“How Academic Peer Effects Do (and Don’t) Work”, with Nicolas Salamanca.
“Quantifying Aspirational Poverty Traps”, with Nicolas Salamanca.
Abstract: Recent theoretical findings from behavioral economics argue that, aside from classical poverty traps, there exist aspirational poverty traps, a specific type of behavioral poverty traps in which low–effort equilibria are perpetuated not by people’s lack of resources but by their imperfect aspiration–formation mechanisms. These findings have important implications for the design of anti-poverty programs, since they suggest that some people can be pulled out of poverty by shifting their aspirations – an alternative and arguably cheaper approach than traditional monetary transfers. There is, however, little empirical evidence of the existence and economic relevance of these aspirational poverty traps. In this paper, we derive testable predictions which allow us to quantify the potential size of aspirational poverty traps using existing data from large-scale anti-poverty program evaluations. Using data from the National Longitudinal Survey of Youth 1979 (NLSY79), we aim to assess the size of the different populations which can be subject to aspirational poverty traps. We also aim to characterize these populations so it is easier to target them with aspiration-shifting mechanisms. Finally, we discuss how our results can be extended to inform the design of potentially more cost-effective welfare policies.
JEL Codes: C13, D03, J60
Keywords: behavioral agents; anti-poverty programs; threshold regression
“From Subsidies to Loans: The Effects of A College Financing Reform on Students’ Secondary School Choices”, with Jan Kabátek
Abstract: We estimate how high school students respond to uncertainty shocks in the availability of higher education. We exploit the announcements and implementation timing of a unique reform in the Netherlands which turned universal, unconditional grants for higher education students into income-contingent loans. The Dutch government first announced an intention to amend college financing, and later revealed the details and timing of the reform, thus creating 1) uncertainty surrounding the costs of attending any type of post-secondary education, 2) subsequently resolving this uncertainty with higher expected costs of attending higher education but not vocational education, and 3) at implementation, increasing the cost of higher education. Using individual administrative records, we find that as uncertainty increases, students shy away from college-preparing tracks. As uncertainty is resolved, students further avoid college-preparing tracks, yet those who do select into them choose STEM fields more often. When the reform is eventually implemented, selection away from college-preparing tracks and into STEM within those tracks peak. Using archives and aggregate administrative data from all countries neighboring the Netherlands, we show that our results are not driven by larger trends. Our results reveal forward-looking behavior of students entering high school consistent with debt aversion, with distributional effects across family income and cognitive ability that are important for the design of optimal student loan policies.
JEL Codes: I23, I28, C25.
Keywords: educational choices; information; policy uncertainty; student loans
“Drivers of Educational Decisions: Risk Preferences, Information Differentials and the Conditional Variance of Wages”
Abstract: High-achieving students from low-income families apply less to college compared to richer peers. Important explanations are differences in preferences and in information, founded on the existence of a steep socioeconomic gradient in information. This paper uses a variance decomposition of wages to test whether the college gap can be explained by differences in information and/or differences in risk preferences. I test the hypothesis of differences in information by looking at the patterns of unobserved heterogeneity, and the hypothesis of differences in risk preferences by looking at the patterns of wage uncertainty. Using the NLSY79, I find evidence that within any educational groups, students from different backgrounds are subject to different amounts of wage uncertainty, which makes it difficult to conclude that low-income students are more risk averse. I also find evidence of a gradient in information. This finding suggests that policy-makers should focus on interventions providing students with more information about education and occupations.
JEL Code: J31, J24, J62
Keywords: educational choices; risk preferences; information