Welcome to CFED's behavioral economics blog! Emerging evidence from the behavioral sciences and behavioral economics offers us a richer and more nuanced understanding of human behavior and decision-making. We believe that taking these insights out of lab and into the real world may uncover promising and powerful ways to promote economic opportunity, asset building and financial security. This blog reports on the latest in applied behavioral research, program and product design, and provides a forum for asset building practitioners and researchers to share their experiences applying behavioral economics to their work with their peers.
Designing for Scarcity: A Behavioral Approach to Two-Generation Strategies with Poverty Interrupted
By ideas42 on 09/11/2014 @ 09:50 AM
It’s easy to agree that poverty is a problem, but explaining its causes and prescribing solutions is a far more difficult and complex task. What if children born into poverty were as likely to succeed as their wealthier peers? What if intergenerational poverty were not an unavoidable and enduring evil, but entirely tractable?
These are the challenging questions that ideas42, a unique social enterprise that uses insights from behavioral economics to address tough social problems, is seeking to answer with a new initiative called Poverty Interrupted. By combining effective program design with behavioral insights drawn from the cognitive effects of scarcity, Poverty Interrupted aims to end intergenerational poverty for families with young children. Currently in the research and design phase, Poverty Interrupted will identify behavioral barriers that are preventing better outcomes for people in poverty. Once identified, the findings of this research will be used to develop interventions designed to mitigate those obstacles.
Despite the efforts and expenditures of countless public and private organizations each year, poverty remains a severe problem in the United States. As of 2012, the relative poverty rate in America is 17.4%—well above the OECD average of 11.1% (OECD 2014). Beyond the inherent injustice of that statistic, it is also extremely expensive to society for low-income families to fail. The national cost of poverty—including both public spending and lost earnings—is estimated to be $500 billion per year, or nearly 4% of GDP (Holzer et al. 2007). Children are especially vulnerable; in 2012 more than 21% of Americans under 18 were living in poverty (OECD 2014). Most concerning of all, if current trends continue, many of these children will grow up to head low-income households of their own, and continue a cycle of intergenerational poverty.
Despite these disheartening facts and figures, ideas42 believes that emerging insights from behavioral economics offer hope. Central to the Poverty Interrupted initiative is the recognition that poverty presents inherent cognitive challenges that must be addressed before economic mobility can be achieved. Members of the project team will present initial findings and promising solutions at CFED’s 2014 Asset Learning Conference in an interactive round table session titled Designing for Scarcity: A Behavioral Approach to Two-Generation Strategies on Wednesday, September 17 at 4pm.
The session will be framed by the concept of scarcity – exploring how a shortage of money, time and other important resources taxes our ability to make decisions, pay attention and exert self-control. This cognitive effect has significant implications for programs and policies that serve and support low-income families whose lives are marked by chronic scarcity. The Poverty Interrupted team will discuss breakthrough research on scarcity in behavioral economics and will allow participants a chance to identify drivers of scarcity in their own organizations and communities.
Join the session to learn more about Poverty Interrupted and the unique opportunity to combine ideas42’s expertise in behaviorally-informed design with the existing knowledge in the poverty-fighting space to break the cycle of poverty for millions of people.
Interview: CFED Interns Participate in Poverty Simulation
By Paul Day on 07/25/2014 @ 03:45 PM
Last week, CFED interns Anne Guthrie, Valerie Marshall and Alexander Scarlis played the role of service providers in a simulation for nearly sixty Capitol Hill legislators, staff and interns to experience what it’s like to be in poverty. We sat down with Val and Alex to discuss their thoughts about poverty simulations and how they can be a valuable tool to change perceptions about the working poor.
Q: What did you hope to learn from this? Were your expectations fulfilled?
Alex: I don’t know what it feels like to be in poverty. I was very curious to see how I felt being in this simulation. I was curious to see how members of Congress as their staff reacted and what it felt like for them to be in poverty. It was chaotic and stressful for everyone involved to a certain extent. But I think they found the right balance.
Val: I had never been to a poverty simulation before so I had no reference point. I had no idea what they were going to do. I was looking to get the experience to see what they were going to say about poverty. You had to confront this feeling of helpless on a whole new level. They emphasized key things most people don’t even think about, like transportation. For example, to go from table to table you had to pick up a transportation ticket. So as these public transportation systems start raising their costs, obviously it’s going to have a big impact.
Q: From what you know about poverty in a policy sense, how accurate was the simulation?
Alex: Well I think they nailed all of the different programs that poor people have available but they are in all of these different places so you had to run around trying to gather all of this documentation and maintain a semblance of a healthy family life which means grocery stores, getting your car fixed and getting to work on time. It’s everything we all do plus trying to make sure your benefits stay intact, which is next to impossible.
Val: I thought from the participant side it was really realistic. It was really overwhelming in the sense that you got that constant stress that I think a lot of families living in poverty experience.
Q: Was the stress level of the simulation intentional or accidental by how it was organized?
Val: I think it was designed to mimic the environmental stressors. The families they designed were actually real cases they had from some Department of Health and Human Services database so it really was realistic. In the simulation I had two children plus a younger sister who was under eighteen that I had to take care of and I had to negotiate those responsibilities in addition to my own education and rent. It was definitely a good reminder of why we’re doing the work that we’re doing.
Q: So there were members of Congress there participating as individuals and families in poverty. What were some of their reactions?
Alex: Everyone was trying to keep track of their money and make sure they have everything in order because everything was so chaotic around. It was too bad that more members of Congress weren’t participating, but it was actually great a lot of interns and staff were participating because they’re often the ones writing up the legislation, doing the research or advising the representative.
Val: A lot of members of Congress spoke before and during the event. There was a huge emphasis during the event about the working poor and just because you live in poverty doesn’t mean you’re lazy, dispelling these old stereotypes about who’s on welfare vs. those who aren’t. But there were still some reminders of that archaic way of thinking. There are still areas for progress.
Q: Would it make sense to branch out and do these kinds of simulations in other cities?
Val: I think it would be a good idea. Most politicians come from an affluent background and having this sort of education would counter other things they’ve learned. It’s one thing to hear about an experience from a constituent and another to have the experience yourself.
Alex: If politicians are responsive to their constituents, then theoretically we need to change the perceptions of the constituents. From that perspective, it would make sense to have a lot of these simulations around the country that are accessible to the general public. When you have a mindset shift among people then you’d think that would trickle up to Congress because at the end of the day, they have to be responsive.
Q: Can you talk about where poverty simulations fit in the context of CFED’s work? What insights has this experience given you?
Alex: To have these members of Congress and staffers participate, hopefully when they read our work about asset limits and the struggles people face that translates and something connects. It would be great if this was more widespread in the policy world; otherwise it’s just statistics.
Val: A lot of the research we’ve been looking at in the applied research department has been looking at the role of financial literacy and applying those skills and changing behaviors across the lifespan. It really shows the value of having financial knowledge and knowing which questions to ask, who you need to speak to, all that stuff underscores what we’ve been looking at.
For more perspectives on the recent Capitol Hill poverty simulation, read this latest Washington Post story.
You Might Like
BeFi 2014 Highlights: Challenges to Improving Financial Decision-Making and Behavior
By Pamela Chan, Emily Hoagland and Emory Nelms on 06/19/2014 @ 03:45 PM
On May 30, we participated in RAND’s annual Behavioral Finance forum (BeFi), a day-long conference which brings together research from behavioral science, cognitive psychology, economics and consumer finance. This year's conference presented research on retirement security, financial inclusion, consumer financial protection and financial advice. In particular, the presentations focused on challenges that policy makers and financial service providers face in improving consumers’ financial decision-making and behaviors. Here are a few key takeaways:
- Long-term engagement is difficult to maintain. (Presentation, Wendell). Auto-enrollment and default options have been among the most successful and widely incorporated behavioral insights from past research. However, these interventions only involve one discrete action or decision. What happens when you want a person to complete a series of decisions and actions over the longer-term? Experience tells us that most people tend to be engaged at first but lose interest as time goes on. In an attempted to counter this tendency, HelloWallet, a financial management software company, tested strategies for maintaining long-term engagement in their services. While they’ve identified some promising strategies (notably, asking someone to do less and using mobile technology), they still struggle to maintain high levels of engagement over the long-term, and that’s a challenge for the behavioral finance field to continue grappling with.
- The impact of environmental context on a person’s financial decision-making changes when operating under scarcity. (Presentation, Shah). When people are presented with everyday financial decisions, they often rely on mental accounting. Mental accounting involves drawing inferences from the surrounding context and weighing our different options. In this way, how we assess value is relative to our context. For example, someone sitting on the sand at a beach resort may be willing to buy a drink for $9 even though they’d never pay that much for the same drink at a convenience store. In these situations, behavior changes as the contextual cues changes—the posh-ness of the resort makes it justifiable to pay $9. What’s interesting is the power of environmental context seems to lessen when a person is operating under scarcity (e.g., has less money). In situations of scarcity, people may actually make more consistent financial decisions because their worries of trade-offs are more salient and can override environmental cues that would otherwise be influential. In the beach example, a person operating under scarcity wouldn’t pay $9 for a drink at a resort that they can get for $2 at the convenience store. This insight serves as a reminder to be careful about making assumptions on the strength of environmental context on consumers. It also highlights the complexity of financial decision-making and illuminates why there are rarely easy solutions for policy makers and financial services provider seeking to influence consumer behavior.
- People don’t know who to trust…or potentially trust the wrong people. (Presentation, Huang and Agnew). Knowing where to find reliable and relevant financial information about financial products and services is a major barrier that we face when making a financial decision. Although there are many sources of financial information, about half of those who seek out financial advice said they didn’t know who they could trust to provide it. Even more, people tend to be bad at judging financial advice even when they do seek it out. For example research shows that if advisors give good advice on easy topics followed by bad advice on harder topics, people will still continue to trust them. In general, people also seem more likely to trust younger advisors and those with some form of certification; however, people often cannot tell the difference between real credentials and those the experimenters made up. This insight reinforces the need for products or services that help consumers identify quality sources of financial advice.
Beyond these panels, there was also a wealth of insightful topics on financial literacy both nationally and internationally, how visualization encourages learning, and how the federal government and private financial companies are using behavioral finance to build financial capability and stability. Videos of the presentations will be posted on RAND’s BeFi website.
You Might Also Like
Testing Text Messages to Improve Financial Behavior
By Pamela Chan on 05/21/2014 @ 09:30 AM
How effective would you like your mobile banking alerts to be? Last month, The New Yorker featured a snarky mobile banking program in a humor column by Kelly Stout. Personified through a text conversation with the author, “Mobile Banking” started with helpful reminders about her balance and spending goals. However, as the author fails to live up to the expectation of Mobile Banking in her spending behavior, Mobile Banking devolves into a frenemy, sending judgmental reminders such as:
“Your checking acct ending in 6885 has a balance of $3.03, which is below your $50.00 limit in your Alerts setting. What is going on with you? I take it your $.99 transaction with EZ BUDGET didn’t work out.”
Fortunately for us, we’re far from a world where we have our own versions of this Mobile Banking character in our pockets. There’s been a proliferation of mobile banking and financial management apps that offer frequent reminders to help consumers stay on track with their finances. However, designing effective reminders that prompt people to take actions in managing their finances is a fairly difficult task.
We’ve talked about this in a previous blog post, “Read this Now! The Art & Science of Reminders.” Effective reminders need to be salient enough to grab one’s attention, specific about the action to be made, and timed within a window of time that is both close enough to the point of action and far enough to allow sufficient time for the action. So, how is one to know how to shape reminders?
Researchers at Innovation for Poverty Action are working how text messages affect financial behaviors through a Message Replication Program. Specifically, they are conducting a series of experiments to see how the timing and content of different text messages can shift savings and credit payment behaviors. They already ran pilots in Peru, Bolivia and the Philippines and are now seeking partners to conduct tests in the US. If you’ve got a savings or credit program with over 20,000 clients, you may be able to take part in this pilot (download details here).
While we await the results and recommendations from the Message Replication Program, a good way to figure out how to design effective reminders for your clients is to set up small experiments at your own organization. I recently learned of HealthCrowd, an engagement platform for healthcare providers that serve patients with Medicaid. HealthCrowd utilizes text messages to encourage positive patient actions, such as showing up for a preventative care appointment. According to their CEO, Neng Bing, HealthCrowd’s messaging platform works because they run experiments among existing customers to test whether or not certain messages or the timing of the messages prompt patients to take a desired action.
At CFED, we worked with ideas42 last year and set up experiments to test program features in the BETA Project. The tests conducted last year showed that community-based nonprofits could run experiments and try out new program features. As none of our sites focused on testing the messages and timing of reminders, we’d love to know if any of you are testing out text messages in your organizations. Please share what you’re doing in the comments below. If we get a good response, we can compile some highlights from the field on the use of mobile text messages and apps to shift financial behavior for a future blog post!
Findings and Implications from Testing in the BETA Project
In the social sector, it's not always possible to test the impact of a designed solution. This often leaves us wondering whether resources are spent in ways that have real effects on welfare. The BETA Project was a unique opportunity to design behavioral solutions for three partner sites and test each solution to see whether or not it had an impact on the respective program.
What we found is encouraging. As discussed in our brief from the project, we discovered that small changes to program design, inspired by behavioral economics, can have a real impact on program effectiveness:
- At Accion, we saw that simple reminders reduced Non-Sufficient Funds fees and increased on-time payments—especially among the most vulnerable borrowers. The impact on “risky” borrowers is particularly significant in light of Accion’s mission to reach this underserved population, but it has larger implications as well. Access to credit is important; it can help vulnerable borrowers grow a business, stabilize the amount of money available each month and deal with emergencies. Improving repayment behaviors among this borrower segment could reduce costs and help justify expanding services to this population.
- At Cleveland Housing Network, we saw effects of the raffle on rent payment throughout the month, even after late fees were applied, suggesting a more profound effect on behavior than mere financial motivation. The fact that these effects persisted throughout the month, even after late fees were applied, suggests a more profound effect on behavior than mere financial motivation. Countless programs and products have payment (or repayment) problems in the asset-building field. If we find that this effect scales, it has the potential for massive impact.
- At Neighborhood Trust, we saw that simple plan-making facilitates action, and observed a promising increase in people’s utilization of accounts. While our results at Neighborhood Trust were indicative, and not conclusive due to small sample size, similar efforts to prompt action at Grameen Bank resulted in similar-sized effects on savings account usage but at a statistically significant level. Sites providing financial education should marry their education efforts with concrete steps to assist people in taking actions (such as through simple plan-making) that move them towards a stronger financial position.
These effects are not just statistics. There are real people who benefitted from the BETA project. Moreover, the solutions were relatively inexpensive, with direct costs of less than $5,000 at each organization. The primary cost was the investment in staff time to work with the BETA Project team to diagnose the problem and design solutions. All three sites have decided to continue to use some form of the tested solutions in the future. These findings speak volumes about ways other asset-building programs can make small adjustments to deal with these common problems and have a real impact. In thinking about why the reminders, raffles and simple plan-making activities had a real impact for our partner sites, we were able to uncover three important lessons for program and product design in the asset-building field:
- Be preemptive. Asset-building organizations often start engaging with clients only after a problem, such as a missed payment, arises. While in-person interactions with clients can be expensive, there are many early opportunities for engagement through other communication channels with clients.
- Incentives “buy” attention. Incentives can be a powerful tool for behavior change, but not always because of their financial benefit. Sometimes incentives are helpful because they capture people’s attention. They signal that doing the incentivized activity is important and perhaps encourage someone to prioritize that activity.
- Facilitate action along with information. Information alone rarely leads to behavior change. Simple plan-making activities help to guide clients into action (e.g., active utilization of accounts). Organizations that seek behavior change should provide assistance through action-oriented activities like simple plan-making and direct access to relevant products and services to help clients follow through on their goals.
For more on these insights from the tests at our partner sites and tips for how organizations can design behavior-changing program solutions, please check out the Small Changes, Real Impact brief and report available on the BETA Project Publications and Events page.
Testing in the Social Sector – the Beta Project
What organization doesn’t want to describe itself as “outcome oriented” and “data driven”? These two buzz phrases highlight a growing interest in the social sector for measuring and tracking concrete outputs in order to demonstrate organizational impact.
Beyond the usual metrics, however, is the need to measure the impact of specific program changes or initiatives. One useful tool for testing the effects of a change is the randomized controlled trial (RCT), the “gold standard” for evidence across many domains. In our last blog post, we discussed how RCTs are useful because they cut through the bias introduced by our motivations and allow us to assess the impact of a specific idea. Many good ideas just don’t have an impact in the real world, and testing is important to make sure that resources are spent in ways that have real effects on welfare. In the BETA Project, we conducted RCTs at two of our partner sites, Accion Texas and Cleveland Housing Network.
Change is Hard
In order to learn about the best ways to effect change, we need to test specific programs or interventions. But there are cases all over the social impact world where programs aren’t evaluated, so we have no idea whether they had any impact at all. In cases where rigorous testing has been done, the results have often suggested that the conventional wisdom about “what works” is off the mark. Sociologist Peter Rossi summarized the problem with his “Iron Law of Evaluation”: The expected value of any net impact assessment of any large scale social program is zero. In other words, we could save a lot of effort and have the same effect by never trying out any programs at all!
While the Iron Law is perhaps a dramatization of the difficulty of effecting large-scale change, it encourages us to be careful with our time, our money, and our optimism. Three recent RCTs have produced results that push back on the conventional wisdom of the social impact world.
The Miracle of Microfinance
In 2006, microfinance was on the top of the world. Dr. Muhammed Yunus and the Grameen Bank received the Nobel Prize – not for Economics, but for Peace. In the prize announcement, the Nobel committee stated that “Yunus and the Grameen Bank have shown that even the poorest of the poor can work to bring about their own development.”
Microcredit makes a lot of sense. Many economists believe that lack of access to credit is one of the most important barriers keeping people impoverished. But some economists were skeptical. While microcredit borrowers saw huge gains compared to non-borrowers, it was possible that previous studies were simply catching the effect of being entrepreneurial. The poor with an entrepreneurial spirit were both more likely to succeed and more likely to get a loan.
In 2009, these economists put microfinance to the test – and found that, if microcredit was distributed in a randomized trial, there were no effects on poverty a year and a half later. Of course, this does not mean that microfinance did not work. Many people who received microcredit seem to have been investing in their business. But realizing that investment may take them much longer than we once expected.
Location, Location, Location
How much of poverty is driven by “neighborhood effects? The sociologist Julius Wilson has theorized that harmful effects of neighborhoods could be responsible for high dropout rates in school and in the labor market. For example, in high poverty areas it might be harder to find peers and role models that would encourage study. In 1994 the department of Housing and Urban Development tested this theory with the “Moving to Opportunity” experiment. The experiment gave some families vouchers to help move them from high poverty areas to low poverty areas, with the hope that such a move would substantially improve outcomes.
But the Moving to Opportunity findings were mixed. While the effects on schooling and jobs were lower than expected, there were surprisingly strong effects in health. Obesity rates dropped by 40%. Neighborhoods effects are important – but not in the ways we initially thought.
The Effects of Financial Literacy
In the aftermath of the Great Recession, many have called for personal financial education to become a standard part of the high school curriculum. Surveys show that individuals who know more about personal finance are also more likely to build their savings and limit their debt, allowing them to better weather financial storms.
But once again, randomized controlled trials tell a different story. Several studies have randomly assigned some individuals to receive financial education, while control individuals did not. These studies did not find any changes in savings or debt. This suggests that providing information alone did not put people on the path to financial security, and might encourage us to address other problems limiting financial success, such as limited self-control and attention.
Past Research Informs Future Design
The results above may seem surprising, but it’s not all bad news! Careful evaluation also lets us know when we’ve gotten it right, and can steer us towards new applications of proven solutions. For example, ideas42 has an ongoing project [M1] where we are working to redesign financial education curriculum using "rules of thumb" (a paper on this project will be released in early 2014). In the BETA project, our intervention designs were based on ideas that have worked in the past. For example:
- At Accion Texas, we reminded borrowers to be prepared to pay their monthly bill via text and email reminders. This approach that has been shown to be effective in helping the poor build assets by reminding them to save.
- In our work with the Cleveland Housing Network, we designed a raffle for borrowers who paid their rent by the 1st of the month. In the past, a raffle-based approach has been shown to help heart patients who are especially at risk of forgetting to take their medication.
- At Neighborhood Trust, we helped clients plan their account usage by prompting them to make a plan as to how they would do so. This intervention was inspired by research that shows that helping voters make a plan with information about where, when, and how they plan to vote makes them more likely to actually follow through.
Next BETA Project Post: Findings and Implications from the BETA Project
Our next post on the BETA Project will be final post based on the work completed in 2013 with partner sites Accion Texas, Cleveland Housing Network and Neighborhood Trust. This post and other helpful insights from the BETA Project are available on CFED’s Behavioral Economics blog and BETA Project website.
How to Know What You Know – the BETA project
How do we understand the world around us? As individuals, we have five senses that will reveal some truth about the world, and they serve us pretty well when we’re trying to survive. But how do you see an atom? Can you touch a burgeoning social movement? Can you hear the migratory behavior of the humpback whale? (One guesses that the answer is no.) To gain a deeper understanding of the world around us, we have to gather data that lie beyond what we can directly observe.
The Origins of Scientific Inquiry
In 1747, James Lind conducted what is widely considered to be the first medical clinical trial in his quest to treat scurvy. He gave 12 of his scurvy patients (“Their cases…as similar as I could have them”) differing treatments over the same time period and found that the ones who were fed a steady diet of oranges and limes improved rapidly. This comparison of a group that receives a special treatment (diet of oranges and limes) against a group that receives the standard diet hinted at the rough structure of what we today call a Randomized Controlled Trial (RCT).
The Modern Era
Since then, RCTs have become the “gold standard” for evidence across myriad domains. In the BETA Project, we conducted RCTs at two of our three partner sites, Accion Texas and Cleveland Housing Network.* The modern RCT is more rigorous than Lind’s design, but it is motivated by the same desire to evaluate the effects of trying something new. During an RCT, individuals in a population are randomly assigned to receive a treatment, and they are evaluated in comparison to another group in the same population who (randomly) did not receive the treatment.
In the academic world, doing an RCT is a common way to design a study, but more recently, social impact organizations and government agencies have increasingly begun turning to RCTs to evaluate their programs. The Institute of Education Sciences, an office of the Department of Education, has supported 175 RCTs to date in their effort to evaluate interventions in the school system. Their efforts are part of a broad push to formalize and quantify the effects of new initiatives in public policy.
Why Randomized Controlled Trials?
Why are randomized controlled trials so valuable for evaluating new ideas? If you ask a statistician, they’ll tell you that experimental designs allow us to isolate the effects of a given intervention. The idea of an “intervention” is a broad one – scurvy patients are given citrus; students are taught with a redesigned curriculum; cancer cells are treated with a new chemical agent. With a large enough sample, we can assume that individuals who got the treatment have the same innate characteristics and are affected by the same environments on average as the individuals who didn’t, which means that any differences in outcomes we observe come from the intervention applied.
If you ask a behavioral scientist, however, they’ll be happy to tell you why we need to systematically evaluate interventions for behavioral reasons. First, observers are subject to biases in their intuition and perception—rigorous evaluation provides solid evidence that cuts through these biases. Second, humans behave in ways that are hard to predict. Careful testing provides us with reliable evidence on human behavior, and this evidence allows us to feel confident that social impact interventions are having real effects.
Patterns in the Noise
The behavioral research suggests that people are exceptionally adept at seeing patterns in the noise, especially when those patterns happen to fit neatly with their motivations. In one study, participants were told that they would be asked to drink either some delicious orange juice or a much less appealing vegetable smoothie depending on what a computer display showed them. Some were told that they would drink the orange juice if they saw a letter, while others were told they would get the juice if they saw a number. However, every participant actually saw something like this:
Is this a letter or a number? It’s unclear. But participants who were told that a letter meant they would drink the OJ overwhelming saw a “B”, while those participants who were told that a letter meant they would drink the gross vegetable smoothie saw the number 13 instead. Their motivations fundamentally changed how they perceived the world around them.
This lab study echoes things we see every day – political junkies are sure that the media is biased against their candidate, the refs have made an obviously terrible penalty call when it’s against your favorite team, and your child’s art is objectively the most beautiful in the class. But all of these examples serve to highlight the importance of rigorous evaluation. Without testing, we’d be likely to perceive our interventions as effective in every case – an assumption that can be irresponsibly inaccurate.
*We were unable to randomize at our third partner site, Neighborhood Trust, due to sample size limitations.
Next BETA Project Post: Testing in the Social Sector
Randomized controlled trials help us determine what works, and figuring out what works is especially vital in the social sector. Our next post on the BETA Project will discuss using RCTs in the social sector. This post and other helpful insights from the BETA Project are available on CFED’s Behavioral Economics blog and BETA Project website.
BETA Project Wrap-up: Small Changes, Real Impact
In reflecting back, 2013 was an exciting year for CFED, ideas42 and the Citi Foundation! Through the BETA (Behavioral Economics Technical Assistance) Project, we worked with Accion Texas, the Cleveland Housing Network, and Neighborhood Trust Financial Partners to design and test new solutions for their programs using insights from behavioral economics. We defined the problems to be tackled with our partners and diagnosed why clients may be missing payments or underutilizing credit union accounts. We created new solutions such as raffles, reminders, and planning tools. We then tested the designed solutions to assess their effectiveness—and found that small changes within a program can have a real impact on outcomes for small business loan borrowers, lease purchase program residents and financial education clients.
On December 17, the BETA Project team published the results of these experiments. We also hosted a webinar, Small Changes, Real Impact: Applying Behavioral Economics in Asset Building Programs. During the webinar, we shared insights we gained through the project.
First, our findings affirmed several principles from behavioral economics. For example, our designed solutions seemed to have a greater impact on certain clients, many of whom were considered the most vulnerable. This parallels the recent book on Scarcity by Sendhil Mullainathan and Eldar Shafir. They suggest that living in conditions of poverty imposes a heavy cognitive burden - and that interventions that relieve that burden (like reminders) may be especially powerful.
Second, our partners uncovered many lessons about how practitioners can approach program design from a behavioral perspective. More often than not, asset-building organizations come to us with ideas about what they need to do to improve their programs. However, starting with a new solution without taking the time to better understand the nature of the problem at hand can cause issues down the road (e.g., wasted money on an ineffective solution). To prevent this, we recommend putting the problem before the solution. By taking time to refine the problem and generate theories about what’s going on, you can test whether or not your theories are valid.
Find out more about what we learned with the BETA Project partner sites, plus additional tips for incorporating behavioral economics principles into your own work. Links to our webinar, related brief and full report are available here.
Tomorrow: Small Changes, Real Impact
By Sean Luechtefeld on 12/16/2013 @ 12:00 PM
Applying Behavioral Economics in Asset-Building Programs
Tomorrow! | 3 - 4 pm EST
Last year, the BETA Project was launched to improve the effectiveness of products and services designed to help people bolster their financial security. In the 12 months since, we’ve worked closely with three organizations to understand a problem in their program, design solutions and test them.
Tomorrow, join CFED, ideas42 and the Citi Foundation as we discuss findings from this year-long project. During the webinar, speakers will report on the research conducted at BETA Project partner sites, explore the implications of applying insights from behavioral economics to asset-building program design and provide helpful tips on how to incorporate the behavioral perspective into your organization. More information on the project is available here.
- Daria Sheehan, Citi Foundation
- Katy Davis, ideas42
- Matthew Darling, ideas42
- Pamela Chan, CFED
Advanced registration is required for the event and space is limited. Click here to sign up.
If you have any questions, please don’t hesitate to contact Pamela Chan, Senior Research Manager, at email@example.com.
Michigan's Save to Win Program Demonstrates Successful Way to Encourage Savings
By Kori Hattemer on 10/10/2013 @ 12:00 PM
In 2009, the Doorways to Dreams (D2D) Fund partnered with eight Michigan credit unions to launch Save to Win ™ (STW), the first large-scale prize-linked savings product in the nation. Accountholders who save using the special balance building 12-month certificates of deposit (CDs) are entered into raffles for cash prizes with each deposit of $25 or more. Four years later, 58 credit unions in Michigan now participate in the program and 40,000 unique accountholders across the state have saved $72.2 million from 2009-2012.
D2D has been tracking the program in Michigan since its inception and recently published highlights from 2012, which emphasize the importance of developing innovative ways to encourage financially vulnerable households to save. Highlights from Year 4 of the program include:
- Accountholders appear to be developing long-term savings habits. Each year, accountholders are given the option to reopen or "rollover" their accounts, and a high percentage of accountholders rolled their accounts over from 2011 to 2012. Ninety-one percent of the accountholders who enrolled in 2009 and were still enrolled in 2011 once again rolled their accounts over in 2012. A high percentage of accountholders who signed up in 2010 (83%) and 2011 (77%) also rolled their accounts over from 2011 to 2012.
- Accountholders used their savings for a variety of purposes. While many accountholders rolled their accounts over in 2012, accountholders also used their savings to meet short-term needs. Account balances decreased in May and September, so accountholders may have used the funds to pay for summer child care or back to school costs. D2D also found that there is a cyclical dip in account balances during the rollover period each year, which may indicate that accountholders are making planned withdrawals between account years.
- STW continues to positively impact financially vulnerable accountholders. D2D defines financially vulnerable individuals as those who are single parents, asset poor, non-savers, or low-to-moderate income. In 2012, these accountholders had nearly identical rollover rates as their non-financially vulnerable counterparts, demonstrating the importance of STW in helping financially vulnerable individuals save.
As other organizations and financial institutions look for ways to empower financially vulnerable individuals to save, STW provides an innovative model for designing a savings product that is engaging and encourages savings habits. The success of STW in Michigan has motivated other states to launch similar programs, and D2D continues to advance prize-linked savings products as a strategy for helping low- and moderate-income individuals save.
To read more about the success of the STW program in Michigan in D2D's recent report, click here.
To learn more about D2D's pioneering prize-linked savings work, click here.
From “To Do” to “Done”: Simple Plan-Making Strategies
Take a look at your to-do list. Which task do you expect to complete first? Probably something fairly simple and concrete: “Buy groceries,” “Pay phone bill,” “Clean bathroom.”
Now, which task always gets pushed to the bottom of the list? Learning conversational Spanish would be fun and would make you a stronger job candidate, but as the day goes on, it’s easily put off for another day.
Despite fully intending to, we may have trouble accomplishing tasks—even very important ones—that we fully intend to do. The tasks we define for ourselves can be vague, contain multiple steps or involve obstacles that we just don’t want to think about. These are situations where a technique called simple plan-making may be useful.
Simple plan-making is a strategy to turn our intentions into action. In a study aiming to improve voter turnout at elections, eligible voters received a phone call asking them if they intended to vote. However, some voters were also asked: (1) when they would vote, (2) how they would get there and (3) where they would be beforehand. Adding these three plan-making questions more than doubled the effect on voter turnout. The act of creating a well-constructed plan helps us think through potential obstacles, plan how to overcome them, and remember to execute that plan at the right time.
In the BETA Project, we incorporated plan-making into our design at partner site Neighborhood Trust Financial Partners (Neighborhood Trust) to help clients follow through on their intentions to open and use credit union accounts. Below, we share some elements of our plan-making designs at Neighborhood Trust.
Why Did We Choose to Use Simple Plan-Making at Neighborhood Trust?
Through Neighborhood Trust’s financial education course, a significant number of participating clients open credit union accounts. Our behavioral diagnosis at Neighborhood Trust revealed that some clients open an account, intending to use it; however, they never take that first step towards actually using the account.
Neighborhood Trust appeared to be an especially promising candidate for plan-making tools based on some key characteristics:
- The intentions already exist. Making a plan won’t work if we don’t actually want to perform the action. At Neighborhood Trust, many clients open accounts during the class, suggesting that they intend to use those accounts…eventually.
- Using an account is complicated. The more vague or complex the steps, the more likely we are to procrastinate. Clients at Neighborhood Trust may not think through the large number of steps required to use their credit union account, from learning how to use a debit card to planning regular trips to the ATM. We saw an opportunity to break out each step to make them more concrete and achievable.
- Timing matters. In order to act on an intention, we must remember what we planned to do at the right time. For clients, fitting in a trip to the credit union amid a busy work schedule may be daunting or easy to forget. Creating a plan to go to the credit union could help clients set an explicit intention for completing the action and writing it down could help them to remember their intentions at the right time (“I’ll go to the ATM right before I shop at the grocery store…”).
- There are opportunities for commitment and enforcement. There are numerous forces pulling our attention away from our intended goals, and it takes a lot to keep ourselves on track. We all need ways to stay committed to completing our intended goals. Neighborhood Trust’s five-week financial education course presents an opportunity for clients to publicly commit to their plans during class discussion, and report back during the next class.
We often tell ourselves that the more important a task is, the more likely we are to complete it. However, because we assume that our strong intentions will be enough to carry us to completion, we fail to take the necessary steps towards completing our goal. Plan-making can be MOST effective (and usually most needed) when intentions are strong. It helps to remind ourselves of all the things that need to happen in pursuit of our goals and to make sure we take action to complete them all.
The BETA Project Design
We designed a set of plan-making activities for Neighborhood Trust clients to use during the five-week financial education course. The activities included bringing the necessary documents to open an account, locating the nearest ATM and making an initial savings deposit.
To make sure we were helping clients follow through on an existing intention, we instructed clients to choose a plan rather than be assigned a plan.
We made sure plans were actionable by making them concrete and granular.
To discourage clients from giving up on a plan, we allowed for flexible timing: “If I don’t finish it this week, I can work on it again next week.”
Financial Advisors acted to guide and enforce the plan-making activities by helping clients select and create a plan at the end of each class and checking in on their progress during the next class.
Make Your Next Plan
Now look at your to-do list again. Is there anything on that list that needs to be rewritten as a plan? When will you read our next blog post? Where will you be? What will you be doing immediately beforehand? Make a plan—and stick to it.
Next BETA Project Post: Testing, Testing
After we define problems, diagnose behaviors and design solutions, we test those solutions. Our next post on the BETA Project will discuss our testing methodology. This post and other helpful insights from the BETA Project are available on CFED’s Behavioral Economics blog and BETA Project website.
Designing for Difficulty – The BETA Project
“Make It Easy” – it’s not just a Staples advertising gimmick, but a key design principle from behavioral economics. How can we make sure people sign up for 401(k) savings accounts? Make it easy by setting a default plan. How can we get people to eat right? Make it easier by designing an intuitive food ‘plate’ instead of a food pyramid. How can you make sure you get to the gym? Pay the extra money for an on-site locker so you do not have to remember your work-out clothes. These may seem like different types of interventions, but they have one thing in common: they make it easier for us to take the actions we want to take.
For the most part, the BETA Project team shies away from designing something that’s hard to use. When we want people to take an action, we make it simple. But sometimes we want to make it hard to take an action. One of our designs for Cleveland Housing Network (CHN) called for exactly that.
Late Fees & Mental Accounting
A few weeks ago, we discussed how we diagnosed problems at CHN. As part of our diagnosis, we found that some CHN residents didn’t seem overly worried about having to pay the late fee. Many lumped the $25 late fee into the rent payment.
When you think about it, this is odd. Though $25 isn’t a huge amount of money, it’s not pocket change, either, especially for the low-income households that CHN services. You would get angry if you received a $25 fine for a parking ticket or for returning a library book late. Why weren’t residents reacting to a fine for a late rent payment?
Behavioral economists call this phenomenon mental accounting. For an example, compare the following two passages:
Imagine that you are about to purchase a jacket for $125 and a calculator for $15. The calculator salesman informs you that the calculator you wish to buy is on sale for $10 at the other branch of the store, located 20 minutes away. Would you make the trip to the other store?
Imagine that you are about to purchase a jacket for $15 and a calculator for $125. The calculator salesman informs you that the calculator you wish to buy is on sale for $120 at the other branch of the store, located 20 minutes’ drive away. Would you make the trip to the other store?
To an economist, both passages are asking the same thing – would you be willing to spend 20 minutes to save $5? But people are much more willing to make the drive if they read the first passage. We tend to think about savings in relative terms. Reducing the price of a $15 calculator to $10 seems like a great bargain, while reducing the price of a $125 calculator to $120 seems trivial.
When we think about a $25 fine for a parking ticket, we’re comparing it to a “fine” of $0 (i.e., not being charged a fine at all). But CHN residents have to pay their late fee simultaneously with their rent, so they’re comparing something like $500 (the median rent) to $525 ($500 plus the $25 fine). Just like in the jacket and calculator problem, they think about the fine in relative terms, instead of absolute terms. $25 represents only 4% of their rent, so the fine seems relatively small.
The BETA Project Design
Because metal accounting was making the late fee seem small relative to the price of rent, we wanted to try to reframe the decision and disconnect it from the rent payment process. We couldn’t arrange for the late fee to be billed separately from paying rent, but we could try to isolate the process of making the decision to pay the rent late.
We did this by creating a late fee waiver. Residents were issued a one-time waiver, which they could use to cancel a late fee. This changes the decision process by adding another step. Rather than thinking, “Should I pay my rent today? It’s just $25 if I pay late,” residents now first have to decide whether they want to use the late fee waiver or not. This is a subtle, but important, difference. Rather than deciding whether to pay late and thereby increase the cost of rent payment from $500 to $525, the decision is now whether to spend $25 or use the waiver.
An interesting wrinkle is that we wanted the waiver to reframe the decision that people made, but we didn’t necessarily want them to use the waiver. The waiver had two potential effects. First, as discussed above, it could reframe the decision to make residents less willing to pay their rent late. But it also had a potential negative effect—it reduced the penalty of late payment, thereby reducing the disincentive.
This put us in an unusual position: we wanted to make a form that people would not want to use. So, we made it harder for residents to use the waiver.
How did we make the waivers harder to use?
To make the waiver harder to use, the BETA Project team had to make sure it both felt harder to use and actually was harder to use. We did this in several ways:
- First, we tried to induce the concept of scarcity. The waiver could only be used once, and we tried to frame it as something for residents to use in an emergency, when they really needed it. We hoped that most residents would save it until it expired.
- Second, we introduced several hassle factors in order to make it difficult to use. Even small hassles make it unlikely that something is used. We added big hassles—most notably, we required residents to come into the CHN offices with their waiver in order to activate it. They could not use it over the phone or over the internet, even though CHN administrators could have easily verified its one-time use.
Third, we incorporated several small design tweaks that made the waiver seem more difficult to use than it actually was. For example, each waiver had a unique serial number (even though it was unnecessary, since CHN could track who used it using tenant ID numbers). We also required a signature to use it.
Finally, we incorporated a small, but important phrase: “You choose when to use it.” We hoped that this would induce a feeling of autonomy and power over the rent payment process.
Were we successful? We’ll be reporting on the outcomes of the BETA experiments before the end of the year, but here’s something to think about: over the four months of the BETA project, 70 waivers were used out of the 373 that were issued. That’s 19%, which means that 81% of CHN residents were not taking up “free money.”
Next BETA Project Post: Simple Plan-Making Strategies
Our next post on the BETA Project will discuss how to create an effective plan—and stick to it. This post and other helpful insights from the BETA Project are available on CFED’s Behavioral Economics blog and BETA Project website.
Read this Now! The Art & Science of Reminders
Have you ever meant to do something so important or so forgettable that you created a reminder for yourself, only to find that you still failed to follow through? Maybe you wrote a note reminding yourself to pick up the dry cleaning on the way home from work, but you completely forgot after a busy day. Or maybe you wrote a phone number on your hand last night, but unfortunately (or fortunately) didn’t write down the name or the reason to call.
Reminders are helpful for overcoming prospective memory failures — failures that occur when people forget to do planned actions. But why do some reminders work, and some fail? We have found that when reminders fail, it is often for one of the following three reasons: (1) the wrong action was prompted, (2) the reminder was not specific or (3) the reminder came at the wrong time.
In diagnosing the behavioral bottlenecks for Accion Texas, the BETA Project team looked at Accion’s loan repayment process through the clients’ eyes. Taking this perspective allowed us to see that reminders may be helpful to encourage on-time payments as borrowers seemingly intended to pay, but failed to follow through. When designing reminders for Accion Texas, we made sure the reminders were actionable, specific and timely.
Prompting the Right Action
Accion Texas’ existing monthly statement emphasizes the importance of making a payment by the payment due date. However, most borrowers are enrolled in automatic payment withdrawals, and actually need to make sure that they make a deposit to have sufficient funds. Since it can take a couple of days to process payment deposit, they also need to make sure that any necessary funds are deposited well before the due date.
We were concerned that borrowers may not consciously realize how their Accion Texas loan payment process differs from many other monthly payments. As such, we designed our reminders to emphasize the action of making a deposit rather than making a payment.
To refocus borrowers on the action of making a deposit, we redesigned the monthly statement to include a Post-it note to help borrowers (1) create a detailed plan about when they will make the deposit before the withdrawal, and (2) commit to the plan. We also provided an emphatic “suggested deposit date” on the statement to help borrowers plan out when and how to make a deposit.
The email and text messages reinforced the statement and the Post-it note by explicitly directing borrowers to make sure they have enough funds in their account for the payment.
In each component of the design, our reminders didn’t simply tell the borrowers to “pay on time,” but were intended to elicit very specific behaviors.
When reminders are received is just as important as what the reminders say. Accion Texas borrowers need to ensure there are sufficient funds in their accounts before their due date when the withdrawal occurs. But because only the due date is made salient in the loan contract and monthly statements, a borrower may be wrongly anchored to the due date, when it is too late to take action.
Our email and text reminders were deliberately timed to be far enough before the due date to provide sufficient time to act, and sufficiently close to maintain the salience and urgency to make a deposit. The emails were sent ten days prior to a given borrower’s due date each month. The text reminders were sent three days prior to the due date. Both messages were sent in the morning so that borrowers could make a deposit during daytime business hours.
Without appropriate timing, even the best-worded reminder can fail.
Now that you’ve learned about the art of reminders, try making one to remind yourself to come back and check our next blog post!
Next BETA Project Post: Designing for Difficulty
All of us get fed up with the hassles in our lives. But sometimes, hassles can actually make our lives better. Our next post on the BETA Project will discuss the intentional design of hassles. This post and other helpful insights from the BETA Project are available on CFED’s Behavioral Economics blog and BETA Project website.
Take a Walk in Someone Else’s Shoes – The BETA Project
At what point can we say that we understand another person’s behavior? "Before you abuse, criticize and accuse," singer Joe South told us, "Walk a mile in my shoes." If we walked a mile in the shoes of every client that our three BETA test sites serve, we would need to cover approximately 3,797 miles.
Luckily, in order to diagnose a behavioral problem, we don’t need to exhaustively understand the mindset and behavior of every single client at our partner sites. Human behavior is shaped by a complex blend of contextual features and internal neuro-cognitive processes into which we have limited access, and different people demonstrate different behaviors. We simply aim to better understand the contextual features that become “behavioral bottlenecks” preventing many clients from reaching their goals.
In previous blog posts, we shared behavioral diagnosis tactics that include proving ourselves wrong and looking for unexpected details. Another useful technique that may unearth important contextual details is taking the perspective of the end-user, to the extent possible. We may not be able to walk a full mile in each client’s shoes—but we should at least take a quick stroll.
Diagnosis Tactic #3: We experience the end-user’s perspective.
State the problem. At Accion Texas, we set out to solve the following problem:
Borrowers have difficulties making consistent, on-time repayments using the Automatic Clearing House (ACH) electronic withdrawal system.
Generate ideas. Our first step in tackling this problem was to map out the loan application, disbursement and repayment process at Accion Texas. Discussions with staff members enabled us to render a detailed process map of the entire process from start to finish. Based on this representation of the borrowing process, the behavioral bottleneck appeared to be at the actual moment of repayment, when many borrowers seemingly intended to pay, but mysteriously failed to follow through at the last minute.
Look for clues. As we conducted site visit interviews with additional employees, a fuller picture came into focus. Numerous staff members at Accion, including the Collections and Accounting teams, reported that changing the payment date was one of the most common loan adjustments that borrowers perform. We also heard from the Collections team that many of these borrowers do not make the change until they have encountered an issue with payment, even if they already know that the previous payment date doesn’t work for them. For instance, the payment may be scheduled soon after their rent is due and money is tight. Finally, we found that the timing of the monthly statements sent to borrowers can be inconsistent, meaning that these statements may not be serving as effective reminders to make a payment.
Change your perspective. Piecing together each of these perspectives allows us to better understand the process, but we also need to make sure we are viewing them through the right lens. We ultimately did a full remapping of the user experience (from the end-user’s perspective rather than the organization’s perspective), conducted client interviews to fully understand process details and viewed actual materials (like the monthly statement) in precisely the form that clients would see them. Through this process, we saw, through a borrower’s eyes, how small details in the application process (such as assigning an arbitrary repayment date to borrowers) and repayment process (such as sending out statements with inconsistent timing) can contribute to late repayment.
Next BETA Project Post: Design
At our partner sites, we used multiple tactics to gain insight into what contextual features of each program might be contributing to these behavioral problems. Although diagnosis is an important phase of our methodology, we must remember that the entire sequence is iterative and continuously changing as our understanding grows. We will revisit all of our assumptions in later stages, and we need to continue to ask whether we are asking the right questions.
This post and other helpful insights from the BETA Project are available on the Behavioral Economics blog and the BETA Project website. Our next post will look at how behavioral diagnosis sets the stage for our next phase: design. We will discuss how we use our behavioral diagnosis to design innovative solutions for our partner sites.
Look for the Unexpected – The BETA Project
Part of diagnosing a behavioral problem is realizing that you don’t always know where to look for the “symptoms.” In medical diagnosis, symptoms are at least limited to the physical human body. Human behavior, on the other hand, is shaped by a complex blend of contextual details and internal neuro-cognitive processes into which we have limited access.
Nevertheless, there are a few tactics we can use to pinpoint the “behavioral bottlenecks” that may be preventing someone from reaching their desired outcome. One method we use in the BETA Project is to generate hypotheses and try to find evidence to prove ourselves wrong. Below, we share another tactic to diagnose the underlying behaviors and psychologies at play for a given problem.
Diagnosis Tactic #2: We look for overlooked details.
State the problem. At partner site Neighborhood Trust, we set out to tackle the following issue:
Low-income individuals sign up for accounts with affiliated credit unions during Neighborhood Trust’s financial education course, but do not fully utilize them.
Generate ideas. During our preliminary diagnosis process, we wondered how frequently Neighborhood Trust clients used their accounts after they were first opened. Perhaps, we thought, clients didn’t use their accounts often and long enough for it to become a habit before they graduated from the financial education course.
Look for clues. In fact, client interviews conducted during our site visit suggested that some clients may never visit the credit union or use their account, even once, after account opening. One client reported that she intended to enroll in direct deposit with her employer, but never quite got around to it before she lost that job. She continues to use money orders to pay her bills rather than her account, which remained dormant.
Look beneath the surface. In our initial hypothesis, we thought that some clients may not have used their accounts enough. During site visits, we found that some clients may not have used their accounts at all. This finding prompted us to dig a bit deeper into the earlier stages of the account opening process and course content, with an eye out for counterintuitive, unexpected details.
Through observation, interviews and analysis, we discovered that Neighborhood Trust is incredibly successful at making it easy for clients to open credit union accounts. Account applications are included in the course curriculum, there are recurring and predictable opportunities for clients to gather documents and open an account, and course instructors can provide direct assistance.
Account usage, on the other hand, remained largely outside the purview of the course. Actions related to account usage like finding the nearest ATM or credit union branch, learning how to use online banking and activating a debit card for the first time were riddled with small inconveniences.
This led us to believe that the course was very effective at helping clients take action to open an account. However, the course was lacking in the later stages of guiding clients from account opening to active account usage. Even though these steps appear to be simple, small barriers can have a surprisingly large effect. These "hassle factors” could prompt a client to procrastinate and put off a task that seems difficult in favor of more familiar options, like money orders.
Only by diving in to examine the gritty details of the client experience were we able to detect this “behavioral bottleneck.”
Next BETA Project Post: Take a Walk in Someone Else’s Shoes
As mentioned in “Don’t Suppose, Diagnose,” we use a range of tactics to elicit insights during the behavioral diagnosis process. Our next post on the BETA Project will discuss another strategy we use in the field: looking from the perspective of the end-user. This post and other helpful insights from the BETA Project are available on CFED’s Behavioral Economics blog and BETA Project website.
Being Wrong is Sometimes Right – The BETA Project
Sometimes it’s good to make mistakes. As soul singer Joss Stone says, “I've got a right to be wrong. My mistakes will make me strong.” In behavioral diagnosis, as in life, being wrong is sometimes helpful – especially when it stimulates new insights on the problem. One of our favorite strategies to test hypotheses in behavioral diagnosis is looking for clues to try and prove ourselves wrong.
Diagnosis Tactic #1: We try to prove ourselves wrong.
State the problem. At Cleveland Housing Network, we started off with the following problem:
Despite multiple payment options, residents in the Lease Purchase Program fail to pay their rent on time.
Generate ideas. Lease Purchase Program residents who pay their rent late are assessed fines, starting with a $25 late fee and going up to $150 in court fees if the resident enters the eviction process. One of our initial hypotheses was that some residents may not be fully aware of the amounts of the fees they incur when paying late. If residents don’t attend to how much they are paying in fees, we conjectured, they may not be motivated to make the effort to pay rent on time. We often ignore small fees when making decisions. For example, we will sometimes happily spend $2 in ATM withdrawal fees rather than walk across the street or plan our spending ahead of time.
Look for clues. Our site visit to Cleveland proved us wrong. Through interviews with a range of residents and employees, we discovered that residents are well aware of the amount and timing of the fees. In fact, some residents anticipate paying a late fee and consider it a minor cost to accompany their monthly rent check. For someone paying the median rent ($500), the additional fee might not seem like much. The difference between paying $500 and $525 seems trivial because they had anchored to the high cost of rent—even though it adds up over the cost of the year.
Revisit your initial ideas. We had discovered a new behavioral bottleneck: the fee may not deter some residents from paying late, even if they are fully aware of it. Residents were letting the extra $25 go because it didn’t seem like very much compared to the cost of rent. This is a common cognitive error that affects all people, as we have a tendency to think in terms of percentages, rather than totals. We’re happy to spend 5 minutes haggling for a $2 discount on a $5 pair of sunglasses, but fail to spend the same time negotiating for $10,000 off a $300,000 house. In the first case we are negotiating for 40% off, but we are only negotiating for 3% in the latter case. If we do not think about the actual monetary values involved, our intuitions can lead us astray. As a result, the loss—in CHN’s case, the late fees—had a minor impact on behavior.
Another interesting human reaction to fines may also be at play here. Fines can change the nature of the transaction, turning the undesirable behavior into something that a person can choose to engage in for a cost, rather than something they should try to avoid. In the absence of a fine, a resident may think that paying rent on time is something that “good” residents do and will pay their rent accordingly. However, when a fine for paying late is introduced, the resident may subconsciously think, “Great, it’s okay if I’m late with rent because I’m buying the privilege by paying the late fee.” The desire to pay rent on time so that they can remain a “good” resident is replaced by a cost-benefit calculation where it may be worth it to pay the fine for the ability to be late (especially since it probably seems like a small amount compared with the total rent payment).
With these new insights, it’s likely our original hypothesis was incorrect. But, we wouldn’t have known that if we hadn’t taken the steps to formulate an initial theory and seek information to prove ourselves wrong.
Next BETA Project Post: Look for the Unexpected
As mentioned in, “Don’t Suppose, Diagnose,” we use a range of tactics to elicit insights during the behavioral diagnosis process. Our next post on the BETA Project will discuss another strategy we use in the field: looking for unexpected details. This post and other helpful insights from the BETA Project are available on CFED’s Behavioral Economics blog and BETA Project website.
Don’t Suppose, Diagnose! – The BETA Project
The last time you visited the doctor with an illness, what did your physician do? After taking vital signs, your doctor probably examined you, asked questions about your symptoms, possibly asked about your diet, lifestyle or recent events, and perhaps scheduled follow-up tests. This combination of data and contextual details allowed your doctor to make an informed guess about what type of treatment might help your condition, rather than make uninformed suppositions.
The behavioral diagnosis process follows similar principles of data collection and discovery, but applies them in a very different way. In the BETA Project, behavioral diagnosis is the second phase of a four-stage problem solving process: define, diagnose, design and test.
The goal for the diagnose stage is to tease out factors and contextual details that might be contributing to the behavioral problem identified during the define stage. Diagnosis is an iterative process for charting the decisions and actions an individual must take to reach the desired outcome, constructing informed hypotheses about psychologies at play and looking for evidence in the field that helps us refine those hypotheses. Think of the TV show, House. In each episode, a patient comes into the hospital with a condition. The team of doctors work on the case throughout the episode, observing symptoms and reflecting back on past cases and medical research to create hypotheses regarding what might be causing the symptoms. The first hypothesis is never right, but as the team tries to test each hypothesis, they learn more about the patient’s condition and get closer to the proper diagnosis.
Behavioral diagnosis is rarely complete with one hypothesis, but requires multiple iterations. Additionally, behavioral diagnosis is unlike medical diagnosis (especially the kind that happens on TV) in that the process will very rarely yield a single, clear answer for why people are behaving a certain way. Moreover, we recognize that different people demonstrate different behaviors. Our goal here is not to pinpoint a precise cause of a precise behavior, but to better understand the key “behavioral bottlenecks,” or places where human behavior may be preventing someone from reaching their goals in a given context.
With something as nebulous and complex as human behavior, how does one even start a diagnosis? You could ask yourself, “why are people doing [the problematic behavior]?” to generate hypotheses on potential causes of the problem. You could also look for clues to assess whether or not any of the ideas are on track. In our next three blog posts, we will share a few of the tactics we use to diagnose the underlying behaviors and psychologies at each of our BETA Project test sites.
Next BETA Project Post: Being Wrong is Sometimes Right
Our next post on the BETA Project will discuss one of our favorite strategies we use in the field: proving ourselves wrong. This post and other helpful insights from the BETA Project are available on CFED’s Behavioral Economics blog and BETA Project website.
Defining Problem Statements in the BETA Project
While defining the problem is the first step towards a solution, crafting the right problem statement is inherently difficult. As we mentioned in our last post, a well-crafted problem statement should not be defined too broadly, too narrowly or with hidden presumptions. Here’s how we refined the problem statements at each of our three pilot sites in the BETA Project.
Neighborhood Trust Financial Partners
The initial problem statement from Neighborhood Trust’s application to the BETA Project was:
Barriers such as inconvenient locations of credit unions and banks keep “unbanked but bankable” clients from using asset-enhancing bank accounts and remain reliant on expensive fringe financial services.
This problem definition seemed too broad to our team because it concentrated on the use of fringe financial services. It presumed that bank account use is a perfect substitute for fringe financial services, while these services can actually address very different financial needs (e.g., using a checking account to pay bills vs. taking out a payday loan). We also felt that focusing too narrowly on “unbanked but bankable” clients would unnecessarily restrict the sample of clients we seek to serve through the project and prevent other client segments from potentially benefitting from an intervention.
Our refined problem statement is:
Low-income individuals sign up for accounts with affiliated credit unions during Neighborhood Trust’s financial education course, but do not fully utilize them.
In contrast to the original statement, the refined statement focuses purely on client account use, without any assumption that this will necessarily lead to reduced use of fringe financial services, and includes the full client population.
Accion Texas, Inc.
In contrast to the Neighborhood Trust example, Accion Texas’ application contained a problem statement that seemed too narrow:
For individuals with credit scores lower than 522, having separated personal and business checking accounts almost doubles the probability of repayment in comparison to individuals that only have one checking account or have no bank account at all. Accion hopes to help the underbanked borrowers improve loan repayment rates so they can build or improve their credit and move up the asset-building chain.
The original problem statement directed us to look at loan repayment rates, but when we gathered more information about delinquencies, we found that Accion’s customers repaid their loans at rates on par with, if not better than, the rates expected for a nonprofit microlender. Additionally, Accion’s problem statement concentrated specifically on underbanked individuals with low credit scores. Segmenting borrowers based on their credit scores or banking status precluded an accurate diagnosis of the problem. A person’s financial status can be in a state of flux: credit scores go up and down, bank accounts are opened and closed. Furthermore, while the underbanked may face unique challenges in repayment, we did not want to exclude the possibility of interventions targeting all of Accion’s clients.
Based on conversations with Accion staff, we learned that late payments took a significant amount of staff resources to manage and that there were negative consequences for borrowers. Customers were charged late fees and Non-Sufficient Funds fees and were at risk of damaging their credit scores if Accion reported their delinquencies. Additionally, most borrowers were enrolled in automatic electronic payments, but sometimes did not have enough money in the account on the date their payments were withdrawn. This led the BETA Project team to revise the problem statement to target on-time loan payments:
Borrowers have difficulties making consistent, on-time repayments using the Automatic Clearing House (ACH) electronic withdrawal system.
This changed the focus to preventing borrowers from becoming past due and entering the collections process, rather than on borrowers who are already past due on their repayment.
Cleveland Housing Network
Finally, problem statements can contain hidden presumptions that limit the possibilities for diagnosis and design. Cleveland Housing Network’s original problem statement contained this kind of presumption:
Barriers such as poor marketing inhibit residents in the Lease Purchase Program from paying rent online.
This statement made two presumptions. First, it presumed that poor marketing of the program is what is preventing online payment. Second, it presumed that if more people signed up to pay rent online, they would be more likely to pay their rent on time. This second presumption is hidden—meaning it is implied rather than directly stated. In our initial discussions with the Cleveland Housing Network, management indicated that on-time rent payment was the goal, and that online payment was a means to reach that ultimate goal. We felt the presumptions would limit the range of possible solutions to the ultimate goal of repayment. Further, we checked the second assumption against one month of data and found that many people paid their rent late even if they had signed up for the online payment system.
After further discussion with the Cleveland Housing Network, we revised the statement:
Despite multiple payment options, clients in the Lease Purchase Program fail to pay their rent on time.
The revised problem statement eliminates these presumptions. It also presents us with a different target: we attempt to encourage residents to pay their rent on time, rather than focus only on online payments.
Next BETA Project Post: Diagnosis
This post and other helpful insights from the BETA Project are available on the Behavioral Economics blog and the BETA Project website. Our next post will look at how reworking the problem statements for the BETA project sets the stage for our next phase: diagnosis. We will discuss how we started the process of moving from these problem statements to a diagnosis of the underlying behaviors and psychologies that may be preventing clients from achieving their desired outcomes.
Helpful Tactics to Define a Problem
Behavioral economics is, ultimately, about how we think of people. The assumptions we make about people change how we approach problems related to their behavior. If we assume that their actions follow their intentions, we will design programs that attempt to change intentions. If we think that people take an action if they are informed of the consequences, we will design programs to educate them. However, if we take people as they are—fallible yet clever, short-tempered yet patient, the paragon of animals and yet the quintessence of dust—we can design interventions that work.
This approach makes defining a problem (that gets to the core issue at hand and that can be addressed by behavioral diagnosis and design) really hard. Crafting a problem definition is as much an art as a science. As such, there are no step-by-step set of instructions to follow. However, we will share a few of our favorite tips and tricks that have proven useful in the BETA Project.
Tactic #1: Change the Scope
A problem statement often starts off as a piece of a bigger problem or as a collection of smaller problems. When a problem is defined too broadly or too narrowly, a change of scope is necessary. A problem statement that is too broadly defined often falls beyond that project’s scope of work and can feel overwhelming and daunting. In the BETA Project, to check for this, we would ask ourselves, “What are the components of this problem? Of these, what is the highest priority and achievable?”
A narrow problem statement, on the other hand, limits investigators from exploring other areas that may ultimately prove relevant. It feels like solving it won’t actually get you to the desired goal. To check for a statement that is too narrow, we would ask, “Is this part of another problem? Would fixing this problem just be one of many other fixes necessary to solve that other problem?”
While there isn’t one right answer to any of these questions, reflecting upon them in the BETA Project often helped us detect a hazy problem definition. It also set us up to practice the following two problem definition tactics.
Tactic #2: Remove Assumptions
In April, we posted a summary of the 99 problems presented to the BETA Project. It was really interesting to see how these problems looked from the applicants’ perspectives, but the problem statements often contained hidden assumptions about the challenge posed.
For example, some applicants reported that they experienced low take-up of their program because their promotional materials are not well designed. This assumes that their problem is bad advertising and that low take-up is due to a lack of knowledge about the program. Looking at the problem with these assumptions sets someone up to try to fix the advertising, without thinking about whether or not there could be other reasons for the low take-up. Assumptions limit the exploration of possible solutions in many domains, from the field of asset building to Antarctic exploration.
Consider the classic behavioral problem of getting people to save more for retirement. For decades, human resources professionals have tried to encourage employees to save more for retirement. They typically used one of two approaches: either increasing the employer match, or encouraging employees to attend seminars. They rarely thought about how the problem was defined and instead focused on costly incentives and time-consuming education as potential solutions.
These approaches have built-in assumptions about why people were not saving. They assume that people are not saving because of a lack of motivation (which would be solved by increasing the employee match) or education (which would be solved by classes). Incentives and education are powerful tools for a program designer, but they should be considered two tools among many.
By removing these assumptions and asking “How can we get people to save more?” we open up the range of possible solutions to try out. New possible diagnoses (such as limited attention) present themselves, and they imply novel solutions (such as “opt out” 401(k) programs).
Tactic #3: Change Representation
Another useful tactic for problem solving is changing how the problem is represented. Imagine that you and a friend are playing a game. In the game, you lay out nine cards (an ace and all the numbered cards, two through nine), and you alternate turns picking up cards from a table until one person has three cards that total exactly fifteen. You start by laying out the cards in a row and start to play.
How should you even start? It’s really hard to determine the best way to play when the cards are laid out this way. More likely than not, you’ll end up with a game that looks something like this where there’s no way you can win—or block your friend from winning.
How could you avoid this situation? What if you went back to the beginning of the game and re-thought the problem? You realize that the object of the game is to lay out the cards and be the first to get three cards totaling exactly fifteen. There’s nothing that says the cards initially need to be laid out in a row. By laying out the game and representing the problem that way, you made it hard for you to play. So, you rethink the layout of the cards and position them in a square where each row, column, and diagonal totals 15.
With the numbers arranged in a “magic square,” the problem becomes simple because it’s set up like a tic-tac-toe game. As long as you pick a corner card first, you can’t lose (watch “How to Win Tic Tac Toe Every Time” if you don’t believe us). Changing how we represent the problem makes it easier to see the path towards solutions. For practitioners designing programs, changing representations can be done by thinking about the situation from a different perspective. For example, you could ask yourself, “How does my client view this situation? Would they think there is a problem? How would they define it?”
Next BETA Project Post: Defining Problem Statements in the BETA Project
By using each of these tactics in the BETA Project, we were able to refine the problem statement at each site to arrive at problem statements that are not defined too broadly, too narrowly or tangled with hidden presumptions. Our next post on the BETA Project will look at the original and final problem statements for each site and how we refined them. This post and other helpful insights from the BETA Project are available on CFED’s Behavioral Economics blog and BETA Project website.
The First Step towards a Solution
EDITOR’S NOTE: Today’s post is the first of a three-part series on defining problems for behaviorally informed interventions. Come back tomorrow and Thursday for the other two posts in this series.
At each of our pilot sites, the BETA Project uses a four-stage problem-solving process: define, diagnose, design and test. In the coming months, we will highlight interesting findings from our work at each stage over this past year.
We’ll start by discussing the define stage, where we attempt to correctly define a problem that can be addressed by behavioral diagnosis and design.
In order to solve a problem, you need to know what the problem is. Individuals and organizations often fail to take adequate time to define their problem. Instead, they leap directly into creating solutions. Only when those solutions fail do they go back to the drawing board and attempt to figure out exactly what problem they are trying to solve.
There are examples of failures to properly define problems in many domains. An especially illustrative example comes from the failure of the famous Arctic explorer Robert Falcon Scott’s “Terra Nova” expedition.
Scurvy, Scott & Getting to the South Pole
Robert Falcon Scott wanted to be the first explorer to reach the South Pole. His first “Discovery” expedition in 1902 was thwarted by difficulties with starving sled dogs and an outbreak of scurvy among men on the team. Scurvy is a horrible disease with symptoms that include extreme tiredness, disintegrating gums, immobilization and open wounds that won’t heal. Given his experience, in preparation for Terra Nova eight years later, Scott attempted to prevent scurvy by seeking the latest advice from doctors and other Arctic explorers. He was told that scurvy was caused by bacteria in tainted canned meat, and took great care to prepare the Terra Nova team accordingly. Unfortunately, these precautions didn’t help, and Scott and his team ended up in an icy grave.
Today we know that scurvy is caused by a vitamin deficiency and can be prevented and treated by eating fresh fruits, vegetables and certain animal products that contain vitamin C. While this specific fact wasn’t known to Scott and many of his contemporaries, medical research in the 1700s concluded that "scurvy is solely owing to a total abstinence from fresh vegetable food and greens; which is alone the primary cause of the disease." By 1740, the British Royal Navy mandated that every sailor receive a daily dose of lemon or lime juice to forestall the disease. If it was known for almost 200 years that lemon or lime juice could prevent and cure scurvy, why in the world was Scott trying to guard against bacteria in tainted meat?
One major conundrum is that Scott and the scurvy experts he consulted defined the problem incorrectly. While doctors in the 1700s took a wide lens to the problem and asked “How do we prevent and treat scurvy?” with the development of germ theory in disease research by the early 1900s, Scott and his contemporaries took a more narrow lens. They assumed that scurvy had a bacteriological origin (borrowed from germ theory), and consequently defined the problem as “How do we ensure that our meat is not tainted with bacteria?” For many diseases, like cholera, germ theory was an important medical advancement because bacteria are the actual cause. But for scurvy, it was a step backward, as scurvy is caused by a vitamin deficiency and not the presence of bacteria.
When a problem is poorly defined, it is difficult to correctly interpret information. Experienced Arctic explorers such as Scott noticed that when explorers ate fresh seal meat, the symptoms of scurvy lessened. However, when they preserved the meat, scurvy re-surfaced. We now know that this is because the process of preserving and cooking meat leached away vitamin C. But since they had defined the problem incorrectly, Scott’s expedition believed this was evidence that they had failed to properly preserve the meat. Tragically, the expedition would throw out meat that was perfectly well-preserved (though lacking vitamin C) under the mistaken belief that it was tainted with bacteria.
The advance in medical knowledge was, in this case, a curse. Rather than exploring what actually causes scurvy, practitioners in the field started to take intense efforts to rid their food of bacteria, making Scott’s efforts to protect his team from scurvy entirely ineffective. The problem was defined too narrowly, and it contained the assumption that scurvy was caused by bacteria. With a broader definition of the problem, Scott’s might have returned home safely, rather than resting in a cairn somewhere in the Ross Ice Shelf.
Next BETA Project Post: Helpful Tactics to Define a Problem
How can we, as practitioners in the consumer finance field, avoid similar pitfalls of poor problem definition? In our next BETA Project blog post, we will highlight some problem definition tactics that were especially helpful to us and may be helpful to leverage findings from behavioral theory for your own program. These posts and other helpful insights from the BETA Project will be available on CFED’s Behavioral Economics blog and the BETA Project website.
Currently reading page 1 of 4.