Class-size reduction is among the most expensive standard interventions in compulsory schooling, and the most credible quasi-experimental evidence on its long-run effects points in opposite directions in two neighboring, institutionally similar countries. Exploiting Sweden's maximum class-size rule, Fredriksson, Öckert and Oosterbeek (2013) find that smaller classes in upper primary school raise cognitive and noncognitive skills, completed education, and adult wages. The same design applied to Norwegian administrative registries yields precise nulls (Leuven and Løkken 2020; Falch, Sandsør and Strøm 2017). The mechanisms behind class-size effects — when and why class size is effective — remain unknown, and the Norwegian–Swedish contrast survives tests of parental-background heterogeneity, school size, competitive pressure, teacher quality, and local fiscal and labor-market conditions. The objective is to identify the institutional, cultural, or behavioral channels through which class size affects student skills and adult outcomes, and to explain why the effect emerges in some settings but not others.

Background

Economists model schooling as an education production function that maps school inputs — teachers, peers, and class size — into students' skills and, ultimately, their adult earnings. Class size is the input policymakers most often manipulate: class-size reduction programs have been implemented in the United States, the United Kingdom, the Netherlands and elsewhere, at large fiscal cost. Yet decades of research have not produced a consensus estimate of its effect. The core identification problem is that class size is not random: parents, schools and education authorities sort children into schools and classes in ways correlated with their outcomes. The most credible studies therefore exploit exogenous variation, such as the Tennessee STAR experiment (a large-scale U.S. experiment that randomly assigned kindergarten pupils to small or regular-size classes; Krueger 2003, Chetty et al. 2011) or 'maximum class size rules' of the kind used in Israel (Angrist and Lavy 1999) and Scandinavia — administrative rules that force a class to split when enrollment crosses a threshold, generating abrupt, enrollment-driven changes in class size that can be used in a regression-discontinuity / instrumental-variables design (a design that compares students just above and below the enrollment cutoff, where class size differs but students are otherwise comparable).

Applied to long-run outcomes, this design family has produced a sharp Scandinavian contrast. Fredriksson, Öckert and Oosterbeek (2013) exploited Sweden's rule (classes capped at 30 pupils, with splits at enrollment thresholds of roughly 30, 60 and 90 pupils in grade 4) in administrative registries and found that smaller classes in grades 4–6 improve cognitive and noncognitive ability at age 13 (noncognitive ability meaning skills such as motivation, self-discipline and social behavior, as distinct from test-measured cognitive ability), achievement at 16, completed education, and wages and earnings at ages 27–42, with effects concentrated among children from lower-income families and large enough to pass a cost–benefit test. Their companion work shows that families and teachers respond to class size — higher-income parents increase homework help when classes grow, some parents move their child to another school district, and teachers in larger classes delegate more responsibility to pupils — so behavioral responses form part of the production mechanism (Fredriksson, Öckert and Oosterbeek 2016).

Leuven and Løkken (2020) applied the analogous Norwegian rule (maximum class size 28 in primary school and 30 in middle school, effective through the 2002/03 school year) to Norwegian registries following cohorts to age 48. They find no beneficial effects of smaller classes on completed schooling or earnings: their most precise estimates rule out gains from a one-pupil reduction larger than about 0.12 percent (primary school) and 0.15 percent (middle school) in income. The null is mirrored by independent Norwegian evidence on short-run achievement (Leuven, Oosterbeek and Rønning 2008) and on long-run outcomes (Falch, Sandsør and Strøm 2017). Leuven and Løkken show the contrast with Sweden is not an artifact of population differences in parental background, school size, or competitive pressure, and conclude that the effect of class size 'is not a structural parameter': little is known about when and why class size is effective, and they call for opening the 'black box' of education production. Cross-country comparisons had already documented large international heterogeneity in short-run class-size effects (Wößmann and West 2006), and a re-analysis of the original Israeli evidence found precisely estimated nulls in newer data despite the same rule (Angrist, Lavy, Leder-Luis and Shany 2019). The Norwegian–Swedish contrast is thus the clearest instance of a general puzzle: the same institutional device produces different effects in different school systems, and the channels mediating those effects are unidentified.

Problem Statement

Identify the mechanisms — the institutional, cultural, or behavioral channels — through which class size in compulsory school affects students' skills and adult outcomes, to the point of explaining when and why class-size reduction is effective; in particular, explain why near-identical quasi-experimental designs based on maximum class-size rules in Scandinavian administrative registries yield economically meaningful positive long-run effects on completed education and earnings in Sweden but precise null effects in Norway. Identify specific mediating channels — for example teacher behavior and classroom practices, parental responses, peer composition and classroom disruption, or school organization — with credible causal evidence, whether assembled from the cited quasi-experimental literature or from original analysis, that these channels mediate the class-size effect, and show that the identified channels account for the observed cross-country divergence, where a directional argument grounded in that causal evidence suffices and a quantitative decomposition is not required, including why the candidate mediators already tested in the literature (parental-background heterogeneity, school size, competitive pressure, teacher quality, school-district fiscal conditions, local labor-market conditions) failed to reconcile the two bodies of evidence. Non-linear or threshold effects of class size, and effects on non-cognitive skills or other human-capital dimensions not captured by test scores, schooling and earnings, are within scope insofar as they bear on the mechanism.

References

  1. Edwin Leuven and Sturla A. Løkken (2020), 'Long-Term Impacts of Class Size in Compulsory School', Journal of Human Resources 55(1):309–348. DOI: 10.3368/jhr.55.2.0217.8574r2. https://doi.org/10.3368/jhr.55.2.0217.8574r2
  2. Peter Fredriksson, Björn Öckert, and Hessel Oosterbeek (2013), 'Long-Term Effects of Class Size', Quarterly Journal of Economics 128(1):249–285. DOI: 10.1093/qje/qjs048. https://doi.org/10.1093/qje/qjs048
  3. Torberg Falch, Astrid Marie Jorde Sandsør, and Bjarne Strøm (2017), 'Do Smaller Classes Always Improve Students' Long-run Outcomes?', Oxford Bulletin of Economics and Statistics 79(5):654–688. DOI: 10.1111/obes.12161. https://doi.org/10.1111/obes.12161
  4. Peter Fredriksson, Björn Öckert, and Hessel Oosterbeek (2014), 'Inside the Black Box of Class Size: Mechanisms, Behavioral Responses, and Social Background', IZA Discussion Paper No. 8019 (March 2014) / SSRN working paper. DOI: 10.2139/ssrn.2409541. https://docs.iza.org/dp8019.pdf
  5. Peter Fredriksson, Björn Öckert, and Hessel Oosterbeek (2016), 'Parental Responses to Public Investments in Children: Evidence from a Maximum Class Size Rule', Journal of Human Resources 51(4):832–868. DOI: 10.3368/jhr.51.4.1114-6779r1. https://doi.org/10.3368/jhr.51.4.1114-6779r1
  6. Edwin Leuven, Hessel Oosterbeek, and Marte Rønning (2008), 'Quasi-experimental Estimates of the Effect of Class Size on Achievement in Norway', Scandinavian Journal of Economics 110(4):663–693. DOI: 10.1111/j.1467-9442.2008.00556.x. https://doi.org/10.1111/j.1467-9442.2008.00556.x
  7. Joshua D. Angrist and Victor Lavy (1999), 'Using Maimonides' Rule to Estimate the Effect of Class Size on Scholastic Achievement', Quarterly Journal of Economics 114(2):533–575. DOI: 10.1162/003355399556061. https://doi.org/10.1162/003355399556061
  8. Joshua D. Angrist, Victor Lavy, Jetson Leder-Luis, and Adi Shany (2019), 'Maimonides' Rule Redux', American Economic Review: Insights 1(3):309–324. DOI: 10.1257/aeri.20180120. https://doi.org/10.1257/aeri.20180120
  9. Ludger Wößmann and Martin R. West (2006), 'Class-Size Effects in School Systems Around the World: Evidence from Between-Grade Variation in TIMSS', European Economic Review 50(3):695–736. DOI: 10.1016/j.euroecorev.2004.11.005. https://doi.org/10.1016/j.euroecorev.2004.11.005
  10. Raj Chetty, John N. Friedman, Nathaniel Hilger, Emmanuel Saez, Diane Whitmore Schanzenbach, and Danny Yagan (2011), 'How Does Your Kindergarten Classroom Affect Your Earnings? Evidence from Project STAR', Quarterly Journal of Economics 126(4):1593–1660. DOI: 10.1093/qje/qjr041. https://doi.org/10.1093/qje/qjr041
  11. Alan B. Krueger (2003), 'Economic Considerations and Class Size', Economic Journal 113(485):F34–F63. DOI: 10.1111/1468-0297.00098. https://doi.org/10.1111/1468-0297.00098
  12. Trine Filges, Christoffer Scavenius Sonne-Schmidt, and Bjørn Christian Viinholt Nielsen (2018), 'Small Class Sizes for Improving Student Achievement in Primary and Secondary Schools: A Systematic Review', Campbell Systematic Reviews 14(1). DOI: 10.4073/CSR.2018.10. https://doi.org/10.4073/CSR.2018.10
  13. Matej Opatrny, Tomas Havranek, Zuzana Irsova, and Milan Ščasný (2025), 'Publication Bias and Model Uncertainty in Measuring the Effect of Class Size on Achievement', Journal of Labor Economics (advance online publication, August 2025). DOI: 10.1086/737989. https://doi.org/10.1086/737989
  14. Marie Connolly and Catherine Haeck (2022), 'Nonlinear Class Size Effects on Cognitive and Noncognitive Development of Young Children', Journal of Labor Economics 40(S1):S341–S382. DOI: 10.1086/718328. https://doi.org/10.1086/718328
  15. Maximilian Bach and Stephan Sievert (2025), 'Birth Cohort Size Variation and the Estimation of Class Size Effects', Journal of Human Resources 60(2):578–606 (published online August 2024). DOI: 10.3368/jhr.0919-10430r2. https://doi.org/10.3368/jhr.0919-10430r2
  16. Hirotake Ito, Makiko Nakamuro, and Shintaro Yamaguchi (2020), 'Effects of Class-Size Reduction on Cognitive and Non-cognitive Skills', Japan and the World Economy 53:100977. DOI: 10.1016/j.japwor.2019.100977. https://doi.org/10.1016/j.japwor.2019.100977
  17. Graham McKee, Katharine R. E. Sims, and Steven G. Rivkin (2015), 'Disruption, Learning, and the Heterogeneous Benefits of Smaller Classes', Empirical Economics 48(3):1267–1286 (published online July 2014). DOI: 10.1007/S00181-014-0810-1. https://doi.org/10.1007/S00181-014-0810-1
  18. Cuiping Schiman and Jeffrey C. Schiman (2025), 'Class Size, Student Disruption, and Academic Achievement', The B.E. Journal of Economic Analysis & Policy 25(3):607–623. DOI: 10.1515/bejeap-2024-0181. https://doi.org/10.1515/bejeap-2024-0181
  19. Anne Brink Nandrup (2016), 'Do Class Size Effects Differ Across Grades?', Education Economics 24(1):83–95 (published online November 2015). DOI: 10.1080/09645292.2015.1099616. https://doi.org/10.1080/09645292.2015.1099616
  20. Karl Fritjof Krassel and Eskil Heinesen (2014), 'Class-Size Effects in Secondary School', Education Economics 22(4):412–426. DOI: 10.1080/09645292.2014.902428. https://doi.org/10.1080/09645292.2014.902428
  21. Oddny Judith Solheim and Vibeke Opheim (2019), 'Beyond Class Size Reduction: Towards More Flexible Ways of Implementing a Reduced Pupil–Teacher Ratio', International Journal of Educational Research 96:146–153. DOI: 10.1016/j.ijer.2018.10.008. https://doi.org/10.1016/j.ijer.2018.10.008
  22. Trude Gunnes, Lars J. Kirkebøen, Lena Lindenskov, and Marte Rønning (2025), 'Small- and Large-Group Instruction and a Didactic Method in Mathematics for Low-Performing Adolescents: Results from a Randomized Field Experiment', Scandinavian Journal of Economics 127(3):613–652. DOI: 10.1111/sjoe.12581. https://doi.org/10.1111/sjoe.12581