Many poor people do not benefit from the transfers and services they are eligible for. Researchers consider this problem, often referred to as the non take-up problem, as a major source of poverty persistence in rich societies. Yet, the reasons why potential beneficiaries do not take up those services are not fully understood. Recent works have made it clear that the traditional explanations in terms of stigma, lack of information or cost of application only capture a small part of the picture. Saurabh Bhargava and Dayanand Manoli recently conducted a field experiment that sheds new light on that phenomenon. The experiment is about the taking up of earned income tax credits (EITC).
The EITC program is the largest US federal cash transfer program. It benefits more than 25 millions of households. It consists of income subsidies that depend on the earned income. Only households earning less than a certain threshold are eligible, and the threshold, as well as the amount of income subsidy, depend on the size of the household and favour households with children.
According to estimates based on administrative data and tax returns, 6.7 million eligible people do not claim their credits. The forgone credit is worth $1,096 on average. Each year, based on tax returns, the Internal Revenue Service (IRS) identifies a subgroup of non-claimants tax filers that are likely eligible for a tax credit. A few weeks after IRS receives the tax returns, it sends a notice to these tax filers, reminding them that they should claim the credit. This mail contains the form to fill and send back. In California, where the experiment took place, 59% of those people receiving the IRS reminder did not send the form back and, therefore, did not claim for their due credit. They are the sample on which the experiment was conducted.
A few months later, the authors, with the help of the IRS, send to all subjects a new mail encouraging them to apply for the tax credit. The mail was composed of an envelope, a notice and a worksheet. In the control group of subjects, all three components were similar to the ones sent by the IRS months before. There were several treatments, consisting of modifying one or several of the components. The authors tested different kinds of modifications, trying to capture different possible explanations of non take-up. Here are the main findings of the experiment.
The first surprise of the authors was that no less than 22% of subjects from the control group did send their form back and apply for the tax credit. That means that a simple second mail with exactly the same information as the first one already convinced a significant set of people to request the benefit.
The largest treatment effect consisted in mentioning the potential benefit of the tax credit on the notice. That increased by 8 points the percentage of applying subjects. In some notices, the benefit was expressed in terms of the maximum possible tax credit, which is a function of the composition of the household. In other notices, the benefit was expressed in terms of the average possible tax credit. Whether the maximum or the average amount was mentioned had no impact on the rate of application. The likely explanation given by the authors is that non-claimants underestimated the potential gains of participating in the program, and mentioning it in the notice made them realize that it was worth trying.
The second largest treatment effect is about the complexity of the notice. Whereas the baseline notice was a one page letter clearly divided into four sections entitled: “summary,” “what you need to do,” “next steps” and “additional information,” the complex notice was a two-page one containing basically the same information but making it more difficult to understand what subjects had to do. Based on a supplement survey of low and moderate income tax payers, the authors explain the effect of the complexity of the notice by the fact that “it altered the attention recipients paid to the information as well as inferences made with respect to the costs and benefits of the program” (page 3509). Increasing the complexity of the worksheet had the same deterrence effect as increasing the complexity of the notice.
Some treatments had surprisingly no effect. For instance, adding a “Good news for you” mention on the envelope of the mail had no effect. Adding an information flyer in the mail, describing how the EITC program works had nothing but a deterring effect, probably because it increased the perceived cost of participating in the program. Adding mentions such as “You may have earned a refund due to your many hours of employment” or “Usually, four out of every five people claim their refund,” dedicated to lower the psychological and social stigma of claiming credit had a surprising significant negative effect on applications. The authors think that those mentions increased the perceived complexity and belief in the likelihood of an audit.
Even if it is hard to generalize the findings of this experiment to other social programs, it suggests that repeating the message that people are eligible, announcing the amounts that people are likely to obtain and making notices and forms as simple as possible could have an impact.
The authors conclude their study by mentioning that the right behavioural model to think of when one tries to understand why poor people don’t request the benefits they are eligible for is not a model of global evaluation of costs and benefits of applying. It is rather a model in which people are sensitive to what psychologists, after Kurt Lewin, call “hassle costs”, according to which in the subjective process of evaluating costs and benefits, people may bump into seemingly details, such as form complexities, confusion or uncertainty regarding eligibility or benefits, which lead them to avoid or postpone claiming. It would be interesting to know if such a model would also lead us to better understand other cases of non take-up.
 S. Bhargava and D. Manoli 2015, “Psychological frictions and incomplete take-up of social benefits: Evidence from an IRS field experiment,” American Economic Review 105(11) 3489-3529.