Altonji and Zhu estimate causal effects of specific graduate degrees, such as an MBA or an MS in Electrical Engineering, on labor market outcomes. Moreover, they study how college major and characteristics of students and graduate schools influence the payoff to graduate education. The researchers use alternative fixed effect regression models to control for endogenous selection into graduate programs and in addition use propensity score weighting to construct suitable control groups. Altonji and Zhu use a version of Dale and Krueger's strategy to estimate differences across schools in the value of specific degrees. Their analysis takes advantage of the size and richness of the Texas Administrative data, and the fact that it can be used to track students through high school, college, graduate school and the labor market.
Li, Linde, and Shimao propose an easily-computable measure named Major Complexity Index (MCI) that captures the latent skills taught in different majors and that are required by different occupations. By applying the method-of-reflection to the major-to-occupation network, the researchers construct a scalar measure of the relative complexity of majors. Their measure provides strong explanatory power of average earning and employment variations across majors. Further evidence suggests that MCI is strongly related to advanced skills such as critical and analytical thinking. Li, Linde, and Shimao also provide a two-stage algorithm to partial out selection on observables, which opens up possibilities of applying the complexity measure in various contexts.
This workshop will be dedicated to advances in experimental economics combining laboratory and field-experimental methodologies with theoretical and psychological insights on decision-making, strategic interaction and policy. Brown, Kaur, Kingdon, and Schofield would invite papers in lab experiments, field experiments and their combination that test theory, demonstrate the importance of psychological phenomena, and explore social and policy issues. In addition to senior faculty members, invited presenters will include junior faculty as well as graduate students.
Books shape how children learn about society and social norms, in part through the representation of different characters. To better understand the messages to which children have been exposed, Adukia, Eble, Harrison, Runesha, and Szasz introduce new machine-led methods for systematically converting images into data, along with using established text analysis methods. They apply these tools to measure the representation of race, gender, and age in children's books commonly found in US schools and homes over the last century. The researchers find that books selected to highlight people of color, or females of all races, have increasingly over time depicted characters with darker skin tones; whereas "mainstream" books over the last two decades have increasingly depicted chromatically ambiguous characters with an increase in lighter skin tones. Children are consistently depicted with lighter skin than adults, despite no systematic differences in skin tones by age. Comparing images and text, Adukia, Eble, Harrison, Runesha, and Szasz find that females are more represented in images than in text. There is a persistent disproportionate representation of males, particularly White males, and lighter-skinned people relative to darker-skinned people. Their data provide a view into the "black box" of education through children's books in US schools and homes, highlighting what has changed and what has endured, and the tools Adukia, Eble, Harrison, Runesha, and Szasz introduce help reveal the path toward more equitable representation.
Public schools face substantial constraints on when and whom they can hire. Jones, Kraft, Papay, and Wedenoja study Boston Public Schools' (BPS) efforts to reduce these frictions by eliminating seniority privileges and forced placements as well as accelerating the open hiring timeline. Using a difference-indifference framework where some BPS schools enjoyed hiring autonomies prior to the reforms, the researchers find that mutual consent hiring moved up the median hire date by more than two months and increased the experience, diversity, effectiveness, and retention of both internal and external new hires. These findings illustrate the large returns to conducting early, open searches in the public teacher labor market.
Income share agreements (ISAs), which cover college attendance costs in exchange for claims on future income, may attract students with poor earnings potential (adverse selection) or distort labor supply among enrollees (moral hazard). In this study, Herbst, Palacios, and Yannelis conduct a field experiment to separately test for moral hazard and adverse selection by varying the contract terms offered by a large ISA provider in South America. Specifically, the researchers randomly offer existing ISA enrollees the opportunity to lower their pledged income shares by 5 or 10 percentage points in exchange for flat monthly fees ranging from 1,000 to 70,000 pesos. To identify moral hazard, Herbst, Palacios, and Yannelis estimate the treatment effect of accepting a contract with a lower income-share obligation, using contract offers as instruments. To identify adverse selection, they compare individuals who received different menus of offers but ultimately pledged the same income share. Estimating treatment effects and selection patterns in earnings, repayment, and risk factors enables us to quantify the welfare losses associated with information asymmetries in equity-like contracts and helps inform the debate over how to fairly and efficiently finance human capital investments.
Higher education can play a crucial role for innovation and growth by by providing people with the frontier knowledge that is necessary to innovate. This paper proposes a novel approach to quantify the provision of this type of knowledge across and within schools: the education innovation gap, a measure of the distance between the content of a university course and frontier knowledge. Biasi and Ma calculate the gap for 1.75 million courses taught at US institutions by mapping the text of their syllabi to the text of 20 million old and new academic publications. The researchers show evidence of substantial variation in the gap, both across and within schools. In particular, the gap is smaller in more selective schools and those serving fewer disadvantaged or minority students. The gap of a given course significantly declines when a new instructor takes over the course, and more research-active instructors teach courses with smaller gaps. Importantly, these differences bear a strong relationship with students' graduation rates and incomes after graduation, especially in less selective schools and those serving more disadvantaged students.
The college experience involves much more than credit hours and degrees. Students likely derive utility from in-person instruction and on-campus social activities. Quantitative measures of the value of these individual components of the college experience have been hard to come by. Leveraging the COVID-19 shock, Aucejo, French, and Zafar elicit students’ intended likelihood of enrolling in higher education under different costs and possible states of the world. These states, which would have been unimaginable in the absence of the pandemic, vary in terms of class formats (i.e., in-person vs. remote instruction) and restrictions to campus social life. Aucejo, French, and Zafar show how such data can be used to recover student’s willingness-to-pay (WTP) for college-related activities in the absence of COVID-19, without parametric assumptions on the underlying heterogeneity in WTP. Their data come from a module that Aucejo, French, and Zafar fielded to approximately 1,500 undergraduate students at one of the largest public universities in the United States. The researchers find that the WTP for in-person instruction (relative to a remote format) represents around 4.2% of the average annual net cost of attending university, while the WTP for on-campus social activities is 8.1% of the average annual net costs. Aucejo, French, and Zafar also find large heterogeneity in WTP, which varies systematically across socioeconomic groups. Their analysis shows that economically-disadvantaged students derive substantially lower value from university social life, but this is primarily due to time and resource constraints. Aucejo, French, and Zafar also validate their approach by showing that the heterogeneity in the estimated WTPs is systematically correlated with students' actual enrollment decisions in the subsequent semester. Beyond providing an explanation for why college persistence rates may differ by socioeconomic background, their results have implications for how college costs should be structured.
This paper was distributed as Working Paper 28511, where an updated version may be available.
In 2013/14 the arrangements for setting teacher pay in England were radically reformed. A system of seniority-based progression was replaced by a decentralised system allowing schools to pay teachers based on individual performance and local labour market conditions. Using a data-driven strategy, Anders, Bryson, Horvath, and Nasim classify schools into four groups according to what form of flexibility they adopted, then use this classification to analyse how pay flexibility affected teachers’ pay and retention. About 9% of schools speeded up pay progression relative to what would have been expected in the seniority-based scheme, while about twice as many slowed it down. Teachers in these schools saw wages increase (decrease) by 4 (3) per cent on average. These wage changes seem to have little effect on teacher retention, suggesting school-specific labour supply in England is highly inelastic, implying large monopsony power for schools.
Most college-going students believe that the Advanced Placement (AP) program offers a pathway to selective colleges. This paper presents evidence on this assumption with data on over 1,800 students that Conger, Long, and McGhee randomly assigned the offer of enrollment in an AP science course. Results suggest that taking an AP science class has no substantial impacts on students' plans to enroll in college and no impact on college entrance exam scores. Yet students taking AP science have very low pass rates on the AP exam and a lower probability of enrolling in a competitive college. Exploratory analyses suggest that the negative treatment effect is found only among students who take the AP course in their senior year and may be driven by reduced aspirations as a result of AP exam failure.
Tincani, Kosse, and Miglino study the incentive effects of preferential admissions on pre-college study effort, when students do not hold correct beliefs about their college admission chances. The researchers exploit a unique Government-backed randomized experiment in Chile: in 2016, high-schools with low college admission rates and a high share of disadvantaged students were randomly assigned to participate in a policy that guarantees a college admission to the top 15 percent of students in the school. Tincani, Kosse, and Miglino collected data in the schools that participated in this experiment and merged them with detailed administrative records to build a longitudinal dataset with information on educational outcomes, choices and beliefs. Tincani, Kosse, and Miglino find that the policy increased enrollment by 35 percent, but it decreased achievement (-0.10 st. dev.) and study effort (-0.08 st. dev.) prior to college admission. They show that high-school students' subjective beliefs about their school rank and about their admission chances differ from objective values in a way that is consistent with these policy impacts. Using simulations from an estimated dynamic model that incorporates subjective beliefs, Tincani, Kosse, and Miglino show that the endogenous effort response compressed the policy impacts on enrollment: had students not altered their study effort, the policy impacts on enrollment would have been higher by 13.6 percent, because the policy had negative admission and enrollment effects on some students who lowered their study effort in the incorrect anticipation of a guaranteed admission. When students hold biased beliefs, admission policies aimed at fostering upward mobility of disadvantaged students can harm some of the students they intend to help.
A central point in the market design case for the use of strategyproof assignment mechanisms in school choice is that they relieve applicants of the need to strategize on the basis of beliefs about admissions chances. This paper shows that beliefs about admissions chances shape choice outcomes even when the assignment mechanism is strategyproof by influencing the way applicants search for schools, and that "smart matching platforms" providing live feedback on admissions chances help applicants search more effectively. Motivated by a model in which applicants engage in costly search for schools and over-optimism can lead to under-search, Arteaga Ossa, Kapor, Neilson, and Zimmerman use data from a large-scale survey of choice participants in Chile to show that learning about schools is hard, that beliefs about admissions chances guide the decision to stop searching, and that applicants systematically underestimate nonplacement risk. Arteaga Ossa, Kapor, Neilson, and Zimmerman then use RCT and RD research designs to evaluate live feedback policies at scale in the Chilean and New Haven choice systems. They find that 22% of applicants submitting applications where risks of nonplacement are high respond to warnings by adding schools to their list, reducing their nonplacement risk by 58%. These results replicate across settings and over time. Arteaga Ossa, Kapor, Neilson, and Zimmerman conclude that reducing the strategic burden of school choice requires not just strategyproofness inside the centralized system, but also choice supports for the strategic decisions that inevitably remain outside of it.