Education Program Meeting

November 19-20, 2009
Caroline Minter Hoxby, Organizer

Glenn Ellison, MIT and NBER; and Ashley Swanson, MIT
The Gender Gap in Secondary School Mathematics at High Achievement Levels: Evidence from the American Mathematics Competitions

Ellison and Swanson use a new data source, American Mathematics Competitions, to examine the gender gap among high school students at very high achievement levels. The data bring out several new facts. There is a large gender gap that widens dramatically at percentiles above those that can be examined using standard data sources. An analysis of unobserved heterogeneity indicates that there is only moderate variation in the gender gap across schools. The highest achieving girls in the United States are concentrated in a very small set of elite schools, suggesting that almost all girls with the ability to reach high math achievement levels are not doing so.

Ofer Malamud, University of Chicago and NBER; and Abigail Wozniak, University of Notre Dame
The Impact of College Education on Geographic Mobility: Identifying Education Using Multiple Components of Vietnam Draft Risk

Malamud and Wozniak examine whether higher education is a causal determinant of geographic mobility using variation in college attainment induced by draft-avoidance behavior during the Vietnam War. They use national and state-level induction risk to identify both educational attainment and veteran status among cohorts of affected men observed in the 1980 Census. Their 2SLS estimates imply that the additional years of higher education significantly increased the likelihood that affected men resided outside their birth states later in life. Most estimates suggest a causal impact of higher education on migration that are larger in magnitude but not significantly different from OLS. Their large reduced-form estimates for the effect of induction risk on out-of-state migration also imply that the Vietnam War led to substantial geographic churning in the national labor market. They conclude that the causal impact of college completion on subsequent mobility is large and they provide evidence on a range of mechanisms that may be responsible for the relationship between college education and mobility.

Tahir Andrabi, Pomona College; Jishnu Das, The World Bank; and Asim Ijaz Khwaja, Harvard University and NBER
What Did You Do All Day?

Female education levels are very low in many developing countries. Does maternal education have a causal impact on children's educational outcomes even at these very low levels of education? By combining a nationwide census of schools in Pakistan with household data, Andrabi, Das, and Khwaja are able to use the availability of girls' schools in the mother's birth village as an instrument for maternal schooling to address this issue. Since public schools in Pakistan are segregated by gender, the instrument affects only maternal education rather than the education levels of both mothers and fathers. The researchers find that children of mothers with some education spend 75 minutes more on educational activities at home compared to children whose mothers report no education at all. Mothers with some education also spend more time helping their children with school work; the effect is stronger (an extra 40 minutes per day) in families where the mother is likely the primary care-giver. Finally, test scores for children whose mothers have some education are higher in English, Urdu (the vernacular) and Mathematics by 0.24-0.35 standard deviations. The authors find no relationship between maternal education and mother's time spent on paid work or housework - a posited channel through which education affects bargaining power within the household. Nor do they find a relationship between maternal education and the mother's role in educational decisions, or in the provision of other child-specific goods, such as expenditures on pocket money, uniforms, and tuition. The data therefore suggest that at these very low levels of education, maternal education does not substantially affect her bargaining power within the household. Instead, maternal education could directly increase the mother's productivity, or affect her preferences towards the children's education in a context where her bargaining power is low.

Matthew Chingos, Harvard University; and Martin West, Harvard Graduate School of Education
Do More Effective Teachers Earn More Outside of the Classroom?

Chingos and West examine earnings records for more than 90,000 classroom teachers employed by Florida public schools between the 2001-2 and 2006-7 school years, roughly 20,000 of whom left the classroom during that time. A majority of those leaving the classroom remained employed by public school districts. The earnings distribution of former teachers is wider than their earnings while teaching, even when excluding likely part-time workers. Among teachers in grades 4-8 leaving for other industries, a single standard deviation increase in estimated value-added is associated with 3-5 percent higher earnings outside of teaching. High school math, science, and social studies teachers also earn 11-14 percent more after leaving for other industries than do former English teachers. The relationship between both effectiveness and subject-area expertise and earnings appears to be stronger in other industries than for the same groups of teachers while in the classroom, suggesting that current compensation systems do not fully account for the higher opportunity wages of effective teachers and teachers in high-demand subjects.

Philippe Morin, University of Ottawa
Gender and Competition: From the Lab into the Classroom

Evidence from the experimental economics literature suggests that females perform less effectively than males in competitive environments. Morin assesses the external validity of this finding in a regular non-experimental setting: the classroom. The 1997 Ontario Secondary School reform created a "double cohort" of secondary school graduates, drastically increasing the number of university applicants in September 2003. Given the limited number of places available in universities, the quality of accepted students was significantly higher in that year than in previous years, increasing competition for high grades in the classroom. Examining academic performance of the 2001 and 2003 entering cohorts at a large Ontario university, Morin finds that male students coped better with the increased competition than females. In particular, the male university average increased relative to females, as did the proportion of male students graduating "on time". These results emphasize the presence of gender differences in performance under increased competition in important real-life situations; supporting the findings of the experimental economics literature.

Jane Cooley, University of Wisconsin; Salvador Navarro, University of Wisconsin; and Yuya Takahashi, University of Wisconsin, Madison
A Framework for the Analysis of Time-varying Treatment Effects: How The Timing of Grade Retention Affects Outcomes

Cooley, Navarro, and Takahashi develop a method to estimate time-varying treatment effects in situations where dynamic selection into treatment may confound estimates of the treatment effect. In so doing, they draw attention to an important policy tool, the timing of treatment, and the associated challenges with determining how treatment effects vary over time. Their preferred method assumes that the unobservables that jointly determine selection into treatment and the treatment effects can be modeled through a factor structure. Importantly, this allows for heterogeneous treatment effects across unobservable types and recovers a much richer set of counterfactuals than can be recovered with existing methods. The authors then apply their method to the study of grade retention using the Early Childhood Longitudinal Study of Kindergartners. They find dynamic selection and that the effect depends both on the time at which the student is retained and the time elapsed since retention.

Leah Platt Boustan, UC, Los Angeles and NBER
Desegregation and Urban Change: Evidence from City Boundaries

In the early 1970s, courts mandated that many urban school districts outside the South desegregate by race while exempting the surrounding suburbs. Boustan estimates the marginal willingness to pay to avoid desegregation by comparing housing values in city districts, some of which faced court-ordered desegregation, to their adjacent suburbs over the 1970s, focusing on areas close to the school district boundary. The average desegregation plan reduced urban housing prices and rents by 3-7 percent. Aversion to desegregation is due to increased exposure to cross-race peers and to student reassignment to non-neighborhood schools.

Joshua Goodman, Harvard University
The Labor of Division: Returns to Compulsory Math Coursework

Labor economists know that a year of schooling raises earnings but they have little evidence on the impact of specific courses completed. Goodman identifies the impact of math coursework on earnings using the differential timing of state-level increases in high school graduation requirements as a source of exogenous variation. The increased requirements induced large increases in both the completed math coursework and the earnings of blacks, particularly black males. Two-sample instrumental variable estimates suggest that each additional year of math raised blacks' earnings by 5-9 percent, accounting for a large fraction of the value of a year of schooling. Closer analysis suggests that much of this effect comes from black students who attend non-white schools and who will not attend college. The earnings impact of additional math coursework is robust to changes in empirical specification, is not driven by selection into the labor force, and persists when earnings are conditioned on educational attainment. The reforms close one fifth of the earnings gap between black and white males. Estimates for whites are similar to those for blacks but are much noisier because of the reforms' weaker impact on white students' coursework. These results suggest that math coursework is an important determinant of the labor market return to schooling, that simple minimum requirements largely benefit low-skilled students, and that more demanding requirements might be necessary to improve the outcomes of high-skilled students.

Winnie Chan; and Robert McMillan, University of Toronto and NBER
School Choice and Public School Performance

Chan and McMillan measure the effects of increased competition from private schools on public school performance. They take advantage of the clear exogenous increase in choice associated with Ontario's 2002 tuition tax credit, which eased access to private schools throughout Canada's most populous province, as well as the exogenous reduction at the time the policy was unexpectedly cancelled. Building on the idea that the policy would have differential competitive effects for public schools in districts with a significant private school presence versus those without, the authors construct a measure of the increase in competition that is specific to each school, combining the province-wide reform with local variation in private school availability. Relating this measure to school outcomes, their results indicate that public school performance improved for schools facing the greatest competitive pressures following the introduction of the policy, controlling for a host of other relevant factors; when the policy was switched off, such gains were eliminated. The estimates imply that a single standard deviation increase in private school competition generates about one-tenth of a standard deviation improvement in average public school performance. A positive relationship of similar magnitude persists when accounting for fixed unobservables using the within-school gain, and also when instrumenting for private school presence. To assess whether the positive effect is primarily due to increases in productivity, the analysis accounts for a series of alternative mechanisms. Increased sorting is unlikely to be responsible for the public school improvement because there were no significant outflows of Ontario public school students to the private system; and very slight compositional changes in observable public school student characteristics suggest that unobservable changes also are likely to have been minor . There is no evidence of any significant "gaming" on the part of public schools, nor of differential changes in school resources that might explain performance differences. Overall, the Ontario evidence supports the view that increased competition from private schools raises public school productivity, of benefit to students remaining in the public system.

Christopher Avery, Harvard University and NBER; and Sarah Turner, University of Virgina and NBER
Playing the College Admissions Game: Critical Moves and the Link to Socio-Economic Circumstances

High-achieving students from families with low incomes are less likely to enroll at selective private colleges or at public flagship colleges than their more affluent peers. Using original survey data for high school seniors from Virginia during the college admissions cycle, Avery and Turner explore the systematic differences in student application patterns related to family circumstances. They find little evidence to support explanations related to differences in admission probabilities or differences in preferences by family circumstances. Instead, low-income students have less information through the application process than do high-income students and this lack of information is connected to application choices students in the study.

Rema Hanna, Harvard University and NBER; and Leigh Linden, Columbia University
Measuring Discrimination in Education

Hanna and Linden illustrate a methodology for measuring discrimination in educational contexts. They ran an exam competition in India in which children competed for a large financial prize. They recruited teachers to grade the exams and then randomly assigned child "characteristics" (age, gender, and caste) to the cover sheets of the exams - this was to ensure that there was no systematic relationship between the characteristics observed by the teachers and exam quality. The researchers find that teachers give to exams that are assigned to be "lower caste" scores that are about 0.03 to 0.09 standard deviations lower than exams that are assigned to be "high caste." The effect is small relative to the real differences in scores between the high and lower caste children. Low-performing, low caste children and top-performing females tend to lose out the most because of discrimination. Interestingly, the authors find that the discrimination against low caste students is driven by low caste teachers, while teachers who belong to higher caste groups do not appear to discriminate at all. This result runs counter to the previous literature, which tends to find that individuals discriminate in favor of members of their own groups.

NBER Videos

National Bureau of Economic Research, 1050 Massachusetts Ave., Cambridge, MA 02138; 617-868-3900; email:

Contact Us