11
Findings Across the Policy Domains
The committee’s review of research from the six selected domains allowed us to look for themes, commonalities, lessons, and unexpected connections in the evidence. As our charge directed, we hoped to learn about evidence that certain intervention strategies or approaches are effective across multiple contexts, domains, or circumstances, as well as what might be learned from approaches that so far have not been widely effective. We looked for evidence about how the core principles described in Chapter 3 affect decision making in the context of specific real-world applications and how the intervention strategies described in Chapter 4 operate in varied domains, contexts, and populations.
The domains we selected cover important public policy areas, but these are by no means the only domains in which behavioral economics interventions have been studied. Even for the areas discussed in this report we did not have sufficient resources for a comprehensive review: rather, we undertook a broad overview of what has been learned. Other domains, such as voting behavior and charitable giving, have been studied in similar ways in the United States, and evidence from other countries—work on compliance with tax policies and regulations in the United Kingdom is an example—would also contribute valuable evidence. Nevertheless, review of these six domains provided evidence about application of behavioral economics ideas across quite varied settings.
FINDINGS
Our investigation revealed clear specific findings as well as several consistent themes, both of which highlighted the need for further research. It is important to note that some of the domains we examined have been the subject of more thorough research than others. While we had no systematic way to compare coverage, it was clear that, for example, a much larger volume of research has been produced in the domain of health than in any others we studied. Table 11-1 provides an overview of the findings from each of the domains for which the strongest evidence is available.
Overall, behavioral economics research has produced significant and growing evidence in five of the domains; the domain of criminal justice has not yet been extensively studied. The strength and volume of the evidence—and the degree to which it has been generalized, replicated, and applied at a broad scale—varies. The importance of the five core principles is demonstrated by this work, although not all principles have been used in interventions in all the domains. For example, in the criminal justice domain, the work has not included interventions to explicitly address or appeal to social norms. In the domains of social safety net benefits and retirement benefits, reference dependence and framing have not been extensively addressed. These gaps may reflect omissions in our necessarily brief overview of the work in each domain; they may be the result of publication bias (i.e., studies with null findings have not been published); or they may be an indication that some of the principles may be more or less relevant in different contexts. More important is that this body of work has confirmed in real-life policy settings that people’s actions and decisions are strongly affected by behavioral factors discussed in Chapter 3. This result may not be surprising, but it is a strong and necessary foundation for further progress in designing interventions that address the complex factors involved in human behavior.
Looking at interventions, four strategies have strong empirical evidence of effectiveness: defaults, simplification or removal of hassle factors, immediacy, and framing of choice sets. Two other strategies have shown mixed or less robust evidence: behaviorally informed incentives and implementation prompts.
The interventions that show the strongest effects all have fairly precisely targeted specific behavioral issues that are relevant to the problem the intervention addresses. The intervention strategies for which the evidence is mixed highlight the importance of applying human-centered design and related approaches to the selection and refinement of behavioral interventions in specific contexts discussed in Chapter 4. That is, these interventions appear to show effectiveness when tailored precisely to specific circumstances and populations: this effectiveness is clearly shown in the studies discussed in Chapters 5–10. Thus, evaluations of behavioral interventions
TABLE 11-1 Overview of Findings on Behavioral Economics from Six Domains
| Domain | Primary Findings |
|---|---|
| Health | Medication adherence: structural approaches to lowering barriers (such as forgetfulness, inattention, and inaccurate or incomplete information) are the most promising interventions. Physical activity: behaviorally informed incentives using loss framing have shown promise for increasing physical activity. Colorectal cancer screening: default or opt-out interventions are the most successful among interventions tried for increasing rates of screening. HIV prevention and adherence to treatment: behaviorally informed financial incentives, promotion of testing and treatment services, and provision of information about HIV risk and treatment effectiveness are all promising interventions for increasing HIV prevention and adherence to treatment. Vaccination: three strategies show the most promising behavioral evidence for increasing vaccination rates—making information about vaccines more salient, offering incentives, and changing defaults. Medical provider behavior: changing default settings and providing comparative information on the performance of peers are the most promising interventions to influence provider behavior to be more in line with guidelines. |
| Retirement Savings | Strong influence of core principles: in making decisions about retirement saving benefits, people are strongly influenced by the core principles—limited attention and cognition; incorrect beliefs; present bias; and, to a slightly lesser degree, reference dependence. Use of default designs by employers and firms: default interventions have a major impact on saving for retirement, with little evidence of significant downsides for employees. These positive effects have been demonstrated across a wide variety of settings, making them one of the behavioral designs with the largest generalizability. Other characteristics of retirement plans, including framing, wording, and visuals, have also been shown to affect choice of plan and have the potential to at least modestly increase retirement assets. |
| Social Safety Net Benefits | The evidence on the effectiveness of low-cost nudges to encourage participation in social safety net programs is mixed, with some interventions showing modest effects on take-up but others showing no significant effect or any effect at all unless the study population includes households that have previously participated in the program. A few more costly large-scale interventions, particularly those implemented by the U.S. Department of Agriculture for the Supplemental Nutrition Assistance Program, have shown positive effects on take-up by providing additional information and by simplifying application forms and otherwise reducing the administrative burden. |
| Domain | Primary Findings |
|---|---|
| Climate Change | Providing information to consumers to address present bias and limited attention and cognition and using nudges related to social norms and preferences have been shown to encourage energy conservation, climate-friendly transportation decisions, and engagement in land use conservation programs if carefully targeted to the concerns of the target population. |
| Education | Parent–child interactions: the use of low-touch nudges, such as goal setting, reward programs, text reminders, and efforts to increase parents’ engagement with their young children to improve literacy, can directly address issues with limited attention and present bias to increase the effectiveness of those interactions. Postsecondary education: several interventions have increased college application and applications for financial aid—reducing the administrative burden of the applications; providing support to students for these tasks; and setting up default choices for high school students, such as automatic registration for college admissions tests. Teacher performance: loss-frame incentives for teacher compensation could be a promising avenue to improve teacher performance, although the research is in its infancy. |
| Criminal Justice | Behavioral factors influence all points in the criminal justice system: determinants of criminal behavior, policing practices, court proceedings, judicial decision making, and incarceration. Although the evidence is limited, enough research has been conducted to suggest considerable potential for their effectiveness. |
are most valuable when they incorporate rigorous examination of specific group treatment effects.
More specifically, several findings stand out:
- Retirement savings: The strongest result in our review of the six domains is that making retirement savings a default choice for employees is consistently effective. A substantial body of research has confirmed this finding and tested variations in how it can be done. That evidence has been so strong that it has influenced federal legislation.
- Health: Evidence demonstrates the value of modest, low-cost interventions that target very specific challenges. The evidence also shows the cumulative value of small-scale, low-cost interventions.
- Climate change: Across a range of efforts, the evidence demonstrates the high value of targeting specific concerns, as well as the cumulative value of multiple small-scale, low-cost interventions.
- Social safety net programs: Broad-scale interventions to reduce the administrative burden and better reach the neediest populations who can benefit from social safety net programs can be effective when carefully targeted.
It is important to note that most of the available evidence that demonstrates positive effects is for interventions that bring about short-term or one-time effects. Although some one-time changes can have lasting effects, ways of bringing about persistent or sustained change have not often been examined; achieving longer-lasting effects is likely to prove much more challenging. It is difficult to change habits with a single light-touch or nudge intervention; multiple, repeated interventions are likely necessary. It is also challenging to change behaviors when the costs or outcomes of a decision are removed from the point at which it is made. For example, in the context of daily decisions about energy usage, the effect on a person’s electric bill—or on climate change—is quite far removed from an instance of forgetting to turn off a light. In this case, making the desired behavior change salient and making rewards or benefits of the behavior change immediate are promising strategies, although they may require repeated interventions.
In general, much more is known about promising interventions than about ways to translate, adapt, and implement them in varied contexts and at a broad scale. Many individual studies of interventions have been carried out across the six domains, but there has been far less work to follow up promising results with replication studies and with rigorous research on implementation and scaling (discussed in Chapter 12). Implementation and scaling studies are a crucial next step for identifying the intervention strategies that successfully scale and the contextual, institutional, or policy factors that predict successful implementation.
At the same time, methodological innovations are needed to uncover ways to tailor interventions to their targets and develop more robust and systematic approaches to understanding behavioral barriers—both of which have important implications for intervention design. Continued basic science, such as controlled laboratory studies that identify a promising mechanism of behavior change, is an important and needed input to the intervention design process.
Finally, it is worth noting that although there is a substantial body of work to consider, there are stones that remain unturned: it is important to distinguish between having evidence of no effects from an intervention and not having evidence of effects one way or the other because the intervention has not been adequately tested. A complicating factor is that research that demonstrates no effects is less likely to be published and accessible (discussed in Chapter 12). One possible factor in the variation in research attention across the six domains may be professional incentives. For example, funders may be more willing to support research in the health domain than in other domains, particularly replication and generalizability studies, and there are more journals and other media for studies in this domain.
In sum, a striking takeaway from our review of the impact of behavioral economics on policy is that real policy impact will happen only when
interventions that show effectiveness in a laboratory setting are translated for real-world application, carefully designed for the intended context, accepted by policy makers, and implemented at scale (discussed in Chapter 13). An excellent example of interventions for which such work has been done is the work on increasing usage of Supplemental Nutrition Assistance Program benefits (see Chapter 7); more such work is needed.
CONCLUSIONS
The committee’s findings from across the six domains led the committee to several conclusions about policy design.
Conclusion 11-1: Core principles of behavioral economics have been tested repeatedly across six domains—health, retirement benefits, social safety net benefits, climate change, education, and (to a lesser extent) criminal justice—and the evidence for their importance and value in the design of policy interventions is well established.
Occasionally, it is possible to achieve headline effects—those that are large and enduring, such as the impact of default settings on retirement savings. However, such results are rare: more often, effect sizes are smaller, but it is important to keep this point in context. The objectives policy makers pursue can be both grand (improving the nation’s cardiovascular health, improving student achievement and attainment, mitigating the effects of climate change) and precise (encouraging customers to conserve energy, boosting 4th-grade reading scores). Interventions that are effective, albeit on a modest scale, are a vital component in an array of strategies, particularly if they are low in cost. Indeed, as the research on interventions related to climate change demonstrates, pursuing grand policy objectives requires a variety of approaches. Small interventions are no substitute for structural changes or strong economic incentives, but that is no reason to ignore the accumulated benefits that small, low-cost interventions can contribute in addressing major problems.
Moreover, it is clear that knowledge about behavior is indispensable to the design of effective policies in any domain in which the objective is modifying human behavior and decision making.
Conclusion 11-2: The strong evidence that the five core principles play a significant role in human decisions means that the design of effective policies intended to influence decision making requires careful, expert attention to behavioral factors.
Experts trained in behaviorally based policy development are needed for the development of effective policies in part because there are no easy shortcuts in the design and application of behaviorally informed policies (discussed in Chapter 14).
Conclusion 11-3: There is clear and strong evidence that specific interventions based on behavioral economics principles have been effective at changing targeted behaviors, but matching the tool to the challenge, the circumstances, and the target population is critical to success and requires careful attention. To achieve policy goals, interventions that show effectiveness in laboratory or highly controlled settings need to be translated for real-world application, carefully adapted for the intended context, accepted by policy makers, and implemented at scale.
Part III discusses the implications of these conclusions for training, funding, policy making, and further research.
This page intentionally left blank.
