Financial student aid improves educational outcomes and reduces social inequality, yet many eligible students do not take it up. To examine whether correcting misperceptions increases take-up, I conducted an RCT with 6,225 university students across Germany who were not receiving aid. I find that 63% of students systematically underestimate the financial benefits and overestimate repayment obligations of student aid, and 86% misperceive their eligibility. Providing combined information about the program conditions and individual eligibility significantly corrected misperceptions after six months and increased take-up by 46% after one year. This increase is particularly strong among disadvantaged students. After take-up, students report higher available income while reducing earnings and parental support. These findings suggest that correcting misperceptions can reduce social inequality by alleviating financial constraints among disadvantaged students and their parents.

The newest version of the paper is available here.
SSRN: https://dx.doi.org/10.2139/ssrn.5364685

Media coverage: Die Zeit, Spiegel, Wiarda-Blog

While take-up of social benefits is low across the world, it remains unclear if people do not know they are eligible (eligibility barrier) or actively decide against take-up (application barrier). Using survey data from 22,222 students in Germany, we identify students who are eligible but do not take up means-tested student aid. Among them, we find an eligibility barrier of 82.5% students who do not believe to be eligible, and an application barrier of 13.0% of students who know about their eligibility but still do not take up aid. Both barriers are severely driven by misperceptions. At the eligibility barrier, 59% of students think they will not pass the means-test, but 61% of them underestimate the eligibility conditions. At the application barrier, 63% do not take up aid due to debt aversion, but 76% of them overestimate the repayment. The results show that misperceptions are a main inhibitor of social benefit take-up at both barriers.

Paper available on request.

Most interventions fail to change behavior. We argue that the reason for this failure is that the interventions do not address the underlying problem. Identifying the type of underlying problem requires systematic pre-intervention diagnostics. Yet, a systematic, parsimonious, and generalizable survey tool (anamnesis) to identify problems is hitherto missing. Our anamnesis classifies the underlying type of problem along three fundamental diagnoses: awareness, intention, and implementation problems. We then test whether we can use our anamnesis results to predict the effectiveness of three typical interventions (reminders, incentives, simplifications) across three different settings in an online experiment with 7,500 subjects. We find high context-specificity of the interventions across the three different settings, which can be predicted by our anamnesis results. On average, a predicted effectiveness of 100% ex-ante translates into an actual effectiveness of an intervention of 89.2%. Choosing an intervention based on our diagnosis increases the effect size by around 58% compared to randomly choosing one of the tested interventions.

Manuscript in preparation.

In a series of three experiments, we investigate intrinsic preferences for timing and skewness of information resolution in a "compound lottery" (CL) and "information structure" (IS) environment. In the first experiment, we rule out predeterminedness, i.e., realization before making the decision, of the outcome as a driver of different timing preferences between the environments. In the second experiment, we show that skewness preferences in a two-outcome setting differ between the two environments. In the third comprehensive experiment, we integrate all three dimensions (timing, skewness, and environment) into a three-outcome setting, which allows a rigorous treatment of skewness. With respect to timing, participants prefer to receive information early in both environments. In case of a gradual resolution, participants strictly prefer positively skewed information in the IS environment, but symmetric information in the CL environment. Our results show that skewness and environment must be jointly taken into account when studying preferences for information resolution.

Revised manuscript available here.

We study utility-based intensity measures of higher-order risk preferences and their relation to field behavior in a representative German sample of 2970 adults. Using a new method based on the elicitation of certainty equivalents and a machine learning approach, we find that 85%, 77%, and 62% express risk-averse, prudent, and temperate behavior, respectively. Interestingly, also among risk seekers, more than 70% behave prudentially, which extends to the intensive margin and documents mixed risk aversion and mixed risk-seeking behavior. Age, gender, education, and religion are significantly related to risk aversion, prudence, and temperance. Our intensity measures of risk preferences are significantly related to financial decision-making, job choice, and addictive behaviors. Importantly, field behavior that according to economic theory is insufficiently explained by risk aversion alone is significantly predicted by prudence, where risk aversion mostly remains insignificant.

Manuscript in preparation.