development Research Group Macroeconomics and Growth Team January 2018 Policy Research Working Paper 8317 Abstract A core insight from early behavioral economics is that much Many researchers have connected cultural mental models of human judgment and behavior is influenced by “fast to economic development, yet they rarely identify their thinking” that is intuitive, associative, and automatic; very research findings as “behavioral” economics. This research little human thinking resembles the rational thinking that constitutes a second strand of behavioral economics that characterizes homo economicus. What is less well-recog- illuminates the tight interlinkages between preferences, nized is that innate reliance on cognitive shortcuts means culture, and institutions and points to new policy opportu- that cultural mental models—categories, concepts, social nities. It brings the discipline almost full circle back to 18th identities, narratives, and worldviews—profoundly influ- and 19th century perspectives. This essay cautions against ence judgment and behavior. Individuals have a cultural strong reductionism in which sociological influences on “toolkit” or “repertoire” of mental models that they use to decision making are squeezed into a rational actor model. perceive and interpret a situation and construct a response. This paper is a product of the Macroeconomics and Growth Team, Development Research Group. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at khoff@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team The Making of Behavioral Development Economics Allison Demeritt (University of Washington) and Karla Hoff (World Bank)*  Keywords:      behavioral economics, culture, schema, collective decision making,  psychological and social influences on decision making     JEL codes:  O12, Z13   _______  *The authors would like to thank Jean Ensminger, Ravi Kanbur, Barry Schwartz, Joseph E. Stiglitz, and two anonymous referees for valuable comments.  A shorter version of this essay will appear in History of Political Economy 50 (2018) as part of a special issue on The political economy of development economics: A historical perspective, edited by M. Alacevich and M. Boianovsky.   Table of Contents 1.  Behavioral development economics in historical perspective  2.  Connections of behavioral development economics to broader social science  3.  Development economics for the quasi‐rational actor  Example 1.  Simplifying the ballot process to increase the voice of the poor  Example 2.  Harnessing loss aversion to amplify teachers’ incentives  4.  Development economics for the quasi‐rational, enculturated actor  Example 1.  The plough, patriarchal institutions, and female labor force participation  Example 2.  Medieval city‐states and modern civic culture in Italy  Example 3.  Schooling scripts and achievements in learning  Example 4.  Political reservations in India for women and gender attitudes    5.  Moving beyond the reductionist quasi‐rational actor  6.  Conclusion  REFERENCES      2        1. Behavioral development economics in historical perspective    From the earliest work in economics, many writers recognized that people are not always rational and that institutions and cultural contexts shape their character (Smith, 1982 [1759]; Hirschman, 1982). Yet with W.S. Jevons’ introduction of mathematical methods into economics and the creation of neoclassical theory in the 1870s, core research and much of the practice of economics adopted a different theory of the economic actor.1 Neoclassical theory assumes an actor with unbounded rationality and exogenous preferences—he is called the rational actor. Neoclassical theory until the 1970s investigated economic activity independent of institutions (other than property rights) and independent of culture. Few economists believed that the theory accurately described how individuals made decisions. Instead, adopting neoclassical assumptions was widely accepted as the best means of modeling behavior in economic contexts. In many circles, invoking preference changes as an explanation for behavioral changes was condemned as bad science (Stigler and Becker, 1977, p. 89). Development and growth economists could explain a lot with neoclassical theory, and powerfully so. Central neoclassical applications are capital accumulation and markets with information problems: 1) Capital accumulation and learning— In the Solow model of growth, the cause of underdevelopment is a shortage of capital and skilled labor. Empirical work drawing on household-level panel data in poor countries                                                              1  Throughout the 20th century, prominent researchers directed attention to violations of rationality. Keynes discussed “animal spirits” and Simon detailed many aspects of “bounded rationality” (Keynes 1937; Simon 1957). However, this work was often marginalized or ignored because there was little interest in building on a behavioral perspective that lacked a unifying theoretical core.  3      demonstrates the additional problem of imperfect knowledge about technology (Foster and Rosenzweig, 1995). 2) The economics of asymmetric information— Introduced by Akerlof (1970) as “a struggling attempt to give structure to the statement: Business in underdeveloped countries is difficult,” the economics of asymmetric information explains inefficiencies in markets in which potential buyers and sellers have different information about what is being exchanged—that is, essentially all markets. Three extensions of neoclassical theory in the 20th century studied the formation of institutions . Economists considered institutional responses to asymmetric information, including sharecropping (Stiglitz, 1974), trade-credit interlinkages (Aleem, 1993), and group lending (Besley and Coate, 1995). Based on her discovery that common resources, such as fisheries and grazing lands, are often managed successfully by the people who use them, the political scientist Elinor Ostrom (1990) developed theoretical underpinnings for institutional responses to common-pool resources. An adage that shows how her work challenged earlier beliefs about the “tragedy of the commons” is captured by the eponymous law, “A resource arrangement that works in practice can work in theory” (Fennell, 2011): 3) The new institutional economics (NIE)— This work investigates how particular institutions emerge and change. Under rational choice theory, it shows that asymmetric information and costly enforcement explain the emergence of non-market institutions. But NIE also shows that individual rationality will not, in general, bring about an efficient global market equilibrium. NIE focuses on particular institutions. It does not focus on history or power. The influence of the distribution of power on the kinds of institutions that emerge and the growth that results was recognized in the analysis of the “quasi-natural experiment” of European colonialism. Sokoloff and Engerman (2000) and Acemoglu, Johnson, and Robinson (2002) showed that differences across the colonies in initial inequality of wealth, human capital, and 4      political influence could explain the differences across the former colonies in economic performance in the 20th century (a survey is Hoff, 2003): the initially richest areas—rich because of plantations and mines—were characterized by the highest levels of inequality and became the poorest in modern times. There was a “reversal of fortune”: 4) The colonial origins of underdevelopment— Across the regions once colonized by European powers, the vast differences in the 18th and 19th centuries in the fraction of the population with access to the vote, schooling, financial markets, and property rights help explain the vast differences in economic performance today. This work focuses on the impact of history on macro-performance. Research in the colonial origins of underdevelopment draws on a large sample of economies. But many questions about development cannot be studied across large samples. To make it possible in those cases to build cumulative knowledge, economic historians introduced an approach called “analytic narratives.” A theoretical model is built based on a specific historical problem. For example, Greif (1994) explains why the city-state of Genoa in the early medieval period established a rule of law to solve the problem of overseas agents behaving opportunistically during long-distance trade, whereas the Maghribis, who traded all over the Muslim Mediterranean world, did not. Although based on specific cases, models in analytic narratives generate testable hypotheses applicable to many cases (Bates et al., 1998). 5) Analytic narratives— Economic historians enlarged neoclassical theory by deductive rational choice modeling of large-scale historical phenomena. History influences the salience of the possible equilibrium outcomes and, therefore, which equilibrium is selected.      Most work in NIE and analytic narratives retains rational choice theory while bringing social factors back into economic analysis. As the precision of empirical work in economics grew, it became increasingly clear that there remained much of importance that rational choice 5      theory could not explain. There were systematic violations of rational decision-making with large consequences for welfare, e.g., underuse of fertilizer and under-responsiveness to free vaccination programs for children (Duflo, Kremer and Robinson, 2008, 2011; Banerjee et al., 2010). Esther Duflo put to herself and others the question: when naturally occurring data does not provide the evidence needed to answer a development question, why not implement an experiment (Parker, 2010)? To pursue their vision and advise governments on policy design in all its detail, Duflo and her colleagues founded a lab at MIT. They viewed randomized controlled trials (RCTs) in field experiments as the best way to learn how to promote economic development and reduce poverty. The Jameel Poverty Action Lab (JPAL), and many other organizations, as well, have undertaken RCTs in countries all over the world: 6) RCT-based project evaluations — This approach – in which individuals are randomly allocated to treatment and control groups – has been described as an effort to “transform…development economics, one experiment at a time… Because of the randomness, [treatment and control] groups, if large enough, will have the same complexion: the same mixture of old and young, happy and sad, and every other possible source of experimental confusion. If, at the end of the study, one group turns out to have changed—become wealthier, say—then you can be certain that the change is a result of the treatment...” (Parker, 2010) Deaton (2010) and other critics contend that RCT-based project evaluation will not lead to a better understanding of the development process. The success of a treatment may depend on many mechanisms. Which mechanisms are in place in which contexts is not learned from RCT- based project evaluation. The critics argue that it is necessary to go back to theory to build a cumulative research program in economic development. JPAL has indeed increased its use of “mechanism experiments,” which are RCTs to identify the behavioral mechanisms central to well-specified policy questions (Deaton, 2010; Jensen, Kling, and Mullainathan, 2011). 6      In trying to explain behavior, economists had in many cases reached the limit of what they could do with structural models in rational choice theory. The next approach, which is the focus of this paper, is based on a psychologically and sociologically more realistic model of decision making. It began with the work of two psychologists—Daniel Kahneman and Amos Tversky. In the 1970s and 1980s, they documented predictable violations of the rationality assumption. They showed that changes in the presentation of choices or in contextual factors that would be irrelevant to a rational actor could reverse the preferences of experts and non-experts alike (Tversky and Kahneman, 1981). For example, in two treatments of lung cancer—radiation treatment and surgery—the same statistics were presented to some respondents in terms of mortality rates and to others in terms of survival rates. The respondents then indicated their preferred treatment. The information about the two options was the same, and so neoclassical economics would predict that the preferences would be the same across formulations. But the different framing produced a marked effect by changing what automatically came to mind (“fast” thinking). “The advantage of radiation therapy over surgery evidently looms larger when stated as a reduction of the risk of immediate death from 10% to 0% rather than as an increase from 90% to 100% in the rate of survival” (p. 596). Kahneman and Tversky’s research led to new bodies of work, including ‘behavioral finance,’ ‘behavioral macroeconomics,’ and ‘behavioral game theory.’ Most recently, it has given rise to ‘behavioral development economics’: 7) Behavioral development economics— Behavioral development economics studies problems of economic development using psychologically realistic models of both decision making and preference formation, rather than rational choice theory. Some of the work is informed by the sociological 7      perspective that “take[s] it as obvious that individuals’ preferences are formed by society and that society, so to speak, exists within persons.”2 For the first time, research across disciplines is producing rigorous evidence that context, history, and culture affect decision-making. New techniques have expanded the ability to infer causality from non-experimental data. Research findings are that preferences are influenced by social networks and experience and exposure through which individuals learn mental models (such as categories, identities, and narratives) that they use to process information. Myopic preferences can explain why the wrong types of leaders emerge and well-intentioned reformers fail to obtain legitimacy (Khemani, 2017). Table 1: Frameworks to explain economic development outcomes                                                 Focus on:  Rational choice  Institutions  Culture  Empirical validation    with fixed  of individuals’  Approach  preferences  intentions    - Early economics No  Yes  Yes  No  - Neoclassical growth theory         Yes  No  No  No  - Neoclassical price theory, including the economics of asymmetric information - NIE* and the colonial origins of         underdevelopment Yes  Yes  Some  No  authors, not  - Analytic narratives others  - RCT-based evaluation of projects No  No  No  Yes    - Behavioral development economics No  Yes  Yes  Yes  * Douglass North’s work beginning in the 1990s recognizes the influence of bounded rationality, ideo- logy, and cultural mental models to process information, which makes it an outlier in the category of NIE.                                                              2  Personal communication from the sociologist, Peter Hedstrom, cited in Fehr and Hoff (2011). An early survey of work on endogenous preferences is Bowles (1998). A history of behavioral economics is Heukelom (2014). 8      Table 1 lists the approaches to development economics that we have discussed. The table makes it easy to see one of the themes of this essay: modern development economics has circled back to insights made in the 18th and 19th centuries about the limits on rationality and the impact of institutions and culture on preferences. With the obvious exception of ‘early economics,’ the approaches listed in Table 1 are alive and well today. Thus, development economics is fragmented, as it has always been. Even the individual approaches listed in Table 1 are fragmented. There is no agreement on what constitutes an institution nor, as we discuss later, on the degree to which culture (under various definitions) is a part of behavioral economics. Today, the central focus of research and teaching in development economics is on micro- economics—market failures and RCT-based mechanism evaluations. Graduate students may read everything from the foundational works in neoclassical economics to research on the psychology of poverty. Many are steeped in instruction on RCTs. Within development economics, behavioral development economics is gaining momentum. It is producing a coherent set of principles to improve the design of development projects and policies. The first journal symposium on behavioral development economics was published in 2014 (in the Review of Income and Wealth). In 2015, Harvard University introduced a PhD-level course in behavioral development economics, and the World Bank presented the first book-length synthesis of work in this field (World Development Report 2015: Mind, Society, and Behavior). Kremer and Rao (in progress) are writing the first Handbook survey, entitled “Behavioral Economics and Development.” Development practitioners have applied behavioral insights in many low- and middle-income countries and in poor communities of rich countries. Extended discussions of the successes of these interventions are Datta and Mullainathan (2012), Demeritt and Hoff (2015), 9      and the World Development Report 2015. The successes launched behavioral development economics. 2. Connections of behavioral development economics to broader  social science   In terms of methodology, behavioral development economics mirrors changes in the economics discipline as a whole. Hamermesh (2013) studied historical patterns in the methodologies of papers published in the top three general economic journals (Figure 1). The share of theory papers declined from 62% in 1983 to 28% in 2011. In the same period, the share of empirical papers rose from 38% to 64%. Experiments became an important part of the field for the first time in the mid-1980s and comprised 8% of the papers in 2011. Figure 1: Methodologies of papers in the top three economics journals, 1983-2011 Share of Paper by Methodology 70 63.9 61.6 60 Percentage of Papers 50 37.6 40 27.9 30 20 8.2 10 0.8 0 1983 2011 Type of Paper Theory Empirical Experiment Source: Data compiled from Hamermesh (2013) 10      Within the field of development economics, the current generation of researchers has focused on empirics (Ray, 2008). At the Richard T. Ely lecture at the 2017 American Economic Association Annual Meeting, Duflo (2017) urged economists who are helping governments design new policies to act not only like scientists and engineers, but also like plumbers. Plumbers lay pipes and fix leaks and get things to work properly on the ground by adopting an experimental mindset. Economists who design policy need to recognize that humans are boundedly rational and are embedded in cultural contexts. Therefore, what worked “there” may not work “here.” We next turn to the relationship between behavioral development economics and other social sciences with respect to “grand theory.” Behavioral development economics does not offer a new grand theory as an alternative to neoclassical theory. What it offers instead are realism-improving middle-range theories. Rabin (2013) calls them “PEEMS”—portable extensions of existing models. They incorporate features of psychology— e.g., limited attention, limited will power, framing effects, salience (the details that leap out at you), loss aversion, social identity, and concern with fairness and morality— into an otherwise standard economic model that translates them into testable predictions. PEEMs address an inconsistency that Kenneth Arrow noted over 30 years ago: [A]n economic theorist …toils for months to drive the optimal solution to some complex economic problem, and then blithely assumes that the agents in his model behave as if they are capable of solving the same problem. (Thaler, 2016, p. 162) A few subfields of behavioral economics, with an example of a PEEM or middle-range theory in that field in parenthesis, are behavioral finance (prospect theory), behavioral macroeconomics 11      (present bias), behavioral game theory (limited strategic thinking), and behavioral law and economics (the availability heuristic). While an expressed need for mid-level generalizations is relatively new in economics, it has long propelled work in other social sciences. Robert Merton, one of the ‘founding fathers’ of modern sociology, penned an essay in 1949 titled “On Sociological Theories of the Middle- Range.” The essay was a reaction to scholars’ desire for a “total system of sociological theory,” which Merton argued offered “the same exhilarating challenge and the same small promise as those many philosophical systems which have fallen into deserved disuse” (p.453). He advocated “middle-range theory” firmly tethered to empirical data and hypothesis testing. Modern-day social theorists build on this idea (e.g., Hedstrom and Swedberg, 1998; Elster, 2007; Hedstrom and Ylikoski, 2010; an argument particular to development is Deaton, 2009). The behavioral sciences are converging on a shared view of the value of middle-range theory. Such theories do not match the scope and power of rational choice theory but are necessary for understanding many behaviors. They provide a framework within which policy tools are designed, as discussed later. Many economists have been wary about treading into this territory. The concept of choice is very ingrained in economists’ thinking. One way of thinking about behavioral economics is that it tries to clarify the domain of choice. Choice is influenced by the context of decision making, over which the decision-maker has limited control, and by the categories, concepts, and narratives through which an individual processes information, over which he has limited awareness. We will later present four examples (on female labor force participation, civic culture, learning in education systems,, and gender attitudes) to illustrate how social constructs emerge and how they constrain or enable choice. 12      This essay distinguishes two strands of behavioral development economics, following Camerer (2005) and Hoff and Stiglitz (2016). Strand 1 investigates the ‘quasi-rational’ aspects of the decision maker in which he ‘thinks fast’ and uses rules of thumb as cognitive shortcuts that cause him to make errors (Section 3). This strand links economics primarily with work in psychology and ignores the social and cultural factors that shape important aspects of psychology. Strand 2 investigates the ‘enculturated’ aspects of the decision maker, examining how individuals are jointly constituted with their social environments, with consequences for both individual decision making and group outcomes (Section 4): “modes of interpreting situations…become characteristic of a given culture…[A]t a given moment in history it is a group’s beliefs and ideology, rather than immediate features of its objective circumstances, that may hold the key to its subsequent development” (Ross and Nisbett, 1991, 176).3 Strand 2 links economics with key ideas in sociology and anthropology. Although many people associate Strand 1 with the term ‘behavioral economics,’ this line of work is fundamentally linked with Strand 2. The “fast thinking” of Strand 1 requires for many of its operations the cultural material of Strand 2. For example, experimenters can induce dishonesty among bankers by reminding them of their banking identity before they play a game (Cohn et al. 2014). The identity prime shifts bankers’ minds from one socio-cultural frame (in which their banking identity is not salient) into another (in which it is) and this shift recruits new associations, meanings, behavior patterns, preferences, and norms that govern behavior. The reason the shift happens seamlessly and without individuals’ conscious awareness is their long history of experience operating in the cultural milieu of banking. Absent that experience, the                                                              3  Guiso, Sapienza, and Gonzales (2016), discussed below, and Mokyr (2017) provide examples.  13      authors show that the prime has no effect. Many Strand 1 effects rely on culturally-specific mental architecture that is now being investigated as part of Strand 2 behavioral economics. Whether fast thinking leads you automatically to, for example, overlook women for leadership roles, or to respond aggressively to an assertion of authority, depends on the experience and exposure you have had (e.g., Alesina et al., 2013; Heller et al., 2017). “People think and feel and act in…ways that are shaped by particular patterns of historically derived meanings, practices, products and institutions” (DiMaggio and Markus, 2010, p. 348). Only some features of Strand 1 of behavioral economics, such as procrastination, are common to all known cultures. People universally do not meet the assumptions of homo economicus that preferences are consistent and fixed, but the mechanisms and manifestations of the departures are, in part, culturally specific. Perception entails construction. The anthropologist Mary Douglas (1966, 45-46) writes that Whatever we perceive is organised into patterns for which we, the perceivers, are largely responsible. Perceiving is not a matter of passively allowing an organ—say of sight or hearing—to receive a ready-made impression from without, like a palette receiving a spot of paint….As perceivers we select from all the stimuli falling on our senses only those which interest us, and our interests are governed by a pattern-making tendency, sometimes called schema [or equivalently, mental model]…In perceiving we are building, taking some cues and rejecting others. The most acceptable cues are those which fit most easily into the pattern that is being built up…Ambiguous ones tend to be treated as if they harmonised with the rest of the pattern. Discordant ones tend to be rejected... A cultural toolkit of concepts, categories, identities, narratives, and worldviews, allows people to “locate, perceive, identify, and label” events in the world around them (Goffman, 1974). A greater appreciation of the effects of culture on cognition has led scholars in sociology, 14      psychology, and anthropology to abandon the “entity” view of culture, in which culture determines a stable and consistent set of preferences, in favor of a dynamic view of culture as something that happens “in action” as individuals interpret a situation using a cultural “toolkit” or “repertoire” of concepts, categories, meanings, identities, and narratives and construct a response (Swidler 1986; DiMaggio 1997). In contrast, most modern economic work that takes culture into account adopts the entity view of culture. Behavioral economics is beginning to embrace the toolkit approach, as we discuss below. The cognitive toolkits people use to assess situations and solve problems can affect institutional persistence and change. When members of a group share monolithic cultural experiences, the beliefs and attitudes associated with those experiences can become entrenched even as available information and technologies change; since attention and perception are shaped by an individuals’ limited experiences, and since beliefs themselves affect self-confidence and performance, “fictions” can be sustained in equilibrium, hampering societal development (Hoff and Stiglitz 2010). Individuals and culture are mutually constituted (Markus and Kitayama 2010), and culture and institutions interact and evolve in complementary ways (Alesina and Guiliano 2015). Thus culture, preferences, and institutions are all closely related (overviews from the perspective of behavioral economics are North, 1994; Fehr and Hoff, 2011; and Hoff and Stiglitz, 2016). North (1990, p. 8) discusses how culture’s effect on individual judgment can block development: Incremental change comes from the perceptions of the entrepreneurs in political and economic organizations that they could do better by altering the existing institutional frameworks at some margin. But the perceptions crucially depend on both the information that the entrepreneurs receive and the way they perceive it. If political and economic markets were efficient (i.e., there were zero transaction costs), then the choices made would always be efficient. That is, the actors would always possess true models or if they initially possessed incorrect models, the information feedback would 15      correct them. But that version of the rational actor model has simply led us astray. The actors frequently must act on incomplete information and process the information that they do receive through mental constructs that can result in persistently inefficient paths. Transaction costs in political and economic markets make for inefficient property rights, but the imperfect subjective models of the players as they attempt to understand the complexities of the problems they confront can lead to the persistence of such property rights.” [emphasis added] Section 5 of this essay critiques the reductionist tendency in behavioral development economics, which reduces culture to shared values and rules of thumb. This narrow definition leaves out of account that culture also provides the cognitive toolkit for perceiving and interpreting a situation—including categories, concepts, identities, causal narratives, and worldviews. The cognitive toolkit influences decision making at three levels: perception, construal, and preferences. 3.  Development economics for the quasi‐rational actor   Richard Thaler (2000, 136), a founder of Stramd 1 of behavioral economics, views the individual as “quasi-rational, meaning trying hard but subject to systematic error”: he has fixed preferences but does not always take actions to achieve his objectives because he is “thinking fast.” What comes automatically to mind (including associations and narratives that make some things salient or meaningful) drives his behavior. In general, the pressures of competitive markets do not correct his mistakes and can make them worse, as when a ‘phisher’ manipulates mental frames — “the stories people are telling themselves”—to promote sales (Akerlof and Shiller, 2015, p. 172). A hard-to-forget example coming from Akerlof and Shiller is the Cinnabon®. This roll is cooked in the mall or airport shop that sells it so that its aroma penetrates the surroundings, attracting potential buyers. People may find it hard to resist the sweet-smelling, 880 calorie bun, which would undermine any plan to eat healthfully. “Much effort and expertise went into understanding our weak moments and developing a strategy to take advantage of them” (p. 2). 16      In middle-income countries, there has been a rapid expansion of door-to-door sales of high- sweetened, low nutritional foods to poor people, expanding rates of obesity, diabetes, and undernourishment (New York Times, Sept. 19, 2017, p. A1). A book titled Nudge (Thaler and Sunstein, 2008) became “the spearhead of behavioral economics” (The Economist, March 24, 2012). The authors define a nudge as a policy that changes behavior without substantially changing incentives or information. One way a nudge works is by changing how easy it is to choose a particular option and by giving immediate feedback that permits individuals to correct their mistakes. We use two examples to illustrate how a nudge can promote economic development. Example 1.  Simplifying the ballot process to increase the voice of the poor      There is often a high return to public investments in the poor, but relatively little is known about how to induce governments to increase such investments. The introduction of electronic voting technology in Brazil in 1998 showed that making it easier for voters to communicate their preferred choices greatly increased the voice of the poor by reducing the number of error-ridden votes, particularly by the less educated. The limited number of the devices available in 1998 meant that the electronic voting technology was introduced only in municipalities with more than a threshold number of voters, while the rest used paper ballots. Brazil requires all adult citizens to vote. Fujiwara (2015) identifies the impact of the machines by comparing outcomes for municipalities just above and just below the threshold. By reducing the number of error-ridden ballots that had to be thrown away, electronic balloting effectively enfranchised 11% of the voters. Political power in municipalities shifted to the Left. One of the things that legislators 17      could quickly affect was funding for health care, which is free in Brazil. Funding for public health care increased by 34% over eight years. The number of pregnant women receiving regular pre-natal care increased by 20 percent and the number of low-weight births decreased by 6%. As the World Bank Group (2015, 37) notes, “[t]his is a major development success, since newborn health, controlling for other factors, predicts lifetime health, education, and income.” Example 2.  Harnessing loss aversion to amplify teachers’ incentives   Poor education of disadvantaged students contributes to the intergenerational transmission of poverty. Better performance by their teachers can help children escape poverty as adults. The rational actor responds to incentive pay, but there is scant evidence that merit pay for teachers is effective. In a disadvantaged community near Chicago, Fryer et al. (2012) investigated how to amplify the effect of incentives. They ran a field experiment that involved 150 teachers of kindergarten through eighth grade in disadvantaged communities. The teachers were randomly assigned to a control group or one of the two treatment groups, which we will call ‘winners’ and ‘losers.’ The ‘winners’ worked under a traditional year-end bonus scheme, under which they could make up to $8,000 extra at year-end for strong performance of the students on year-end standardized tests. The ‘losers’ were given $4,000 at the beginning of the academic year and were told that at year-end, poor performers would have to return a portion and that the strongest performers would receive an additional bonus of up to $4,000. The ‘winners’ and ‘losers’ faced identical financial incentives. Only the framing of the incentive payments differed between the two groups. 18      Who performed better? The ‘losers’ did. The improvement in their students’ achievement was the equivalent of an increase in teachers’ skills by one standard deviation. In contrast, students with teachers in the ‘winners’ treatment made no achievement gains. 4. Development economics for the quasi‐rational, enculturated actor   The second strand of behavioral economics focuses on how individuals are enculturated (in addition to being quasi-rational). Extended exposure to social patterns of, e.g., patriarchy, racism, or corruption, shapes the categories, identities, narratives, and worldviews that mediate an individual’s perception of the world. Following North (1994, 2005) and Denzau and North (1994), we call these cognitive tools mental models. They are “the intuitive set of principles or ideas of how things work, which govern people’s predictions about the effects of change” (Camerer, 2005, 72). They matter for decision-making because problems must be represented in the mind before an individual can solve them. There is always a gap between objective reality and subjective perception since the experience of reality is mediated by mental models. The behavioral economist Michael Bacharach (2003; 63, 71) describes the consequence for economic theory aptly: “one does not just see, one sees as…. [A]n empirically adequate theory of economic decision-making must model the decision-maker’s problem as she herself sees it….” Mental models often enable people to act quickly and effectively (Goldstein and Gigerenzer, 2002). But sometimes they impede the ability to seize opportunities. A striking example are seaweed farmers in Indonesia (Hanna, Mullainathan, and Schwartzstein, 2014). In a random sample of nearly 500 seaweed farmers, nearly all thought that the size of the pod that they planted did not affect productivity. Depending on how pod size was measured, farmers’ average estimated loss of income from using an incorrect pod size on their fields was 19      7% or 38%. Researchers invited farmers to participate in an experiment held on the farmers’ own plots. Farmers were present for the planting and helped with weighing the seaweed and recording the results. Yet simply having access to data on the relationship between pod size and yield did not change farmers’ behavior. Only after the experimenters presented each farmer with a simple summary table of the relationship between pod size and income on his own fields did a significant number of farmers correct their mistake. This experiment shows that the farmers’ way of conceptualizing seaweed farming blocked learning. Their mistake had persisted over many years: the farmers had on average 18 years’ experience and were descendants of seaweed farmers. By expanding individuals’ repertoire of mental models—an outcome that is distinct from simply exposing individuals to new information, as the seaweed example demonstrates— an intervention can make people much better off. North (1994) lamented economists’ failure to study the mental models that hinder economic development. He argued that by distorting perception, mental models may foster the underdevelopment of a subculture or entire society. Consider as an example the problem learning to cooperate. All experiments with U.S. subjects have shown that fixed pairs of players in a repeated assurance game called the Stag Hunt will remarkably quickly form the Pareto efficient convention. It was believed that this pattern would apply universally. In order to test that, a repeated Stag Hunt game was played in India. In the experiment in 10 villages in North India, low-caste men generally formed the efficient convention, as the findings in prior experiments with U.S. subjects would predict, but high-caste men generally did not. High- caste men in North India have inherited from earlier socioeconomic regimes a high concern with status and honor and “acting tough” to deter predators. In the experiment, they seemed to 20      interpret the loss they bore from miscoordination as an insult, rather than as the innocent outcome of not-yet-convergent expectations (Brooks, Hoff, and Pandey, 2017). Retaliation to the perceived insult led to the unraveling of the expectations needed to sustain the efficient convention. Institutions possess a “schematizing power” (Bruner, 1990, p. 58): they are a major source of mental models, the templates needed to interpret situations and construct responses. Although many institutions are born out of rational needs to reduce uncertainty and improve coordination, over time they become part of the fabric of society (naturalized, typified) and provide a ‘taken-for-granted’ reality that frames decision contexts; subjective meanings become objective facticity (Berger and Luckmann, 1966). Thus, institutions not only reflect cognition but also shape it by informing “common sense,” which at times is far from common or sensible. Mary Douglas (1986, 10) describes how institutions create epistemological resources:  Classifications, logical operations, and guiding metaphors are given to individuals by society. Above all, the sense of a priori rightness of some ideas and the nonsensicality of others are handed out as part of the social environment… Epistemological resources may be able to explain what cannot be explained by the theory of rational behavior. We next present four examples in which institutions and scripts have influenced perceptions, with large impacts on economic development. Example 1.  The plough, patriarchal institutions, and female labor force participation     Pre-industrial 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 strength either to pull it or control the animal that pulls it. In areas that were topographically well-suited to use of the plough (that is, well-suited to grow wheat, barley, and rye as opposed to maize, sorghum, and millet), men had an advantage in farming relative to 21      women, and adoption of the plough created gendered occupations: male specialization in agriculture and female specialization in domestic activities. Ethnic groups that historically relied on plough-based agriculture continue in modern times to have more unequal gender norms towards employment outside the home. In surveys, individuals from such groups indicate stronger beliefs in gender inequality. In areas well-suited to use of the plough, compared to areas that are not, female labor force participation in 2000 was more than 20 percentage points lower; there was also lower participation of women in firm ownership and politics (Alesina, Guiliano, and Nunn, 2013). The basic trends hold even among children of immigrants to the U.S. or Europe who are from historically plough-using ethnic groups. This indicates that cultural beliefs and values are at least partly responsible for the effects. Example 2.  Medieval city‐states and modern civic culture in Italy     In The Moral Basis of a Backward Society, Edward Banfield (1958) argued that a culture of nepotism and the absence of public-spirited norms kept southern Italy undeveloped compared to northern Italy. Putnam (1993) took advantage of a natural experiment to show that although the formal institutions of 15 new Italian regional governments created in the 1960s and 1970s were identical under national law, their performance was not. Instead, their performance was correlated with regional differences in civic community which, Putnam argues, date back to the late Middle Ages. Around the turn of the last millennium, southern regions of Italy were brought under Norman rule, with political power based on hierarchical religious authority or divine right. In contrast, northern regions experienced the collapse of the Holy Roman Empire and the rise of 22      free city-states, which encouraged the growth of horizontal networks and early forms of participatory democracy. Guiso, Sapienza, and Gonzales (2016) formally tested Putnam’s hypothesis, measuring civic culture in modern Italian towns by the number of non-profit organizations, the existence of an organ donation organization, and the frequency of cheating on a national exam by fifth-grade students. Looking across 400 Italian cities, they find that civic culture is increasing in the occurrence, duration, and intensity of an area’s medieval experience as a free city-state. Further, children in towns with a free city-state legacy have higher self-efficacy than their peers in towns without such a legacy. The authors suggest that participating in the public life of a free city-state may have bolstered beliefs about self-efficacy, which were passed down across generations. Historical events and institutional shocks influence a “nation’s psyche” (p. 1434). Example 3.  Schooling scripts and achievements in learning     Institutions provide scripts of appropriate structure and action. The scripts carry normative obligations that can spur behavior more consistent with a ‘logic of legitimacy’ than a ‘logic of efficiency.’ In some countries, school systems are driven more by the need to ‘look right’ (i.e., to adopt the legitimated formal structures and scripts of schooling, such as a national ministries and commissions, standardized curricula, enrollment indicators, and record-keeping and inspection systems) than with producing learning (Meyer and Rowan 1977; an application to Botswana is Meyer et al., 1993). The result can be an extensive set of structures and practices associated with education delivery that are nevertheless almost completely decoupled from real learning. 23      In India, the information management and state report card system, which was designed to provide a ‘comprehensive’ and ‘unified’ set of statistics for public schools, is an example: In the 2011/12 report card for Tamil Nadu, I counted 817 distinct pieces of information reported. But of the 817 pieces of information not a single one could be construed as a direct measure of learning of any kind…In the section called ‘Performance Indicators’ there are 24 distinct measures including the ‘per cent of schools approachable by all- weather road,’ ‘per cent with boundary wall,’ ‘per cent with ramp,’ ‘Pupil teacher ratio.’ There was data on teachers…by gender, by caste, by age, by ‘per cent trained’, by formal qualification. But [a measure of] learning—of any subject, at any age, measured in any way? Not a single number. (Pritchett, 2014) India’s school system is far from alone in having succumbed to institutionalized scripts at the expense of productive activity. The primary message of the World Bank’s World Development Report 2018 is that “schooling is not the same as learning.” Its central policy recommendation is the astoundingly simple yet deeply important recommendation that developing countries must begin to act in ways that show that “learning really matters to them” (World Bank, 2017). Political and economic forces can create incentives to maintain inefficient institutions and not take actions to produce learning. But there are also cognitive factors at play: institutional scripts shape judgments of appropriate behaviors and goals, individuals orient their actions towards the scripted goals, and this reinforces the strength and legitimacy of the institution. Societies may get stuck with institutions and systems that few people recognize are broken. Example 4.  Political reservations in India for women and gender attitudes    A 1993 amendment to the Constitution of India reserved for a female, in a randomly selected one-third of village governments, the position of village council head (pradhan). Seven years of experience of living under a woman pradhan erased male villagers’ prejudice against women 24      leaders by many measures—an Implicit Association Test, the evaluation of political speeches, and the assessment of the quality of actual village female pradhans. In villages that had been exposed to women pradhans, parents’ aspirations for their teenage daughters, and teenage girls’ aspirations for themselves, were higher; and girls went to school somewhat longer and did somewhat fewer hours of housework. Political reservations for women as pradhan also made female victims of crime much more willing to report the crimes to the police, and made the police much more willing to record the reports (Iyer et al., 2012). 5. Moving beyond the reductionist quasi‐rational actor    One of the strengths of neoclassical theory is its extreme reductionism: with a few principles, it explains a massive amount of behavior. Since this feature catapulted economics to the top of the social science status hierarchy, it is not surprising that reductionism also influences the newest field in economics. Despite growing recognition that social and cultural factors profoundly influence decision making, economists are generally inclined to squeeze the effects into the conceptual and linguistic parameters of rational choice theory. However, doing so may limit the potential of this new field. We will consider a few examples of reductionist tendencies. Nunn (2012, p. S108) defines culture as “heuristics or rules of thumb that aid in decision making.” This definition seems too narrow. Heuristics and rules of thumb can certainly improve decision making. For instance, an experiment in the Dominican Republic showed that lower-skilled entrepreneurs benefit more from a training program based on rules of thumb such as ‘keep business and household money in separate purses’ than training in a standard accounting program (Drexler et al., 2014). But people cannot be given a ‘heuristic’ or ‘rule of thumb’ to increase community trust, promote women’s agency, reduce corruption, or end discrimination. The mental models that sustain these 25      behaviors are too entrenched – they are socially reinforced and psychologically taken-for- granted—and so they often resist simple attempts at change. Datta and Mullainathan (2014, 15) recommend condensing principles in behavioral economics into a framework of “constraints”: To help navigate the large set of findings, we condense the behavioral literature using one simple perspective about the constraints under which people make decisions. Economists and policymakers – indeed all of us – understand constraints all too well. Resources are limited: there is only so much money, time, staff, or even enthusiasm to go around. Yet we often do not realize that mental resources are also limited…. we often design programs assuming that people have unbounded cognitive capacity.… Behavioral economics can be interpreted as identifying a few more limited resources. (italics in original) We will argue that it is too narrow to classify the cognitive effects of psychological, sociological, and anthropological factors only as resources that enable or limit decision making, since cognitive resources cannot be handed to an individual. They usually must be acquired through experience (sometimes repeated experience) to catch hold in the mind. Consider the authors’ example of diarrhea, which kills over half a million infants in developing countries each year, with most of the deaths easily preventable through the use of a sugar and salt water solution (Oral Rehydration Solution (ORS)). The authors ask, why isn’t ORS used enough? Since parents see a child with diarrhea as leaking fluids, 35-50% of them say that the best treatment is to keep the child ‘dry’ by reducing fluid intake. Datta and Mullainathan describe the failure to use ORS as a “scarcity of understanding” (arising from a false mental model). This statement is true. But the terminology underplays the cognitive shifts that need to happen for behavior to change, and the interventions required to change a mental model. 26      Gawande (2013) describes how Bangladesh attacked the problem. A pilot project team set out to reach 60,000 women in 600 villages: They travelled on foot, pitched camp near each village, fanned out door to door, and stayed until they had talked to women in every hut…Initially, the workers taught up to 20 mothers per day. But monitors visiting the villages a few weeks later found that the quality of teaching suffered on this larger scale, so the workers were restricted to 10 households a day. Then a new salary system was devised to pay each worker according to how many of the messages the mothers retained when the monitor followed up. The quality of teaching improved substantially. The field workers soon realized that having the mothers make the solution themselves was more effective than just showing them. The workers began looking for diarrhea cases when they arrived in a village, and treating them to show how effective and safe the remedy was. When the Government of Bangladesh took the program nationwide, it hired, trained, and deployed thousands of workers, region by region, who went door-to-door to show 12 million families how to save their children. As Gawande describes it, “The program was stunningly successful…the knowledge became self-propagating. The program had changed the norms.” Surveys showed that nearly 80 percent of recent diarrhea episodes were treated with the solution. Most women in Bangladesh now teach their children about oral rehydration, eliminating the need for repeated campaigns. But in countries that adopted an arm’s-length approach, the information campaign failed almost entirely. Policy makers who are asked to remedy a “scarcity of understanding” about diarrhea treatment might reasonably assume that they need simply to get the right information to the right people; it is not clear that the policy makers would come up with anything close to what happened in Bangladesh. Our concerns could be dismissed as a quibble over language, but a theme of behavioral economics is that language and labels make a difference for understanding. This is especially true in fields that are closely tied to policy. In an academic setting, common training and frequent interaction among individuals means that a nuanced understanding of ideas can more 27      easily be assumed, and reductionism simplifies communication in useful ways. But in a field that depends on collaboration among people with different backgrounds and priorities, common understanding should not be assumed, and ‘short-hand’ communication can impede comprehension and backfire. One need not look far to find well-designed programs that failed during implementation because too little attention was paid to the socio-psychological mechanisms behind behavior change. A rule of thumb is a simple tool that is normally easy to teach; and sometimes understanding is impeded by a simple knowledge ‘constraint’ (e.g., the fact that water can be contaminated after it is chlorinated). In contrast, mental models are often deeply intertwined with cultural values and expectations regarding the actions and judgments of others (e.g., the social roles of men and women). Reducing the latter (cultural values and expectations) to the former (a simple knowledge constraint) belies what it takes to change mental models. Akerlof has noted that ideas that fall outside the typical province of economics are generally regarded as “taboo” within the discipline. He argued that although rational choice theory is probably adequate for analyzing most types of economic behavior, a new field of “psycho-socio-anthropo-economics” is “probably applicable to the big problems, to the really big things that we don’t yet understand in economics…it is probably applicable to the question of why underdeveloped countries are so poor” (interview in Swedberg, 1990, p. 68). Studies associated with 21st century behavioral economics borrow ideas from these disciplines, but sometimes screen out useful interpretations and causal connections (Davis, 2013). Successfully addressing dysfunctional organizations and institutions around the world requires recognizing a quasi-rational, enculturated actor. In many ways, the interventions at the frontier of behavioral 28      development economics are as much ‘development sociology’ and ‘development psychology’ as they are development economics. 6. Conclusion  Behavioral development economics is an emerging subfield that is generating increasing interest among researchers and policy makers. This essay covers its brief history, mostly in the 21st century. Many of its insights, by incorporating multiple dimensions of human behavior, can be traced back to the earliest work in economics—by Adam Smith, Vilfredo Pareto, J.S. Mill., Max Weber, and others. What is new in the 21st century is the strength of the evidence behind these insights, produced by recent work using RCT and econometric analysis of naturally occurring data. In recent years, research that sheds light on the mechanisms behind human behavior has increased. From that perspective, development economics has gone micro, instead of the prevailing macro stance of the 1950s and 1960s. The goal of behavioral economists is not to replace the powerful theory of rational choice, but to supplement it with mid-level theories that improve explanation and policy design. The approach is not yet a cohesive field. Some people associate it only with the work of Kahneman and Tversky and the ‘nudge’ paradigm, whereas this essay points to a rapidly growing body of work demonstrating that (1) experience and exposure shape preferences and bias perception; and (2) interventions that harness this power can improve well-being. Some economists would like to see the framework of behavioral economics expanded (for example, by incorporating emotions more fully). But many also feel compelled to squeeze behavioral findings into a reductionist framework in which neoclassical economics is augmented only with 29      shared preferences and rules of thumb. Like development economics, behavioral development economics itself encompasses a diverse array of approaches that remain fragmented. Behavioral development economics is trying to answer many of the same questions investigated in two earlier bodies of work –NIE and research on the colonial origins of underdevelopment. NIE explained, for example, the existence of sharecropping contracts but could not explain why the terms of the contracts within villages or regional “patches” were nearly uniform (despite variable soil quality), separated by boundaries where contractual terms were variable or jumped substantially from one set of terms to another. Young and Burke (2001) explained the puzzle by bringing in the psychology of bargaining: the focal power of particular shares, such as one-half or one-third, and concerns to conform to choices that other contracting parties had made. The work on colonial origins of underdevelopment explained the 19th century differences in growth rates across former European colonies, but could not easily explain the durability of the effect of colonial origins. Behavioral development economics points to the importance of increasing our understanding of the historically derived social machinery that affords certain psychological tendencies that from the perspective of the global North seem irrational, lawless, deficient, and immoral (e.g., pervasive mistrust, protection of family honor, nepotism, cronyism, and corruption). Investigating this social machinery can illuminate the cultural logic of these tendencies and how and why they persist; it can also suggest where and how it might be possible to intervene (personal communication from Hazel Rose Markus). There used to be a clear boundary in economics between what was changeable and what was not—the hardware of the brain versus its software, genetics versus culture. But the boundary is no longer clear. Epigenetics shows that the environment can determine which genes 30      are expressed and which ones are not. Culture determines how we process information—what is absorbed and what is misinterpreted or blocked. The fact that beliefs affect what individuals pay attention to and how they process information can explain why some mistakes are not corrected in a generation (Hanna et al., 2017) or even over generations (Hoff and Stiglitz, 2010 and 2016). Economics is about explaining and predicting change and designing policies to foster welfare-increasing change. The old view of development economics, with strict boundaries separating it from other social sciences, explained development and growth in terms of changes in physical and human capital while human beings and institutions stayed the same. A later view allowed for endogenous institutions. Behavioral economics allows people to change, too. Behavioral development economics can explain the possibility of large social and economic changes without shifts in capital stocks or economic incentives, but it cannot yet predict when they will occur. And it does not know yet what are the limits of change.       31      REFERENCES    Acemoglu, Daron, Simon Johnson, and James A. Robinson. 2002. “Reversal of Fortune” Quarterly Journal  of Economics 117 (4):1231–94.  Akerlof, George. 1970. “The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism.”  Quarterly Journal of Economics 84(3):488‐500.  _________ 1984. An Economic Theorist’s Book of Tales, Cambridge: Cambridge University Press.  _________ and Robert J. Shiller.  2015.  Phishing for Phools. Princeton, NJ: Princeton University Press.  Aleem, Irfan.  1993. Imperfect information, screening, and the costs of informal lending.  In:  Hoff, K.,  Braverman, A., and J.E.Stiglitz. Oxford University Press, 131‐153.  Alesina, Alberto and Paola Giuliano. 2015. “Culture and Institutions.” Journal of Economic Literature  53(4): 898–944  Alesina, Alberto, Paola Giuliano, and Nathan Nunn. 2013. "On the Origins of Gender Roles: Women and  the Plough." Quarterly Journal of Economics 128(2): 469‐530.  Bacharach, Michael. 2003. “Framing and cognition in economics:  The bad new and the good.” In  Cognitive Processes and Economic Behaviour.  Eds N. Dimitri, M. Basili, and I. Gilboa.  London: Routledge.    Banerjee, Abhijit, Esther Duflo, Rachel Glennerster, and Dhruva Kothari. 2010. “Improving immunization  coverage in rural India.” British Medical Journal 340  Banfield, Edward C. 1958. The Moral Basis of a Backward Society. Free Press.  Bates, Robert, Avner Greif, Margaret Levi, Jean‐Laurent Rosenthal, Barry Weingast. 1998. Analytic  Narratives. Princeton, NJ: Princeton University Press.  Berger, Peter and Thomas Luckmann. 1966. The Social Construction of Reality. New York:  Anchor.  Besley, Timothy and Stephen Coate. 1995. “Group lending, repayment incentives and social collateral.   Journal of Development Economics 46(1):1‐18.  Bowles, Samuel. 1998. “Endogenous Preferences: The Cultural Consequences of Markets and Other  Economic Institutions.” Journal of Economic Literature 36(1): 75‐111  Brooks, Benjamin, Karla Hoff, and Priyanka Pandey. 2017.  "High‐caste culture in North India impedes  learning to cooperate.” Manuscript, World Bank.         Bruner, Jerome. 1990. Acts of Meaning. Cambridge: Harvard University Press.  Camerer, Colin F. 2005. "Comments on ‘Development Economics through the Lens of Psychology’ by  Sendhil Mullainathan” in Lessons of Experience: Annual World Bank Conference on Development  Economics, World Bank.  Cohn, Alain, Ernst Fehr, and Michel Andre Marechal. 2014. “Business culture and dishonesty in the  banking industry.” Nature  516, 86–89 (04 December 2014)  32      Datta, Saugato and Mullainathan, Sendhil. 2014. "Behavioral Design: A New Approach to Development  Policy." Review of Income and Wealth Series 60(1)  Davis, John B., 2013. “Economics Imperialism under the Impact of Psychology: The Case of Behavioral  Development Economics.” OEconomia, 3‐1: 119‐138.  Deaton, Angus. 2010. “Instruments, Randomization, and Learning about Development.” Journal of  Economic Literature 48(2):424‐455  Demeritt, Allison and Karla Hoff. 2015. “Small Miracles – Behavioral Insights to Improve Development  Policy: World Development Report 2015” in Contemporary Issues in Development Economics, edited by  Tim Besley, Palgrave Macmillan, Chapter 3, pp. 19‐43.    Denzau, Arthur T. and Douglass C. North. 1994. “Shared mental models: Ideologies and Institutions.”  Kyklos 47(1), pp. 3‐31.  DiMaggio, Paul, 1997. “Culture and cognition.” Annual Review of Sociology 23: 263‐287    DiMaggio, Paul and Hazel Markus. 2010. “Culture and Social Psychology: Converging Perspectives” Social  Psychology Quarterly 73(4): 347‐352    Douglas, Mary. 1966.  Purity and Danger.  New York: Routledge.  ____________. 1986.  How Institutions Think. Syracuse, NY: Syracuse University Press.  Drexler, Alejandro, Greg Fischer, and Antoinette Schoar. 2014. "Keeping It Simple: Financial Literacy and  Rules of Thumb." American Economic Journal: Applied Economics, 6(2):1‐31.  Duflo, Esther, Michael Kremer and James Robinson. 2008. “How High Are Rates of Return to Fertilizer?  Evidence from Field Experiments in Kenya.” American Economic Review 98(2): 482‐488.   Duflo, Esther, Michael Kremer and James Robinson. 2011. “Nudging Farmers to Use Fertilizer: Theory  and Experiments Evidence from Kenya.” American Economic Review 10(6): 2350‐2390  Duflo, Esther. 2017. "Richard T. Ely Lecture: The Economist as Plumber." American Economic Review,  107(5): 1‐26  Elster, Jon. 2007. Explaining Social Behavior. Cambridge: Cambridge University Press  Fehr, Ernst and Karla Hoff. 2011. “Introduction—Tastes, Castes and Culture: The Influence of Society on  Preferences.” Economic Journal 121(556): F396‐F412  Fennell, Lee Anne (2011). "Ostrom's Law: Property rights in the commons". International Journal of the  Commons. 5 (1): 9–27.  Foster, Andrew and Mark Rosenzweig.  1995.  “Learning by doing and learning from others: Human  capital and technical change in agriculture.”  Journal of Political Economy 103(6):1176‐1209.  Fryer, Roland G., Steven D. Levitt, John List and Sally Sadoff. 2012. “Enhancing the Efficacy of Teacher  Incentives through Loss Aversion: A Field Experiment.” NBER Working Paper No. 18237.  33      Fujiwara, Thomas. 2015. “Voting technology, political responsiveness, and infant health:  Evidence from  Brazil.” Econometrica 83(2):423‐64.  Gawande, Atul. 2013. “Slow Ideas.” The New Yorker, July 29.  Goldstein, Daniel G. and Gerd Gigerenzer. 2002. “Models of Ecological Rationality: The Recognition  Heuristic.” Psychological Review 109(1): 75‐90  Goffman, Erving. 1974. Frame Analysis: An Essay on the Organization of Experience. Cambridge: Harvard  University Press  Greif, Avner. 1994. “Cultural Beliefs and the Organization of Society.” Journal of Political Economy  102(5): 912‐950  Guiso, Luigi, Paola Sapienza, and Luigi Gonzales. 2016. “Long‐term persistence.” Journal of the European  Economic Association 14(6):1401–1436  Hamermesh, Daniel. 2013. "Six decades of top economics publishing: Who and how?" Journal of  Economic Literature, 51(1):162‐172.  Hanna, Rema, Sendhil Mullainathan and Joshua Schwartzstein. 2014. “Learning Through Noticing:  Theory and Evidence from a Field Experiment.” Quarterly Journal of Economics 129(3):1311‐1353  Hedstrom, Peter and Richard Swedberg. 1998. Social Mechanisms: An Analytical Approach to Social  Theory (Studies in Rationality and Social Change). Cambridge: Cambridge University Press  Hedstrom, Peter and Petri Ylikoski. 2010. “Causal Mechanisms in the Social Sciences.” Annual Review of  Sociology 36: 49‐67   Heller, S. B., Shah, A. K., Guryan, J., Ludwig, J., Mullainathan, S., & Pollack, H. A. (2017). “Thinking, fast  and slow? Some field experiments to reduce crime and dropout in Chicago.” Quarterly Journal of  Economics. 132(1):1‐54.  Heukelom, Floris. 2014. Behavioral Economics: A History. Cambridge, UK:  Cambridge University Press.  Hirschman, Albert O. 1982.  “Rival interpretations of market society:  Civilizing, destructive, or feeble?”   Journal of Economic Literature 20:1463‐1484.  Hoff, Karla. 2003. “Paths of institutional development:  A view from economic history.” World Bank  Research Observer 19(2): 205‐226.  Hoff, Karla and Joseph. E. Stiglitz. 2010. “Equilibrium fictions.” American Economic Review100: 141‐146.    Hoff, Karla and Joseph E. Stiglitz. 2016.  “Striving for balance in economics:  Towards a theory of the  social determination of behavior.” Journal of Economic Behavior and Organization 126(B):25‐57.  Iyer, Lakshmi, Anandi Mani, Prachi Mishra, and Petia Topalova. 2012. "The power of political voice:  Women's political representation and crime in India." American Economic Journal: Applied Economics,  4(4): 65‐193.  34      Kahneman, Daniel. 2011. Thinking, Fast and Slow.  New York:  Farrar, Straus, and Giroux.Keynes, John  Maynard. 1936. The General Theory of Employment, Interest, and Money. London: Macmillan and Co.,  limited  Khemani, Stuti. 2017. “Political economy of reform.” World Bank, manuscript.   Kremer, Michael and Gautam Rao.” In progress.  “Behavioral economics and development.” In  Bernheim, DellaVigna, and Laibson (eds.), Handbook of Behavioral Economics, in progress  Ludwig, Jens, Jeffrey Kling, and Sendhil Mullainathan. 2011. "Mechanism experiments and policy  evaluations." Journal of Economic Perspectives, 25(3):17‐38.  Markus, Hazel and Shinobu Kitayama. 2010. “Cultures and selves:  A cycle of mutual constitution.”  Perspectives on Psychological Science 5(4):420‐430  Merton, Robert. 1949. Social Theory and Social Structure. New York:  Free Press.  Meyer, John W. and Brian Rowan. 1977. “Institutionalized Organizations: Formal Structure as Myth and  Ceremony.” American Journal of Sociology 83(2): 340‐363   Meyer, John W., Joane Nagel and Conrad W. Snyder Jr. 1993. “The Expansion of Mass Education in  Botswana: Local and World Society Perspectives.” Comparative Education Review 37(4): 454‐475  Mokyr, Joel. 2017.A Culture of Growth:  The Origins of the Modern Economy. Princeton:  Princeton  University Press.  New York Times. 2017. “How Big Business Got Brazil Hooked on Junk Food.” Sept. 17,  A1.  North, Douglass C. 1990. Institutions, institutional Change and Economic Performance, Cambridge:  Cambridge University Press  North, Douglass C.  1994. “Economic performance through time.” American Economic Review 84(3):359‐68.  North, Douglass C. 2005.  Understanding the Process of Economic Change.  Princeton: Princeton  University Press.   Nunn, Nathan. 2012. “Culture and the historical process,” Economic History of Developing Regions, Vol.  27, Supplement 1:108‐126.  Ostrom, Elinor (1990). Governing the Commons: The Evolution of Institutions for Collective Action.  Cambridge, UK: Cambridge University Press.  Parker, Ian. 2010. “The Poverty Lab: Transforming development economics, one experiment at a time.”  The New Yorker, May 7.  Pritchett, Lant. 2014. “The Risks to Education Systems from Design Mismatch and Global Isomorphism”  Center for International Development Working Paper No. 277.   Putnam, Robert. 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton: Princeton  University Press.  35      Rabin, Matthew.  2013. “An approach to incorporating psychology into economics.”  American Economic  Review 103(3): 617‐622.  Ray, Debraj. 2008. “Development Economics”, in The New Palgrave Dictionary of Economics, eds.  Lawrence Blume and Steven Durlauf. Palgrave Macmillan  Ross, Lee, and Richard E. Nisbett. 1991. The Person and the Situation.  London:  McGraw‐Hill  Simon, Herbert A. 1957 [1947]. Administrative Behavior: A Study of Decision‐Making Processes in  Administrative Organization (2nd ed.). New York: Macmillan.  Smith, Adam. 1982 [1759]. The Theory of Moral Sentiments. Indianapolis: Liberty Fund.  Sokoloff, Kenneth and Stanley Engerman. 2000. "Institutions, factor endowments, and paths of  development in the New World." Journal of Economic Perspectives, 14(3):217‐232.   Stigler, G. J. and Becker, G. 1977. De gustibus non est disputandum, American Economic Review  67(2):76–90.  Stiglitz, Joseph E. 1974. “Incentives and risk sharing in sharecropping. Review of Economic Studies 41:219‐255.  Swedberg, Richard.  1990.  Economics and Sociology. Princeton: Princeton University Press  Swidler, Ann. 1986. “Culture in Action: Symbols and Strategies.” American Sociological Review 51(2):  273‐286.  Thaler, Richard.  2000. Misbehaving: The Making of Behavioral Development Economics. W.W.Norton.  ______ and Cass Sunstein. 2008. Nudge. New York: Penguin Books.   Tversky, Amos and Daniel Kahneman. 1981. “The Framing of Decisions and the Psychology of Choice.”  Science. 211(30):453‐457  ___________________. 1986. “Rational Choice and the Framing of Decisions” Journal of Business  59(4:  Part 2). S251‐S278.  World Bank Group. 2015. World Development Report: Mind, Society and Behavior, Washington DC.  World Bank Group. 2017. World Development Report 2018: LEARNING to Realize Education’s Promise,  Washington DC.  Young, H., Peyton, and Mary A. Burke. 2001. "Competition and Custom in Economic Contracts: A Case  Study of Illinois Agriculture." American Economic Review, 91(3): 559‐573.  36