Bigger Chinese cities run fewer, larger schools without cutting teachers per student

As cities grow, their education systems do not simply grow with them. A new analysis of nearly 300 Chinese cities finds that students, teachers, and schools all increase less than proportionally with population size, and that schools show the strongest departure from proportionality of all. The result is a consistent structural pattern: larger cities organize basic education through relatively fewer and larger schools, without a corresponding reduction in teaching capacity relative to the number of students they serve.

The study, published in Computational Urban Science, analyzed the scaling of basic education systems across Chinese cities from 2000 to 2020. Its authors, led by Mengqin Wu and Yuntong Cao with senior contributions from Yike Guo, Gang Xu, and Limin Jiao, set out to answer a question that earlier work had left open: whether student enrolment, teacher inputs, and school facilities respond uniformly to city size. In 2020, all six indicators they examined—students, teachers, and schools at both the primary and secondary stages—scaled sublinearly with population, with exponents ranging from 0.74 to 0.87.

Urban scaling analysis rests on a power-law relationship in which an urban quantity Y is proportional to population N raised to an exponent β. When β is greater than 1, the quantity grows more than proportionally with population; when β is about 1, it grows in proportion; and when β is below 1, it grows less than proportionally, meaning that per capita values decline as cities get bigger. This framework, developed originally for indicators such as gross domestic product, road length, and infrastructure, has been applied to carbon emissions, land-use morphology, and infectious diseases, but its joint application to the demand, staffing, and facility dimensions of a single public-service system is what makes the new study distinctive.

The data came from the China Urban Statistical Yearbook for 2000 to 2020, with 2003 excluded because records were unavailable. The researchers used the municipal districts of prefecture-level cities—the shixiaqu—as their urban units, a choice that focuses on urbanized areas rather than full prefecture territories that often include substantial rural populations. After excluding observations with missing or non-positive values, up to 284 cities remained before outlier screening. The team log-transformed the variables, fitted ordinary least squares regressions for each indicator-year combination, removed observations with residuals exceeding two standard deviations, and re-estimated the models to obtain exponents, 95 percent confidence intervals, and coefficients of determination.

The 2020 cross-sectional estimates reveal a finely graded structure. For primary education, students and teachers shared an exponent of 0.87, with confidence intervals of 0.82 to 0.91 and 0.83 to 0.91 respectively. Secondary students scaled at 0.78 and secondary teachers at 0.79, both with confidence intervals below 1. Schools scaled most sublinearly of all: 0.74 for primary schools and 0.75 for secondary schools. Because student and teacher exponents are nearly identical within each education stage, student–teacher ratios are broadly scale-neutral across city sizes. The lower exponent for school numbers, by contrast, implies that average school size increases with city size, particularly in primary education.

The fit quality also differed between stages. Secondary school numbers showed a strong association with population size, with an R-squared of 0.87, whereas the corresponding relationship for primary schools was weaker, at 0.60. This indicates that population size accounts for a smaller share of the cross-city variation in primary school numbers than in secondary school numbers, a difference the authors note when interpreting the reliability of the primary-school estimate.

The scaling relationships were not static over the two decades studied. In the temporal analysis covering 2000 to 2019, the exponent for primary school numbers showed the largest variation, falling sharply from approximately 0.93 in 2000 to a nadir of 0.76 in 2006, a period accompanied by wider confidence intervals and greater cross-city heterogeneity. The exponent then generally increased after 2009, reaching about 0.93 by 2019. The early decline coincided with the era of large-scale school consolidation in China, suggesting that the restructuring of local school networks may have contributed to the observed change. Secondary education indicators, by comparison, varied less, stabilizing within a narrow sublinear band of roughly 0.87 to 0.95 after an initial gentle decline between 2000 and 2005. The relative positions of the two stages also shifted: secondary exponents were generally higher in the early 2000s, but from around 2011 onward the exponents for primary students and teachers exceeded their secondary counterparts.

Beyond the scaling exponents, the study employed the Scale-Adjusted Metropolitan Indicator, or SAMI, which measures the deviation of a city’s observed value from the level expected for its population under the fitted scaling relationship. Positive SAMI values indicate provision above the population-based expectation; negative values indicate provision below it. Because the indicators generally scale sublinearly, their per capita values tend to decline with population size, so conventional per capita rankings and SAMI-based rankings can differ substantially. Using 2015 to 2020 averages, the rank-rank comparison showed that many megacities, including Beijing and Chongqing, rank higher under SAMI than under per capita measures, particularly for school indicators, while many smaller cities move in the opposite direction. Cities such as Wuhan, Nanchang, Hanzhong, and Hulunbuir moved substantially upward after scale adjustment, whereas Lhasa and Zaozhuang shifted downward. Shenzhen stood out as an exception: in secondary education it ranked higher under per capita measures than under SAMI for teachers and schools, showing that the general size-related pattern is not universal.

The spatial distribution of SAMI values in 2020 revealed pronounced regional clustering. At the primary education level, negative values—resource provision markedly below the level expected for a given population—were concentrated in parts of the eastern coast and in northern and northeastern China, including cities in Liaoning, Jilin, Heilongjiang, and Inner Mongolia. Positive values were more common in parts of central and southern China, including Henan, Guizhou, Guangxi, Jiangxi, and Guangdong. Broadly similar patterns appeared across the three primary education indicators and extended to the secondary stage, although local differences remained; for example, positive SAMI values observed for several primary indicators in Guangxi became weaker for secondary school facilities. The authors note that the broadly corresponding spatial patterns of students, teachers, and schools suggest that educational resources generally follow regional differences in student enrolment, and that such differences may partly reflect regional variation in demographic structure and population redistribution.

The findings carry clear limitations. Excluding observations with missing or non-positive values may introduce selection bias. The six indicators capture quantitative dimensions of the system but not qualitative ones such as teacher qualifications, school quality, or student learning outcomes. The analysis is based on aggregate municipal-district data and identifies statistical scaling relationships rather than causal mechanisms, and scaling laws can differ markedly across the urban–rural divide. Within these constraints, the study concludes that the basic education system of Chinese cities has a broadly sublinear but heterogeneous scaling structure, in which economies of scale arise mainly through the organization of school facilities rather than through reduced teacher provision per student, and that scale-adjusted indicators such as SAMI can complement conventional per capita measures when comparing educational provision across cities of different sizes.

Subject of Research: Urban scaling of students, teachers, and schools in Chinese basic education systems

Article Title: Urban scaling of basic education systems of Chinese cities

Article References: Wu, M., Cao, Y., Guo, Y., Cai, R., Xu, Z., Xu, G., & Jiao, L. (2026). Urban scaling of basic education systems of Chinese cities. Computational Urban Science, 6(1), Article 66. https://doi.org/10.1007/s43762-026-00301-5

Image Credits: AI Generated

DOI: 10.1007/s43762-026-00301-5

Keywords: urban scaling, basic education, China, primary education, secondary education, school consolidation, student-teacher ratios, SAMI, population size, municipal districts, economies of scale, education policy