report Overcoming behavioral obstacles to escaping poverty Christopher J. Bryan, Nina Mazar, Julian Jamison, Jeanine Braithwaite, Nadine Dechausay, Alissa Fishbane, Elizabeth Fox, Varun Gauri, Rachel Glennerster, Johannes Haushofer, Dean Karlan, & Renos Vakis abstract International development policy is ripe for an overhaul. Behavioral science can help policymakers to spur changes in behaviors that are difficult to explain from a conventional economic perspective and impede economic development. We focus here on two well-documented, often coinciding psychological phenomena that have particularly wide- ranging implications for development policy: present bias (favoring immediate rewards over long-term considerations) and limited attention. We present a number of general policy recommendations that are informed by insight into these phenomena and offer concrete examples of how the recommendations can be implemented to help low-income individuals improve their lives and reach their long-term goals. Bryan, C. J., Mazar, N., Jamison, J., Braithwaite, J., Dechausay, N., Fishbane, A., . . . Vakis, R. (2017). Overcoming behavioral obstacles to escaping poverty. Behavioral Science & Policy, 3(1), 81–91. H ow can international development poli- fail to expend small amounts of money, time, or cies induce farmers to adopt improved effort up front to obtain much larger benefits in agricultural technologies, get parents the future. When it comes to trading off between to vaccinate their children, prompt patients immediate and future outcomes, such decisions to comply with treatment regimens, and depend on the relative weight one assigns to encourage poor people to save more? These results achieved now versus later on. The pull seemingly disparate challenges have a common of instant gratification often keeps people from feature: insights from behavioral science can making the optimal choices they say they would help to improve the effectiveness of efforts to have made if someone had asked them to reflect address them. on those decisions when not under the imme- Core Findings diate influence of temptation. In other words, For example, the standard rational perspective present bias—overweighting short-term versus What is the issue? of classic economic theory would predict that long-term rewards—gets in the way. 2 International development offering a higher interest rate should motivate policy should take human people to save more. A recent field experiment This deviation from optimality occurs frequently. psychological phenomena into account as well as in Chile found, however, that a large majority of In the abstract, people often prefer to make the classic economic theory. participants did not increase savings in response long-run investment but then are tempted in the In particular, research to this approach, even though interest rates moment to take the immediate benefit, only to shows that individuals exhibit irrational biases increased substantially, from 0.3% to 5%. By regret the choice later. 3 For example, a parent toward the present, contrast, savings almost doubled when subjects who knows she should be saving for her child’s and poverty limits were able to announce their savings goals to a school fees might falter and purchase a tempting their attention spans. self-help group and had their progress publicly meal if she walks past a restaurant when she Developing interventions that account for these monitored and rewarded in nonmonetary is hungry. Conversely, a small but unpleasant phenomena can boost ways—such as with praise—at the group’s weekly obstacle right now can have a large influence uptake and effectiveness. meetings.1 Thus, a basic understanding of even a on decisions: a parent might want to vaccinate small number of the principles that guide human her child, but the prospect of a long, hot walk to How can you act? Selected behavior can help policymakers to alter behav- the clinic (when she doesn’t know for sure that recommendations include: iors that make little sense from a conventional the clinic will even be open) might lead her to 1) Timing the delivery economic perspective and pose challenges to procrastinate—perhaps indefinitely. of interventions for when people are most economic development. likely to be receptive, Present bias is common to those in rich and poor such as after a harvest We discuss two well-studied psychological countries alike. 2 Behavioral scientists have not 2) Offering programs that lock in or otherwise phenomena that have wide-ranging implica- only documented the phenomenon but have increase commitment tions for international development policy: also worked with international development to savings present bias and limited attention. For clarity, we experts and policymakers to design programs 3) Using cognitive aids begin by explaining the two concepts separately, that take it into account. Many of these programs to remind people of optimal behavior although they operate concurrently in many of have been rigorously tested and proved to be the situations we discuss. effective at changing behavior in ways that lead Who should take to positive long-run outcomes. the lead? Behavioral science Present Bias Deters researchers, policymakers in development Investment in the Future Limited Attention Impairs Investing in the future is critical to people’s well- Decisionmaking being. Such investments can take many forms, To understand poverty, one must recognize that such as saving to buy business supplies without its defining features—the shortage of money, paying exorbitant interest rates to a money- time, and basic necessities such as sleep and lender, purchasing fertilizer to improve next food—affect psychological functioning in year’s crop yield, sending children to school, or nonobvious ways that can undermine poor traveling to get preventative medical care. These people’s ability to escape their circumstances. examples might sound like obvious steps to take, This is true even when policies or programs are but behavioral science reveals that people often implemented that, in principle, provide sufficient 82 behavioral science & policy  |  volume 3 issue 1 2017 opportunities for people to pull themselves can be important barriers to investment in future out of poverty. Everyone has limited atten- well-being. tional bandwidth, but wealthy people, freed from having to spend this precious attention on A key practical policy lesson that flows from this acquiring food, shelter, and other basics, have understanding is that the way to battle procras- more attention available for handling unex- tination in well-being investments is to reduce pected hassles and making strategic decisions and ideally abolish the up-front cost of obtaining to improve their circumstances. In contrast, health products that offer substantial benefits at the challenge of navigating everyday life when reasonable prices but go underutilized. Fifteen one lacks adequate resources is enormous. randomized trials showed dramatic increases in Poor people are often left with little or no spare uptake in response to even small reductions in attentional capacity to devote to such important prices for products such as insecticide-treated things as remembering to take their pills every day or navigating the complicated bureaucratic process to qualify for an assistance program. bed nets (ITNs) for avoiding mosquito-borne diseases, dilute chlorine for disinfecting drinking water, and deworming tablets.6 This principle 4m lives saved in sub-Saharan Making matters worse, poverty directly affects helped catalyze large-scale distribution of free Africa from insecticide- the environment in which people live, which ITNs in sub-Saharan Africa, an effort that is esti- treated bed net interventions, since 2000 often creates additional attentional demands. For mated to have saved 4 million lives since 2000.7 example, lack of access to such basic services as Similar actions could produce cost effective piped water, electricity, child care, and affordable increases in the use of many other prophylactic financial services adds numerous daily decisions products that can increase the well-being of to the cognitive plate of a person in poverty, people living in the developing world. whose attentional bandwidth is already scarce.4,5 Likewise, reducing the up-front costs associated with education could yield outsized bene- the cost of a school Principles for Policymakers fits. One study illustrating this point found that uniform intervention in Kenya associated In general, policies aimed at serving the poor will providing free school uniforms to students in with a 6.4 percentage be more effective if they alleviate the difficulties Kenya at a cost to the state of $6 a student, a point increase imposed by present bias and limited attention. small fraction of the total cost of a child’s educa- in attendance is $6 per student Although both conditions are pervasive across tion, led to a 6.4 percentage point increase in humanity, they take a greater toll on the well- school attendance.8 Helping countries reduce being of those experiencing scarcity than on the or eliminate school fees and giving vouchers for well-being of those who are wealthier. Next, we discuss several policy strategies that can achieve this goal and provide evidence of their effective- free school uniforms are practical and straight- forward policies that could improve school enrollment in places where it is low. 33% increase in immunization ness in a range of sectors. rates in rural Rajasthan Beyond reducing fees for long-term investments, associated with the Reduce the Up-Front Cost of minimizing or eliminating what might seem like provision of free lentils at clinics Future-Oriented Behavior trivial inconveniences can dramatically increase Everyone has some tendency to procrastinate; the uptake of services. This approach could people delay doing what they know is in their include strategies such as reducing or simplifying long-term interest because they usually have paperwork (or better yet, instituting automatic no compelling reason to bear the up-front cost enrollment in programs), minimizing travel today when they can put it off until tomorrow. times required to take advantage of programs, The narrowing of attention produced by and helping with child care and transportation. poverty—focusing on immediately pressing In one instance, helping households to fill out needs to the exclusion of other important but the application for an interest-free loan to cover less urgent needs 5 —aggravates this natural the cost of piped water in Morocco increased present bias. As a result, even minor up-front participation from 10% to 69%.9 (This jump costs, such as small copayments, minor incon- mirrors the U.S. finding that helping families fill veniences, or the need to expend extra effort, in FAFSA forms for federal student aid increased a publication of the behavioral science & policy association 83 “It can sometimes be better to charge a small fee and make a service very convenient than to charge nothing for a very inconvenient service” low-income students’ first-year college atten- is particularly difficult for pregnant women and dance rate by 24%.)10 those with young infants. Ideally, all pregnant women would undergo at least one prenatal The need to travel even modest distances (that checkup (so a medical professional can assess is, more than a 10- to 15-minute walk) is another risk factors and encourage the mother to have type of inconvenience that can powerfully a trained attendant at the birth) and all infants dampen service uptake. In Malawi, the likelihood would receive basic immunizations. Evidence that people would show up to receive the results suggests that use of such services would of an HIV test fell sharply when the distance increase dramatically if they were provided they needed to travel increased by even a small within villages or at least at coordinated central amount. 11 Similarly, in Kenya, the likelihood locations accessible by cheap and easy trans- that people would take advantage of protected portation and if other forms of assistance were springs as a water source that reduced the risk of available (for example, a teen helper coming to diarrhea fell with small increases in the distance the woman’s door to accompany her or watch they had to travel to reach the water.12 her other children while she went for a checkup). Conversely, in situations where logistical Because price and inconvenience are both constraints require that services be provided at barriers to investing in future well-being, poli- less convenient locations, small (but immediate) cymakers should think carefully about the material incentives (for example, a bag of lentils trade-offs between them. One might assume and a set of metal plates) can be an effective that the poor would be willing to endure signif- way to offset inconvenience. In India, free lentils icant inconvenience to avoid even a small increased immunization rates in rural Rajasthan financial cost for services, but this assumption from 6% to 39%.14 has a serious flaw: it fails to appreciate that over- coming inconvenience requires attention (to Time the Delivery of Subsidies for When plan for and solve logistical challenges) that poor People Are Most Likely to Be Receptive people cannot spare. Therefore, it can some- Both present bias and limited attention suggest times be better to charge a small fee and make a that the timing of interventions can be critically service very convenient than to charge nothing important in ways that are not obvious from a for a very inconvenient service. traditional economic perspective. For example, sugarcane farmers in India typically receive their This point is illustrated by the success of a income once a year—at the time of harvest—and nonprofit entrepreneurial program for deliv- therefore tend to be relatively rich right after ering preventive health products in rural Uganda. the harvest and relatively poor right before it. A randomized evaluation found impressive In a powerful illustration of both the attentional community health gains when women sold costs of poverty and the importance of timing, underused health products such as ITNs, water a recent study documented that these farmers purification tablets, and antimalarial drugs door to perform worse on tests of sustained attention in door at a discounted (but nontrivial) price, elimi- the period immediately before the harvest, when nating the hassle of seeking these products out.13 money is tight. The difference in scores trans- lates to roughly 10 IQ points.15 Charging a bit to reduce inconvenience is a very promising approach that deserves to be scaled Traditionally, the timing of subsidies has been up. Notably, it could be expanded to improve determined arbitrarily, presumably on the maternal and child health broadly, because travel assumption that a subsidy delivered now is 84 behavioral science & policy  |  volume 3 issue 1 2017 “At times, people will take elaborate steps to protect themselves from succumbing to short-term temptations” at least as useful as a subsidy delivered later. advance notice of the subsidy and possibly even But recent evidence from behavioral science help with planning and budgeting, to ensure that and development research indicates that this they have money available to pay for expenses approach misses an opportunity to enhance beyond those covered by the subsidy. uptake: it would be more effective to give subsi- dies at times when people are most likely to Offer Programs That Lock In or Otherwise have the attentional bandwidth needed to think Increase Commitments to Savings about and take full advantage of them. It seems People are often well aware that temptation or probable, for instance, that the low-income distraction at critical moments can derail their sugarcane farmers would be in a better mental pursuit of long-term goals. As a result, to keep state to evaluate and accept a beneficial offer themselves on track, they may be willing—even immediately after the harvest, when they face eager—to subject themselves to costly penalties fewer pressing demands. for failing to stick to their goals.18 African farmers living in poverty offer an example of how such Aligning the timing of subsidies with the timing commitment savings approaches can be made of important decisions or expenses is another to work. Impoverished farmers sometimes effective strategy. In Tanzania, promoters of underuse technologies that they say they want health insurance deliberately went to the distri- and know can increase profits. This is probably bution points of a cash transfer program to partly because they get paid at harvest but do sign people up for health insurance when they not need hybrid seed and fertilizer until months received the transfers (and therefore had greater later; holding on to their money that long can liquidity). This deliberate timing contributed to be hard. Offering a small, time-limited discount a 20 percentage point increase in the use of on the cost of acquiring fertilizer (for example, health insurance. 16 Similarly, farmers respond in the form of free delivery) right after harvest, more favorably to the promotion of agricultural when money is relatively plentiful, is a form of products (such as fertilizer and hybrid seeds) if commitment savings that has been found to approached at harvest time, when they have increase purchase rates of fertilizers in Kenya money available for those investments and by 11 percentage points. 19 Estimates suggest when their attentional capacity is not overly that to produce a similar purchase rate later on, taxed by the need to grapple with scarce finan- when fertilizer would normally be bought, a 50% cial resources. Finally, subsidies to encourage subsidy of the purchase price would be needed. education could be timed to coincide with when school fees are due. In a recent demon- At times, people will take elaborate steps to stration of the value of this approach, a program protect themselves from succumbing to short- in Bogota, Columbia, that offered cash condi- term temptations. 20,21 They may choose, for tionally in exchange for reenrolling children in instance, to lock their money away where they school produced higher rates of reenrollment cannot access it for some predetermined length when a portion of the monthly transfer was post- of time. 22,23 Some people may even pay for this poned until just before the reenrollment period. restriction on their freedom, for example, by Moreover, this time-sensitive design was particu- accepting a lower interest rate on money they larly effective for those who needed it the most cannot easily access on a whim. (and whose families were most likely to be facing scarce liquidity and attention): the students from One concern with commitment devices is that the families with the lowest incomes and the they come with a risk: Locking money away lowest participation rates.17 To maximize effec- means it is not available for unanticipated but tiveness, such programs should give parents genuinely important expenses. This worry a publication of the behavioral science & policy association 85 can prevent people from taking advantage of observable. Consequently, they did not consider commitment devices or can constrain people’s strand size in farming decisions and did not even ability to cope if they do commit and then an know what the size of the strands they used was. urgent situation arises. An alternative, inspired The study showed that merely offering farmers by work on the theory of mental accounting, the opportunity to observe how researchers is soft commitments, such as labeling a savings varied the size of the strands and the effect of account for particular expenditures (like educa- that variable on yield was not enough for farmers tion) without a strict constraint on how the to notice the relationship. Only when researchers money in it can actually be spent. 24 In a recent explicitly pointed out the relationship between study in Uganda, researchers compared a strand size and yield did farmers notice it and program in which saved money could only be change their practices. 26 This result has nothing used for educational expenses with a program to do with the intelligence of the farmers. A fact in which the savings were encouraged but is only obvious if the observer has the spare not required to go to education (that is, it was attentional capacity to notice it. 27 possible to simply withdraw the cash). In both cases, families saved more and spent more on Simple reminders are another type of straight- education supplies than a control group did. But forward cognitive aid that can be surprisingly families saved the most money in the latter case, beneficial. All people sometimes forget to do when they knew they could still withdraw the things they meant to do—take pills, mail the money for other things if they needed to. 25 rent check, and so on. But, perhaps unsurpris- ingly, when attention is overtaxed, people are Thus, making commitment devices available even less likely to follow through with intended (and easy to use) can be an effective tool—and actions. When attention is completely taken up one that is even sought out by individuals who with pressing demands, people are unlikely to recognize their susceptibility to short-term temp- step back and ask whether they are forgetting to tation, poor planning, and distraction—especially do something. A policy problem that exemplifies at times of peak demand on people’s limited the worsened intention–action gap that occurs attentional resources. These tools are, however, when bandwidth is constrained is the incom- not useful for all individuals, and softer commit- plete adherence to medical treatment regimens ments, such as earmarking an account for for conditions like tuberculosis or HIV/AIDS. In particular expenses, may be preferable in situa- the case of HIV, patients commonly receive a tions when more flexibility is required. 1-month supply of pills and must remember to take those pills every day. Even when patients Introduce Cognitive Aids understand and genuinely intend to adhere to Because poor people often have to attend to their treatment, they often forget to do so amid multiple pressing needs at the same time, the the chaos of other pressing demands on their limits of their attention are continually strained.5 attention. The consequences of such forgetting Thus, it is not surprising that they may be more can be life-threatening, but a simple fix can help. likely than others to miss crucial information For example, research in rural Kenya demon- or forget to take intended actions that could strated that the percentage of HIV patients who improve their welfare. Sometimes, statements achieved perfect or near-perfect treatment explicitly pointing out what might seem obvious adherence (that is, at least 90%) during the nearly to a person not suffering from attentional scar- yearlong study period increased from 40% to city can make a big difference. In a recent study, 53% when they received weekly text-message experienced seaweed farmers in Indonesia had reminders. 28 noticed that the spacing between their seaweed strands affected their yield, so they paid atten- Similarly, although breastfeeding is considered tion to the spacing when planting the strands. the best practice for nourishing babies (especially But the farmers failed to notice that the size because high-quality infant formula and clean of the strands they planted also affected their water are not available in much of the devel- yield, even though the lower yield was easily oping world), competing responsibilities—such 86 behavioral science & policy  |  volume 3 issue 1 2017 as household chores or caring for older chil- dren—can make keeping it up difficult. Simple “opportunities exist to layer cognitive aids can help, including, for example, physical reminders, such as stickers on bottles, behavioral interventions on top of existing programs” that note that bottles are appropriate primarily for older infants and toddlers. Sometimes aids that might seem unnecessary to a person whose attention is not overburdened thumb for design issues such as timing, length, can be enormously helpful to someone whose and frequency of reminders; mode of delivery; attention is overwhelmed. Simple actions, content; and framing of messages. But even like pointing out well-known facts at the right with more research, general rules can offer only time or sending well-timed reminders, can be limited guidance about the optimal implementa- important tools to improve decisionmaking tion of a policy. It is often difficult to predict how among the poor. Reminder messages in partic- cultural differences and unobserved variation ular have been delivered in field experiments by between contexts might influence the effect of text message, e-mail, postcard, letter, phone, even a well-researched treatment. Thus, wher- and in-person survey. They have been shown ever feasible, any new policy applying behavioral to improve a wide range of outcomes, including principles should be evaluated rigorously in the saving rates in Uganda;25 loan repayment in context in which it is meant to be implemented Bolivia, Peru, and the Philippines;29,30 compli- before being deployed at scale (as should all ance with obligatory child support payments new policies). in the United States;31 vaccination rates in rural Guatemala;32 use of water treatment products A second issue is that although many findings in Kenya;33,34 and payment of delinquent fines in demonstrate that the strategies listed here have the United Kingdom. 35 But reminders must not had significant effects in the short run, little is be too frequent or they risk crossing the line known about how long the effects last. This from useful aid to additional drain on limited uncertainty is immaterial in situations where the attention. 28 Also, they are likely to be especially goal is to encourage one-off actions, such as effective for irregular events, such as immuni- when sending a one-time reminder to get chil- zation visits, for which people are less able to dren vaccinated. It is more of a concern when the form a habit. effectiveness of a policy or program depends on people taking sustained, repeated action to form a new habit, as is the case when daily reminders A Need for Experimentation are sent with the intention of increasing compli- A couple of issues relating to these strategies ance with long-term medical regimens. Further merit consideration. When tested, certain minor research is needed to clarify the long-term variations often work better than others—some- effects of some of these techniques. times in ways and for reasons that would have been difficult to anticipate without testing. This Policymakers are in an ideal position to conduct not only suggests the need for more experi- much of this research. They are often mandated mentation but also underscores the sometimes to implement specific programs in particular surprising impact of subtle design features. For settings and populations, which seems to leave example, not all reminders are equally effective. little room for experimentation of the type Although weekly messages worked very well for described above. But because many of these HIV treatment adherence in rural Kenya, an alter- interventions are inexpensive or free to imple- native design with daily messages did not affect ment, opportunities exist to layer behavioral adherence (presumably because too-frequent interventions on top of existing programs. For messages are ignored—or, worse, become an instance, automated reminder text messages added cognitive burden). 28 Additional research can be sent in bulk at extremely low cost. There- is needed to provide generalizable rules of fore, an existing program to promote vaccination a publication of the behavioral science & policy association 87 Sample behavioral strategies to enhance the effectiveness of development programs and policies Psychological phenomenon Recommended behind policy strategy recommendation Sample policies Reduce the up-front Present bias and • Reduce or abolish copayments for underutilized preventive health cost of future-oriented limited attention products such as insecticide-treated bed nets, hand soap, or family behavior planning products. A • Reduce logistical hurdles and, where relevant, the potential embarrassment associated with the uptake of preventive health and family planning products by organizing entrepreneurs to sell such products (at discounted prices) door to door, increasing convenience and privacy.B • Reduce bureaucratic hurdles to program uptake through automatic enrollment or simplified paperwork.C • Reduce travel times to take advantage of programs such as prenatal health care, either by providing such services within villages or by organizing easy, low-cost transportation to central locations.D Time subsidies for when Present bias and • Offer beneficial but high-cost products or services (for example, health people are most likely limited attention insurance) at times when people have greater liquidity (for example, to be receptive, such as right after a cash transfer) and more spare attentional capacity to when they are making evaluate offers.E important decisions or • Align the timing of cash transfers with the time at which school fees are outlays due to encourage school enrollment.F Offer programs that Present bias and • Incentivize the purchase of farming technologies (for example, fertilizer, that lock in or otherwise limited attention hybrid seed) immediately after the harvest, alleviating the need for facilitate savings farmers to save money from the harvest until the next year’s planting season.G • When the inflexibility of hard commitments discourages participation or risks imposing undue costs on people, offer soft commitments, such as savings programs that are earmarked for specific expenses (for example, education) but still allow the savings to be used for other purposes.H,I Introduce cognitive aids Limited attention • Provide text, e-mail, postcard, letter, or phone reminders of the need for important actions, such as taking HIV medication, contributing to savings accounts, or using water treatment products.F,J–L A. Bhatt, S., Weiss, D. J. W., Cameron, E., Bizansio, D., Mappin, B., Dalrymple, U., . . . Gething, P. W. (2015, October 8). The effect of malaria control on Plasmodium falciparum in Africa between 2000 and 2015. Nature, 526, 207–211. http://dx.doi.org/10.1038/nature15535 B. Guariso, A., Nyqvist, M., Svensson, J., & Yanagizawa-Dott, D. (2016). An entrepreneurial model of community health delivery in Uganda [Working paper]. Cambridge, MA: Abdul Latif Jameel Poverty Action Lab. C. Devoto, F., Duflo, E., Dupas, P., Parienté, W., & Pons, V. (2012). Happiness on tap: Piped water adoption in urban Morocco. American Economic Journal: Economic Policy, 4(4), 68–99. http://dx.doi.org/10.1257/pol.4.4.68 D. Kremer, M., Leino, J., Miguel, E., & Zwane, A. P. (2011). Spring cleaning: Rural water impacts, valuation, and property rights institutions. The Quarterly Journal of Economics, 126, 145–205. http://dx.doi.org/10.1093/qje/qjq010 E. Evans, D. K., Hausladen, S., Kosec, K., & Reese, N. (2014). Community-based conditional cash transfers in Tanzania: Results from a randomized trial. Washington, DC: World Bank. F. Barrera-Osorio, F., Bertrand, M., Linden, L. L., & Perez-Calle, F. (2011). Improving the design of conditional transfer programs: Evidence from a randomized education experiment in Colombia. American Economic Journal: Applied Economics, 3(2), 167–195. http://dx.doi.org/10.1257/app.3.2.167 G. Duflo, E., Kremer, M., & Robinson, J. (2011). Nudging farmers to use fertilizer: Theory and experimental evidence from Kenya. The American Economic Review, 101, 2350–2390. Retrieved from http://dx.doi.org/10.1257/aer.101.6.2350 H. Thaler, R. H. (1999). Mental accounting matters. Journal of Behavioral Decision Making, 12, 183–206. Retrieved from http://dx.doi.org/10.1002/ (SICI)1099-0771(199909)12:3<183::AID-BDM318>3.0.CO;2-F I. Karlan, D., & Linden, L. L. (2014). Loose knots: Strong versus weak commitments to save for education in Uganda (NBER Working Paper No. 19863). Retrieved from National Bureau of Economic Research website: http://www.nber.org/papers/w19863 J. Pop-Eleches, C., Thirumurthy, H., Habyarimana, J. P., Zivin, J. G., Goldstein, M. P., de Walque, D., . . . Bangsberg, D. R. (2011). Mobile phone technologies improve adherence to antiretroviral treatment in a resource-limited setting: A randomized controlled trial of text message reminders. AIDS, 25, 825–834. http://dx.doi. org/10.1097/QAD.0b013e32834380c1 K. Ahuja, A., Kremer, M., & Zwane, A. P. (2010). Providing safe water: Evidence from randomized evaluations. Annual Review of Resource Economics, 2, 237–256. http://dx.doi.org/10.1146/annurev.resource.012809.103919 L. Luoto, J., Levine, D., & Albert, J. (2009). Information and persuasion: Achieving safe water behavior in Kenya [Working paper]. Retrieved from University of California, Berkeley, website: http://faculty.haas.berkeley.edu/levine/papers/Luoto_Marketing_Information_May2011.pdf 88 behavioral science & policy  |  volume 3 issue 1 2017 (such as a vaccination camp) could easily and somewhere in between and is considerably cheaply send text reminders to a randomly more nuanced. Nevertheless, it is now clear that chosen subset of the target population and then behavioral interventions are a valuable tool, and compare the vaccination rates in the groups that when such interventions are combined with did and did not receive the reminders. more conventional policy tools—such as regula- tion, education and training, standard economic Other messaging interventions can be added incentives, and infrastructure—they can help to existing programs in similarly straightforward ameliorate poverty and improve well-being. ways, especially when the program already includes communication with potential recip- ients. For instance, it is trivial to add a request The Long View for a soft commitment to an existing interac- Living in poverty puts additional and often over- tion with the recipient. Similarly, tests of optimal whelming demands on a person’s attention. This intervention timing can often be conducted attentional burden can intensify present bias without additional cost if programs are rolled out and otherwise impair decisionmaking, causing over a period of time. If, say, fertilizer discounts the poor to miss opportunities to improve their are already being made available to farmers, situation. Behavioral insights suggest techniques policymakers might be in a position to vary the to lessen the negative impact of this atten- timing at which these discounts are announced tional tax on the poor. These techniques often in randomly selected areas and thereby learn complement more traditional approaches to about the differential impact of the program easing the burdens of the poor. Applications as a function of offer timing. (This approach is of the principles outlined here offer tremen- a specific example of a more general method, dous promise for improving the effectiveness of called phase-in design, for achieving random- development programs. ization even when programs are to be delivered to every household or individual in a particular area.) Such piggybacking of behavioral interven- tion tests on existing programs would allow even policymakers with strong and inflexible imple- mentation mandates to discover techniques that could improve the effectiveness of the programs they already have in place. author affiliation Policymakers need to experiment, but they also need to be aware of their own biases. Like other The authors were members of the BSPA Working humans, they have limited attentional bandwidth Group on International Development. and often devote too little thought to decisions because they think they already know the answer Bryan: University of Chicago, Booth School of or because their own cultural, political, or moral Business. Mazar: World Bank and University of perspective constrains their thinking in ways they Toronto. Jamison: World Bank and Innovations might not even notice. Indeed, even technically for Poverty Action. Braithwaite: University of trained professionals at the World Bank recently Virginia. Dechausay: MDRC. Fishbane: ideas42. were shown to make more mistakes when eval- Fox: United States Agency for International uating data that were presented as referring to Development. Gauri: World Bank. Glennerster: a controversial topic in their field than they did Massachusetts Institute of Technology. Haush- when the same data were framed as referring to ofer: Princeton University, National Bureau of a neutral topic.36 Relatedly, personal predisposi- Economic Research, and Busara Center for tions might lead some policymakers to presume Economics. Karlan: Yale University and Inno- that behavioral interventions are ineffective and vations for Poverty Action. Vakis: World Bank. others to see those same interventions as “silver Corresponding author’s e-mail: Christopher bullet” solutions for all problems. The truth lies .Bryan@chicagobooth.edu. a publication of the behavioral science & policy association 89 references 1. Kast, F., Meier, S., & Pomeranz, D. (2012). 11. Thornton, R. L. (2008). The demand 21. Schwartz, J., Mochon, D., Under-savers anonymous: Evidence on for, and impact of, learning HIV status. Wyper, L., Maroba, J., Patel, D., self-help groups and peer pressure as American Economic Review, 98, & Ariely, D. (2014). Healthier by a savings commitment device (NBER 1829–1863. https://dx.doi.org/10.1257/ precommitment. Psychological Working Paper No. 18417). Retrieved aer.98.5.1829 Science, 25, 538–546. http://dx.doi. from National Bureau of Economic org/10.1177/0956797613510950 12. Kremer, M., Leino, J., Miguel, E., & Research website: http://www.nber.org/ Zwane, A. P. (2011). Spring cleaning: 22. Ashraf, N., Karlan, D., & Yin, W. (2006). papers/w18417 Rural water impacts, valuation, and Tying Odysseus to the mast: Evidence 2. O’Donoghue, T., & Rabin, M. (1999). property rights institutions. The from a commitment savings product in Doing it now or later. The American Quarterly Journal of Economics, 126, the Philippines. The Quarterly Journal Economic Review, 89, 103–124. 145–205. http://dx.doi.org/10.1093/qje/ of Economics, 121, 635–672. https://doi. Retrieved from http://dx.doi.org/10.1257/ qjq010 org/10.1162/qjec.2006.121.2.635 aer.89.1.103 13. Guariso, A., Nyqvist, M., Svensson, 23. Ashraf, N. (2010). Female 3. Frederick, S., Loewenstein, G., J., & Yanagizawa-Dott, D. (2016). An empowerment: Impact of a & O’Donoghue, T. (2002). Time entrepreneurial model of community commitment savings product in the discounting and time preference: health delivery in Uganda [Working Philippines. World Development, 38, A critical review. Journal of paper]. Cambridge, MA: Abdul Latif 333–344. https://dx.doi.org/10.1016/j. Economic Literature, 40, 351–401. Jameel Poverty Action Lab. worlddev.2009.05.010 Retrieved from http://dx.doi. 14. Banerjee, A. V., Duflo, E., Glennerster, 24. Thaler, R. H. (1999). Mental org/10.1257/002205102320161311 R., & Kothari, D. (2010). Improving accounting matters. Journal of 4. Banerjee, A. V., & Mullainathan, S. immunisation coverage in rural India: Behavioral Decision Making, 12, (2008). Limited attention and income Clustered randomised controlled 183–206. http://dx.doi.org/10.1002/ distribution. The American Economic evaluation of immunisation campaigns (SICI)1099-0771(199909)12:3<183::AID- Review, 98, 489–493. http://dx.doi. with and without incentives. BMJ, 340, BDM318>3.0.CO;2-F org/10.1257/aer.98.2.489 Article c2220. http://dx.doi.org/10.1136/ 25. Karlan, D., & Linden, L. L. (2014). Loose bmj.c2220 5. Mullainathan, S., & Shafir, E. (2013). knots: Strong versus weak commitments Scarcity: Why having too little means so 15. Mani, A., Mullainathan, S., Shafir, E., to save for education in Uganda (NBER much. New York, NY: Macmillan. & Zhao, J. (2013, August 30). Poverty Working Paper No. 19863). Retrieved impedes cognitive function. Science, from National Bureau of Economic 6. Dupas, P., & Miguel, E. (2017). Impacts 341, 976–980. http://dx.doi.org/10.1126/ Research website: http://www.nber.org/ and determinants of health levels science.1238041 papers/w19863 in low-income countries. In A. V. Banerjee & E. Duflo (Eds.), Handbook of 16. Evans, D. K., Hausladen, S., Kosec, K., 26. Hanna, R., Mullainathan, S., & economic field experiments (Vol. 2, pp. & Reese, N. (2014). Community-based Schwartzstein, J. (2014). Learning 3–94). Amsterdam, The Netherlands: conditional cash transfers in Tanzania: through noticing: Theory and evidence North-Holland. Results from a randomized trial. from a field experiment. The Quarterly Washington, DC: World Bank. Journal of Economics, 129, 1311–1353. 7. Bhatt, S., Weiss, D. J. W., Cameron, E., http://dx.doi.org/10.1093/qje/qju015 Bizansio, D., Mappin, B., Dalrymple, U., 17. Barrera-Osorio, F., Bertrand, M., . . . Gething, P. W. (2015, October 8). The Linden, L. L., & Perez-Calle, F. (2011). 27. Simons, D. J., & Chabris, C. F. (1999). effect of malaria control on Plasmodium Improving the design of conditional Gorillas in our midst: Sustained falciparum in Africa between 2000 and transfer programs: Evidence from a inattentional blindness for dynamic 2015. Nature, 526, 207–211. http://dx.doi. randomized education experiment in events. Perception, 28, 1059–1074. org/10.1038/nature15535 Colombia. American Economic Journal: http://dx.doi.org/10.1068/p281059 Applied Economics, 3(2), 167–195. 8. Evans, D., Kremer, M., & Ngatia, M. 28. Pop-Eleches, C., Thirumurthy, Retrieved from http://dx.doi.org/10.1257/ (2009). The impact of distributing H., Habyarimana, J. P., Zivin, J. G., app.3.2.167 school uniforms on children’s education Goldstein, M. P., de Walque, D., . . . in Kenya. Retrieved from World Bank 18. Bryan, G., Karlan, D., & Nelson, S. Bangsberg, D. R. (2011). Mobile phone website: http://siteresources.worldbank. (2010). Commitment devices. Annual technologies improve adherence to org/EXTIMPEVA/Resources/evans_ Review of Economics, 2, 671–698. antiretroviral treatment in a resource- kenya_uniforms.pdf https://doi.org/10.1146/annurev. limited setting: A randomized controlled economics.102308.124324 trial of text message reminders. AIDS, 9. Devoto, F., Duflo, E., Dupas, P., Parienté, 25, 825–834. http://dx.doi.org/10.1097/ W., & Pons, V. (2012). Happiness on tap: 19. Duflo, E., Kremer, M., & Robinson, J. QAD.0b013e32834380c1 Piped water adoption in urban Morocco. (2011). Nudging farmers to use fertilizer: American Economic Journal: Economic Theory and experimental evidence from 29. Karlan, D., McConnell, M., Mullainathan, Policy, 4(4), 68–99. http://dx.doi. Kenya. The American Economic Review, S., & Zinman, J. (2016). Getting to the org/10.1257/pol.4.4.68 101, 2350–2390. Retrieved from http:// top of mind: How reminders increase http://dx.doi.org/10.1257/aer.101.6.2350 saving. Management Science, 62, 10. Bettinger, E. P., Long, B. T., Oreopoulos, 3393–3411. http://dx.doi.org/10.1287/ P., & Sanbonmatsu, L. (2012). The role of 20. Fishbach, A., & Trope, Y. (2007). Implicit mnsc.2015.2296 application assistance and information in and explicit counteractive self-control. In college decisions: Results from the H&R J. Shah & W. Gardner (Eds.), Handbook 30. Cadena, X., & Schoar, A. (2011). Block FAFSA experiment. The Quarterly of motivation science (pp. 281–294). Remembering to pay? Reminders vs. Journal of Economics, 127, 1205–1242. New York, NY: Guilford Press. financial incentives for loan payments http://dx.doi.org/10.1093/qje/qjs017 (NBER Working Paper 17020). Retrieved 90 behavioral science & policy  |  volume 3 issue 1 2017 from National Bureau of Economic Research website: http://www.nber.org/ papers/w17020 31. Baird, P., Reardon, L., Cullinan, D., McDermott, D., & Landers, P. (2015). Reminders to pay: Using behavioral economics to increase child support payments (OPRE Report 2015-20). Retrieved from Administration for Children and Families website: http:// www.acf.hhs.gov/sites/default/files/ opre/reminders_to_pay_using_ behavioral_economics_to_increase_ child_support_0.pdf 32. Busso, M., Cristia, J., & Humpage, S. (2015). Did you get your shots? Experimental evidence on the role of reminders. Journal of Health Economics, 44, 226–237. http://dx.doi. org/10.1016/j.jhealeco.2015.08.005 33. Ahuja, A., Kremer, M., & Zwane, A. P. (2010). Providing safe water: Evidence from randomized evaluations. Annual Review of Resource Economics, 2, 237–256. http://dx.doi.org/10.1146/ annurev.resource.012809.103919 34. Luoto, J., Levine, D., & Albert, J. (2009). Information and persuasion: Achieving safe water behavior in Kenya [Working paper]. Retrieved from University of California, Berkeley, website: http:// faculty.haas.berkeley.edu/levine/papers/ Luoto_Marketing_Information_May2011. pdf 35. Haynes, L. C., Green, D. P., Gallagher, R., John, P., & Torgerson, D. J. (2013). Collection of delinquent fines: An adaptive randomized trial to assess the effectiveness of alternative text messages. Journal of Policy Analysis and Management, 32, 718–730. http://dx.doi. org/10.1002/pam.21717 36. World Bank. (2014). World development report 2015: Mind, society, and behavior. Washington, DC: Author. a publication of the behavioral science & policy association 91