This project highlights heterogeneity in the causal effects of education - depending on underlying abilities and socioeconomic background - and how these interact with policies. We need to better understand this heterogeneity in order to design better education policy such that individuals can achieve their full potential. For example, knowing how much of the income dispersion across fields of college major is driven by initial abilities and skills relative to the skills acquired in college is pivotal in order to know when to target resources. Many argue there is a scarcity of Science, Technology, Engineering, and Mathematics (STEM) skills. If so, it may improve efficiency to target resources towards investments in STEM skills. If the STEM premium mainly reflects initial abilities of those who choose the STEM fields, then resources should be targeted earlier in the life-cycle. Analyzing self-selection and sorting on gains to multiple abilities is important in order to know which types of early interventions are most beneficial.
 
The project has generated four working papers. We summarize the main contribution of each of these four papers in the following four paragraphs. Thereafter, we turn to our ongoing work, impacts of the project, and dissemination.

In the first paper, Humphries, Joensen & Veramendi (2017) we address the question: Does the college major premium reflect returns to prior abilities and skills or college education? We decompose the college major premium into labor market returns to multidimensional abilities (grit, interpersonal, and cognitive), skills learned in 9th grade and high school, and skills learned in college. This allows us to quantify how much of the college major premium is due to sorting on multi-dimensional abilities and how much is due to the differential labor market value of major-specific skills. We find that sorting on abilities accounts for most of the return to enrolling in a college major and about 50% of the return to graduating in a college major. We also provide novel estimates of complementarities and interaction effects between abilities and skills, since the returns to abilities vary significantly across college majors. We document that 40% of students who enter STEM fields change major or drop out. We evaluate counterfactual policies to promote STEM degrees through skill investments at three stages - 9th grade, high school, and college - accounting for the fact that many who start STEM fields do not finish. We find that the labor market return to investing in STEM skills pre-college primarily work indirectly through the increased probability of completing (STEM) college degrees. Earlier investments in cognitive abilities have an even larger impact on degree completion and labor market outcomes.

In the second paper, Facchinello (2017) addresses the question: Does early grading affect educational choices? To answer this question, he exploits the staggered implementation of a reform which postponed grade assignment in Swedish compulsory school. The rich and long panel data enables estimation of both short- and long-term effects of early grading, for students with different academic ability and socioeconomic status (SES). When graded early on, high-ability students (especially if high-SES) perform better, and are more likely to choose advanced courses during compulsory school. Low-ability students react in the opposite way, in particular if low-SES. While high school attainment increases for high-ability low-SES students, college attainment decreases for low-ability low-SES students. None of these effects carry over to the labor market. This suggests that early grades improve the match between early education choices and academic ability, reduce over-investment in education, but exacerbate educational inequality. There is no evidence of demotivating effects for low-ability students, a plausible mechanism through which grades could affect education choices, and the main motivation behind the grading reform. Short-term effects are in line with the predictions of a theoretical model where students learn about their ability from SES and grades.

In the third paper, Facchinello, Joensen & Veramendi (2017) we address the question:  Can surveys affect human capital investments? This paper examines whether individual education choices and outcomes are affected by a survey posing questions related to expectations and forward-looking behavior. We have administrative data for the whole Swedish population to which an extensive education survey was administered to randomly drawn samples of 3rd graders. This constitutes a randomized social experiment for testing whether responding to survey questions alters behavior. We observe complete educational and labor market histories until the individuals are 31-41 years old. We have exogenous variation in the timing of first surveys and when an additional survey was administered to parents. The causal effect of the survey on both short- and long-term outcomes is generally not significantly different from zero. We find, however, that being surveyed increases educational attainment and job stability in the early career for those with low parental education. We also assess heterogeneity in estimated causal effects and sibling spillovers in order to get at potential mechanisms. The patterns indicate the importance of increased awareness.

The fourth paper, Humphries (2017) investigates the causes and consequences of self-employment over the life cycle and evaluates how self-employment decisions can be influenced by policy. The first part of the paper, uses machine learning methods to summarize the patterns of self-employment behavior observed in the data. It reveals that careers involving self-employment fit into a small number of economically distinct groups. Some self-employment spells are short, with minimal capital investment and rapid return to paid employment, while others persist and have substantial capital devoted to the business from the outset. Guided by these descriptive results, he develops and estimates a dynamic Roy model in which self-employment decisions depend on factors such as cognitive and non-cognitive abilities, education, prior work experience, the cost of capital, and other labor market opportunities. The model integrates traditional models of dynamic career choice that feature human capital investment and models of business start-up that feature physical capital investment. The estimated model is used to evaluate policies designed to promote self-employment. Cognitive and non-cognitive abilities, education, and prior work experience are important determinants of the types of businesses individuals start, how much capital they employ, and how long they remain in self-employment. Subsidies that incentivize self-employment are generally ineffective, both in terms of promoting long-lasting firms and in terms of improving the welfare and earnings of those induced to enter self-employment.

In ongoing work, we bring the ideas from the first three papers together to analyze the role of information, intergeneration connections, and early childhood development in order to be able to identify sensitive periods and factors for skill development. Humphries, Joensen & Veramendi (2017) extend Humphries, Heckman & Veramendi (2016) to not only include binary sequential choices of levels of education, but also multinomial choices of fields of education. Our rich measures of abilities also allow us to include not only one latent factor of non-cognitive abilities, but two factors - grit and interpersonal - which both matter at different margins. We are currently also including two cognitive factors - math and language - and preliminary evidence shows that sorting patterns also differ markedly along these two dimensions. In joint work, we are further developing non-linear model specifications of Humphries, Heckman & Veramendi (2016) and non-parametric econometric techniques as detailed in our original project proposal. In addition, we are working on quantifying how relaxing several of the assumptions affects conclusions. Particularly, the importance of dynamics and forward looking behavior. The collaboration on this longer-term project was one important factor for Joensen’s move to the University of Chicago such that the project group can work more closely together without geographic distance and too much time difference, while she is still affiliated with the Stockholm School of Economics as a recurrent visiting researcher and frequently travels back to work with the data.

Facchinello (2017) and Humphries (2017) were both well received on the international job market for Economists as they landed Assistant Professor jobs at Singapore Management University and Yale University, respectively. Facchinello (2017) was awarded the Unicredit & Universities Best Job Market Paper Award (November 2015) and Humphries (2017) was also part of the Review of Economic Studies meetings (ReStud Tour, May 2017) among the eight most promising graduating doctoral students in economics and finance in the world. We have presented the four working papers at many invited seminars; including at Arizona State University, CERGE in Prague, Collegio Carlo Alberto, Cornell University, Harvard Business School, Maastricht University, Northwestern University, Norwegian School of Economics, Princeton University, Purdue University, Singapore Management University, Stanford Graduate School of Business, Stockholm School of Economics (SITE and Department of Economics), University College London, University of Chicago (Department of Economics and Booth School of Business), University of Michigan (CIERS and Department of Economics) and Yale University, as well as international conferences including the annual meetings of the Society of Labor Economists (SOLE) in Raleigh in May 2017 and the Society for Economic Dynamics (SED) in Edinburgh in June 2017. Finally, we also presented two of the papers at the workshop we organized at the Stockholm School of Economics, “Human Capital and Public Policy”, in August 2017. The workshop was organized as a venue to present some of the key project results to experts in the field and also bring them into context and discuss relevant institutional details with local researchers working on the topic. The workshop entailed many enlightening discussions and was also well attended by the PhD students in the Stockholm area.