How Does Academic Leniency Affect Student Effort, Achievement, and Long-Run Human Capital?

Bowden, Rodriguez, and Weingarten (2025) examine whether relaxing high-school grading standards reduces student effort and learning. They use statewide administrative data from North Carolina (2013–2019), linking exact birthdates, course grades, absences, and ACT scores. They find that the shift to a more lenient 10-point scale mechanically raised GPAs by about 0.13 points (roughly 4.8%) but caused meaningful declines in effort: absences rose 22% (about 1.3 days) and numeric grades in math and English fell by about 2 points. Longer-run effects show ACT scores dropping by 0.5 points (2.4%), with the strongest negative impacts among lower-ability students.

Why This Article Was Selected for The Policy Scientist

This article addresses a central policy issue with broad relevance: how shifts in academic standards shape long-run human capital. At a time when rising GPAs coincide with stagnant test scores and declining college enrollment, the question is especially timely. This study contributes meaningfully by isolating the behavioral mechanisms behind grade inflation. The administrative data are unusually detailed and well suited for careful measurement of effort and achievement. The causal design—a fuzzy difference-in-discontinuity framework—is considerably stronger than traditional regression approaches, and the resulting estimates appear credible and potentially generalizable to other states with similar grading reforms.

Full Citation and Link to Article

Bowden, A. B., Rodriguez, V., & Weingarten, Z. (forthcoming 2025). The unintended consequences of academic leniency. American Economic Journal: Economic Policy.  https://www.aeaweb.org/articles?from=f&id=10.1257/pol.20240582.

Central Research Question

The authors ask whether relaxing high-school grading standards—via North Carolina’s shift from a 7-point to a 10-point grading scale—alters student effort, academic performance, and long-run human capital accumulation. They also ask whether these effects differ by prior academic ability. The policy created a statewide, abrupt reduction in the numeric threshold for all letter grades. The central question, therefore, is whether easier grading standards produce mechanical improvements in GPA but unintended behavioral responses that ultimately depress learning and widen achievement gaps.

Previous Literature

The study builds on a long-standing body of work examining how grading standards shape effort, achievement, and longer-run outcomes. Prior studies emphasize that grading schemes convey information about performance and may create incentives that meaningfully alter student behavior. Hvidman and Sievertsen (2021), studying a Danish reform that recoded grades, show that downward grade shocks can increase subsequent effort and achievement, especially among high-ability students. Similarly, Butcher, McEwan, and Weerapana (2024) find that a pass/fail policy at Wellesley reduced numeric grades, consistent with reduced student effort, especially among students with weaker math skills. Earlier teacher-level research (Figlio and Lucas, 2004) finds that stricter teachers induce higher downstream test-score gains, particularly for high-ability students. This work collectively suggests that students adjust their effort in response to grading systems, but prior studies often cannot directly observe effort, cannot isolate statewide shocks, or cannot separate mechanical grade changes from behavioral responses.

This paper advances the literature by directly identifying behavioral mechanisms—particularly school attendance—as the channel linking grading standards to achievement. It does so using a quasi-experimental statewide policy that shifts grading leniency for all high-school courses simultaneously, allowing for clearer causal interpretation than earlier work relying on gradual or teacher-level variation. The authors also contribute by documenting heterogeneous treatment effects, showing that students of different ability levels respond in systematically different ways.

Data

The authors use administrative data on the universe of North Carolina public-school students from 2013 to 2019. These longitudinal records include detailed information on demographics, transcript-level grades, numeric course marks, attendance, and standardized test performance. Crucially, the authors supplement these data with restricted-access birthdate information from the North Carolina Department of Public Instruction, enabling precise assignment relative to the kindergarten entry cutoff—a key element of their identification strategy.

The dataset includes Math I end-of-course (EOC) exam scores, ACT scores for all eleventh graders, annual GPA in core subjects, and complete attendance records. Students with disabilities and those in charter schools are excluded to ensure comparability in grading practices and coursework requirements. Approximately 90% of students have valid eighth-grade math test scores, which the authors use as a pre-treatment measure of ability to examine heterogeneous effects.

Overall, the dataset is unusually rich. It captures both behavioral inputs (attendance) and academic outputs (numeric grades, standardized test scores), allowing the authors to distinguish mechanical GPA changes from genuine learning.

Methods

The evaluation uses a fuzzy difference-in-discontinuity design. This design exploits two sources of quasi-random variation: (1) the statewide grading reform, implemented sharply between academic years, and (2) North Carolina’s kindergarten entry cutoff of October 17. Students born just before or after this cutoff enter high school in different years. Thus, students near this threshold are similar in age and background but differ in whether they experience the old or new grading scale upon entering ninth grade.

Because not all families perfectly comply with the enrollment cutoff—some redshirt or delay entry—the authors use a fuzzy regression discontinuity instrument that relies on exact birthdates to predict grade cohort. They embed this into a difference-in-discontinuity framework, comparing discontinuities across pre-policy and post-policy cohorts and controlling for age effects, demographic characteristics, and potential nonlinearities in the running variable.

This approach is notably stronger than conventional regression-based observational studies. It approximates a causal design by near-random assignment of students to differing grading regimes and provides credible local average treatment effects. The authors perform robustness checks including alternative bandwidths, placebo tests, covariate smoothness tests, and functional-form sensitivity analyses, all of which support the validity of their estimates.

Findings/Size Effects

The grading reform mechanically increased GPAs, but it simultaneously reduced student effort and produced no improvement in learning. For all ninth graders, the policy raised GPA by about 0.127 points, or 4.8%. However, numeric course grades fell: math grades declined by roughly 1.7 points, and English grades declined by a similar margin. These reductions suggest that teachers responded to the policy by altering grading practices, or that students exerted less effort, or both.

The clearest evidence of reduced effort comes from absences. The reform increased ninth-grade absences by 1.3 days on average—a 22% increase—and raised chronic absenteeism by 0.3 percentage points. These changes, unlike GPA, cannot be mechanically induced by the grading scale; they reflect genuine behavioral shifts. Math I EOC scores did not increase, indicating no measurable short-run gains in learning.

Heterogeneous effects reveal that low and medium-ability students experienced the largest reductions in effort. Absences rose by 1.6 to 2.1 days for these groups, while high-ability students showed essentially no change. Despite larger expected mechanical GPA gains for lower-ability students, actual GPA increases were similar across ability groups—suggesting that reduced effort offset much of the mechanical increase in GPA among weaker students.

Longer-run outcomes show compounding negative effects. Absences continued to rise in tenth and eleventh grade, reaching increases of 2.9 days in eleventh grade. GPA gains dissipated: by tenth grade, GPA fell below pre-policy levels, and by eleventh grade students experienced net GPA losses. ACT scores declined by roughly 0.5 points (2.4%), with the largest reductions among low- and medium-ability students. College intentions also fell by about 6.6 percentage points, though high-school graduation rose modestly (3.3 percentage points), likely because the lenient scale made passing courses easier.

Together, these findings show that the policy widened achievement gaps. Higher-ability students responded little to the increased leniency, while lower-ability students reduced effort substantially, producing longer-run deficits in human capital accumulation.

Conclusion

This study provides strong evidence that relaxing grading standards produces unintended behavioral responses that undermine academic achievement, particularly for students already at risk of lower performance. The statewide reform generated mechanical increases in GPA but reduced effort, increased absenteeism, lowered numeric grades, and produced measurable declines in ACT scores and college intentions. Effects were strongly heterogeneous: lower-ability students were most affected, while their higher-ability peers saw minimal behavioral changes.

Because the identification strategy isolates causal effects using sharp statewide variation and a robust quasi-random assignment mechanism, the estimates credibly inform policy discussions regarding grade inflation, standards-based reforms, and educational inequality. The results imply that academic leniency, while superficially increasing measured attainment, may depress actual learning and widen disparities in human capital formation.