The Hidden Value of Elite Schools: A New Model of Curriculum and Student Outcomes

In a groundbreaking study released by the National Bureau of Economic Research (NBER), economists Glenn Ellison and Parag Pathak from the Massachusetts Institute of Technology (MIT) present a compelling new framework for understanding how school systems and curricula can be optimally designed. Their paper challenges widely held assumptions about selective schools and raises important questions about how school effectiveness is measured. In particular, the researchers argue that widely used metrics like standardized test scores may fail to capture the true value added by elite institutions such as Boston Latin School (BLS), which emphasize advanced content not always tested in standard assessments.

Beyond the Test Score Illusion

The study centers on a puzzle in education research: why do students who barely gain admission to elite exam schools like BLS not perform significantly better on standardized tests than those who just miss the cutoff? Conventional interpretations have taken this to mean that selective schools offer no real benefit. But Ellison and Pathak offer an alternative explanation, one rooted in a “curriculum-matching” model. In this model, schools are seen as making strategic curriculum decisions based on the academic abilities of their student bodies. When schools tailor their instruction to match student aptitude, they may focus on advanced skills that standardized exams fail to measure. As a result, the absence of test score jumps at the admissions cutoff could indicate not ineffectiveness, but a highly efficient system that prioritizes skills aligned with long-term student success.

Modeling Schools as Curriculum Designers

The authors develop a formal model in which schools allocate instructional time across a range of skills, with the goal of maximizing student achievement. Students have different learning styles and abilities, and the curriculum must be matched accordingly. When this matching is done optimally, assigning high-ability students to schools that focus on more advanced material, student outcomes improve. Yet, crucially, these improvements might not show up in traditional standardized tests if those tests emphasize basic skills. The model predicts no discontinuity in test scores at admissions cutoffs, but it does forecast changes in the slope of the performance curve, with higher-achieving students benefiting more over time. It also predicts performance differences on specific types of questions and across different types of exams.

Real Evidence from a Real School

To test the theory, the researchers examined seventeen years of data from applicants to Boston Latin School, Boston’s most selective public school. They focused on students applying for entry in the 7th grade, which is the main entry point for BLS, and tracked their academic outcomes through high school and into college. The results confirm earlier findings that there is no significant jump in Grade 10 MCAS scores at the BLS admissions cutoff. However, new evidence from the study points strongly in favor of the curriculum-matching model. In particular, there is a clear increase in the slope of the relationship between entrance exam scores and later performance on SAT and PSAT English tests at the admissions threshold, suggesting that students above the cutoff benefit more from the advanced curriculum.

The researchers also looked at student preferences and found that even students below the cutoff overwhelmingly list BLS as their top choice, with preference increasing steadily with entrance scores. This behavior, hard to reconcile with a zero value-added model, is entirely consistent with a framework where parents understand that BLS provides value for higher-performing students.

Hidden Jumps in Individual Test Questions

Perhaps the most fascinating evidence comes from a granular analysis of individual MCAS questions. While overall scores may appear flat at the cutoff, several specific questions showed dramatic jumps, both upward and downward, in performance. Some questions testing middle school-level material saw BLS students perform worse, likely because those topics were deprioritized in favor of more advanced content. Other questions involving higher-order algebra and geometry saw better performance from BLS students, reflecting the school’s emphasis on college preparatory skills. This finding is crucial because it suggests that selective schools teach different material, not necessarily better or worse, and standard tests may not be designed to capture this nuance.

Advanced Coursework and College Impact

The study also investigates participation in Advanced Placement (AP) exams and long-term college outcomes. Admission to BLS significantly increases the number of AP exams taken and passed, especially in core academic areas like calculus and literature. These increases are particularly pronounced in the most popular AP subjects, where BLS students far outnumber their peers in both participation and success. Though college graduation outcomes showed positive effects for BLS students, the results were not statistically significant, partly due to sample size limitations. Nevertheless, the evidence on AP performance strongly supports the idea that BLS delivers value in areas that align more closely with college readiness and long-term educational goals.

Ultimately, the study urges educators and policymakers to reconsider how school quality is measured and how curriculum design can impact student outcomes in ways that go beyond test scores. Ellison and Pathak’s findings suggest that current metrics may underestimate the benefits of selective schools, especially for high-ability students. Rather than focusing narrowly on standardized test outcomes, evaluation systems must account for how well a school’s curriculum aligns with both the abilities of its students and the broader goals of education. In doing so, we may better understand the real contributions of elite institutions and design policies that support smarter curriculum matching across all schools.