A Behavioral Economics View of Poverty

Standard theorizing about poverty falls into two camps. Social scientists regard the behaviors of the economically disadvantaged either as calculated adaptations to prevailing circumstances or as emanating from a unique "culture of poverty," rife with deviant values. The first camp presumes that people are highly rational, that they hold coherent and justified beliefs and pursue their goals effectively, without mistakes, and with no need for help. The second camp attributes to the poor a variety of psychological and attitudinal short-fallings that render their views often misguided and their choices fallible, leaving them in need of paternalistic guidance. We propose a third view. The behavioral patterns of the poor, we argue, may be neither perfectly calculating nor especially deviant. Rather, the poor may exhibit the same basic weaknesses and biases as do people from other walks of life, except that in poverty, with its narrow margins for error, the same behaviors often manifest themselves in more pronounced ways and can lead to worse outcomes. In what follows, we illustrate the kinds of insights that might be gained from a behaviorally more realistic analysis of the economic conditions of the poor, and we propose that alternative policies for alleviating poverty be considered.


Introduction
Standard theorizing about poverty falls into two camps. Social scientists regard the behaviors of the economically disadvantaged either as calculated adaptations to prevailing circumstances, or as emanating from a unique "culture of poverty," rife with deviant values. The first camp presumes that people are highly rational, that they hold coherent and justified beliefs and pursue their goals effectively, without mistakes, and with no need for help. The second camp attributes to the poor a variety of psychological and attitudinal short-fallings that render their views often misguided and their choices fallible, leaving them in need of paternalistic guidance.
We propose a third view. The behavioral patterns of the poor, we argue, may be neither perfectly calculating nor especially deviant. Rather, the poor may exhibit the same basic weaknesses and biases as do people from other walks of life, except that in poverty, with its narrow margins for error, the same behaviors often manifest themselves in more pronounced ways and can lead to worse outcomes. In what follows, we illustrate the kinds of insights that might be gained from a behaviorally more realistic analysis of the economic conditions of the poor, and we propose that alternative policies for alleviating poverty be considered.

I. Some General Psychological Insights
Behavioral research has documented the persistent yet shocking capacity of simple situational factors to influence behaviors typically presumed to reflect deep dispositions or preferences.
Consider Milgram's (1974) well-known obedience studies, in which decent people administered purportedly dangerous levels of shock to innocent others, or Darley and Batson's (1973)  where seminarians, late to deliver a practice sermon on the Good Samaritan, failed to stop to help a person in need. As it turns out, the pressures exerted by situational factors can create restraining forces hard to foresee and to overcome, as well as driving forces that can be harnessed to great effect. As Ross and Nisbett (1991) point out, where standard intuition would hold the primary cause of a problem to be human frailty, or the particular weakness of a group of individuals, the social psychologist would often look to situational barriers and ways to overcome them.
Indeed, contrary to major interventions that often prove ineffectual, apparently minor situational details, referred to as "channel factors," can have great impact. The opening up of a channel (such as an a priori commitment, or a first step) may facilitate some behaviors, whereas other behaviors can be blocked by closed channels. In one classic study, college seniors were given persuasive messages about the value of an inoculation against tetanus. While the messages were effective at changing the students' beliefs and attitudes, few actually took the step of getting a tetanus shot.
By contrast, when other students received the same messages but were also given a map of campus with the infirmary circled and asked to decide on a particular time, the percentage of students getting the inoculation increased by an order of magnitude. A more recent study of the utilization of public health services found that show-up at a counseling center was better predicted by people's distance from the closest center than by other individual differences. Thus, simple channel factors such as a map, or mere physical proximity, seem to trump the supposedly greater significance of important health messages.
The above examples concern the behavior in a social context of a system -the human information processing system -that is quite idiosyncratic and complex. Among other things, the psychological carriers of value appear to be gains and losses, rather than final wealth, and diminishing sensitivity yields conflicting risk attitudes for losses and gains. People are loss averse (the loss associated with giving up a good is greater than the utility associated with 2 Behavioral Economics & Poverty obtaining it), which yields "endowment effects" and a reluctance to depart from the status quo (Kahneman and Tversky, 2000).
Also, contrary to standard fungibility assumptions, people compartmentalize wealth and spending into distinct budget categories, such as savings, rent, and entertainment, and into separate mental accounts, such as current income, assets, and future income (Thaler, 1999). People typically show different propensities to consume from their current income (where MPC is high), current assets (where it is intermediate), and future income (where it is low). In addition, people often fail to ignore sunk costs or to consider opportunity costs, and have trouble predicting their future moods and tastes or learning from past experience (Kahneman and Tversky, 2000).
In what follows, we consider the relevance of psychological insights to anti-poverty policy, by focusing on two specific examples, financial choices and welfare participation.

II. Banking and Saving
Between 10 and 20 percent of all households in America are without bank accounts (Caskey, 1994). Not surprisingly, nearly all of the un-banked are poor. The material costs of not having a banking account appear quite high. The un-banked face very high costs to cash their checks and pay their bills. Also, they have to save through "cookie-jar" channels that yield no (and sometime negative) interest income. The rational as well as deviant models require large interventions to alter behavior. Large financial subsidies to banks and community groups may be used to create low-cost bank accounts, and legislation may be passed to force banks to maintain or re-open branches in disadvantaged neighborhoods. Financial education may be encouraged to overcome cultural stereotypes and misdirected attitudes.
While these approaches certainly have merit, they focus on "major" factors. Instead, we suggest that small situational barriers often play a decisive role in preventing the opening of a bank account despite huge benefits. These barriers might be a testy bus ride, challenging hours, or the reluctance to face a contemptuous bank-teller. Such barriers are not unlike the embarrassment and anxiety that impede many people, including medical doctors, from administering medical selfexams which they know to be highly valuable.
Mental accounting studies suggest that unlabelled and easily available money will be spent more freely than money that is "accounted for,'' leading to very low saving rates among the un-banked, who may then resort to negative interest saving vehicles, such as lay-away plans or rent-to-own, which are immediately available and less subject to the adoption barriers that come with bank accounts.

Policy Implications
The behavioral perspective suggests several routes to improve banking and saving choices among the poor. First, policies that establish better "defaults" should be explored. Because recipients of government transfers may find it simpler to allow for the opening of a bank account to which money is electronically deposited, recent initiatives for electronic deposit of government checks may alter welfare recipients' default from un-banked to banked. A subsidy to employers who offer bank accounts and direct deposit may help further.
Second, those working to move the poor into the formal banking sector should be trained to explore possible channel factors. Handing out a map with directions to the bank, asking people to specify a time for an appointment, or having a bank representative available on location when people come to learn about subsidized programs such as the First Accounts program may impact take-up. Community groups involved in helping people with their EITC could also link the opening of an account to EITC refund.
Third, policies should aim to reduce the psychological barriers associated with a bank account.
Previous studies have shown that bank cost structure is viewed as too complicated by many poor, who prefer check-cashing places where pricing is transparent. Our own survey suggests that many lower-income people who tried banking were left with bad associations due to unanticipated account fees. A government policy aimed at mandating simpler fee structures may prove highly effective. Even if that leads to higher overall fees, the simplicity of the structure may help to avoid the common tendencies to defer, procrastinate, or renounce altogether because of complications.
People's identity salience may also play a supportive role. As recent social psychological research has shown, people derive their self-identity from the groups to which they belong, and may regularly alternate among different salient identities. Thus, a working mother might think of herself primarily as a mother when in the company of her children, but see herself primarily as a professional while at work. The list of potential identities is extensive, with some identities (e.g.,

5
Behavioral Economics & Poverty "mother") likely to conjure up strikingly different values and ideals from others (e.g., "welfare participant"). In promoting banking, as well as other social programs, particular identities may be triggered that generate greater interest and willingness than others.
Policies that encourage the take-up of bank accounts may naturally lead to improved savings and budgeting in general. For one, the opening of a checking account can provide access to services such as automatic electronic payments that may help reduce the occurrence of neglected bills.
Furthermore, as has been documented among the non-poor, an increased utilization of saving accounts, partially as commitment devices, can increase savings. In open surveys we and others have conducted, many lower-income households report using otherwise very expensive lay-away schemes as a way to commit to saving for a specific goal. Survey research on IDA plans, which offer lower-income people matching funds for savings towards a productive asset, point towards a similar desire for commitment devices in this population, where a large majority of participants report favoring the strict withdrawal rules that characterize most current IDA plans. Bank accounts that offer concrete targets with mild commitment (such as a penalty for early withdrawal) would be much preferable to expensive lay-away schemes. Also, rather than abstract accounts, banks or community groups could try to promote the formation of "dedicated accounts:" a "fridge account," an "education account," or a "car account." Such labeling could be enticing and serve as a reminder of what is being saved for.
Finally, defaults may also help stimulate higher savings. Studies of middle-class savings behavior (Benartzi and Thaler, forthcoming) suggest that savings works best as a default, such as in 401(k)s where cash is automatically deposited into savings. IDA plans offer an easy opportunity to incorporate such default in the saving decision of the poor, for example through automatic deductions from paychecks. In fact, a program analogous to the "Save More Tomorrow" plan 6 Behavioral Economics & Poverty which allows people to choose their deduction levels for future paychecks, could be implemented within the IDA plans.

III. Social Programs
The poor have access to a myriad of transfer programs, which nevertheless show a remarkably low take-up rate. Again, economists' answer to this puzzle has been to look for large economic costs that might enter into the cost-benefit analyses of the poor when they decide not to participate. One oft-cited big cost is the "stigma" attached to such programs.
Other factors could also help explain low program take-up. First, various "small" hassles that can dissuade action appear especially salient in this context. A recent comprehensive study of food stamp applications found dramatic hassle costs. State applications reach up 36 pages and often include incomprehensible questions. The application process often cues negative identities and can induce guilt and alienation. People are finger-printed (to verify that they are not doubledipping in other locations), they encounter perjury threats, they undergo home visits to verify they are "really poor," and they are often condescended to. Such treatment is likely to reinforce the alienation and hopelessness that often discourage this population. As discussed earlier, such hassle factors may appear negligible in a standard cost-benefit analysis, but are the kind of barriers whose removal may open channels for desirable behaviors.
Finally, just as people procrastinate on medical check-ups or signing up for 401(k)s, the poor may procrastinate in signing up for welfare programs. This is likely exacerbated by some of the factors discussed above, and by the knowledge that even if they present themselves at the welfare office today, chances are they will not get "all signed up" today. The apparent cost of procrastination may also appear lower if, for a person not currently enrolled, non-participation is viewed as a 7 Behavioral Economics & Poverty foregone gain rather than a loss. Finally, procrastination may be enhanced by wishful thinking. If people believe they will soon get out of poverty or get a job, not applying for the program could be perceived as bearing a low cost since it will soon no longer be needed.

Policy Implications
What policies could a benevolent, behaviorally-minded policy maker explore to encourage takeup? First, facilitate the process. Reforms to improve the transparency of eligibility rules and userfriendliness of forms should be considered. In fact, a unique but simple eligibility form for all programs (as is already done in some states) may be especially helpful. Also, pre-filled forms (where items unlikely to change are automatically filled) could speed up and demystify the recertification process. This is often used in the private sector, for example, in the making of health plan choices.
Second, welfare programs may benefit from triggering more positive attitudes. For example, evidence suggests that trust may help people feel worthier, and typically encourages reciprocity.
Instead, current programs often present an adversarial tone, accompanied by monitoring concerns, and low expectations. Although money is being transferred, no goodwill is earned. Changes that earn goodwill have the potential to affect not only take-up but also compliance with other aspects of program participation, such as job search, attendance, and further referrals.
This discussion also touches on the decentralization of program administration. The federal government has attempted to decentralize many aspects of welfare programs while maintaining minimum benefit level rules intended to safeguard the system from a "race to the bottom." But alongside minimum benefit rules, there ought to be maximum "hassle" rules. For, with excessive 8 Behavioral Economics & Poverty hassle benefits are often lost. The decentralization process needs to be a guarantee that state and local governments will not dissuade take-up (and save money) through the use of various barriers.

IV. Conclusion
Standard economic/policy thinking attribute to people preferences and motivations that they often lack, and ignore psychological factors that can be highly consequential. Deterrence, for example, plays a key role in the legal determination of punishment but appears relatively ineffective because those who violate the law often tend not to engage in the presumed cost-benefit analyses.
Similarly, policies geared towards the poor are often driven by normative assumptions, rather than empirical facts, in ways that may miss the intended beneficiaries. Standard thinking naturally assume that big effects are due to big causes, and thus merit major intervention. If the poor are deeply hurt by their failure to have a bank account, then there must be compelling reasons for that failure. Behavioral research, on the other hand, has shown that highly consequential behaviors often are triggered by what are deemed to be minor causes.
More empirical research, we believe, should be directed towards testing the effectiveness of behaviorally motivated anti-poverty policies, such as the ones discussed above. The good news might be that simple and inexpensive policies have substantial impact. The cautionary news is that policy makers may need to attend to nuances they often are not trained to attend to: subtle distinctions that from a normative perspective may seem immaterial can have large implications for a policy's eventual success.