Population Aging and Education Spending in Chinese Cities

Population aging may affect public education finance through enrolled cohorts and budget allocation. Using 299 Chinese cities (2000–2020), we estimate total public-budget education spending, its budget share, and spending per enrolled student in a 2SLS design that instruments the elderly share with the 2000 share aged 45–64 interacted with the national aging trend. In the baseline city- and year-fixed-effects model, a one-percentage-point increase in the elderly share is associated with a 4.4-log-point decline in total spending (city-clustered p = 0.005); education-share (+0.0003; p = 0.976) and average per-student (+0.026; p = 0.295) estimates are imprecise. The estimates are specification-sensitive. Province-by-year fixed effects move the total-spending estimate to −0.0022 and the education-share estimate to +0.0394; the average per-student estimate becomes +0.0873, with province-clustered 95-percent confidence intervals that include zero for all three outcomes. Adding the 2000 child-population share interacted with a linear trend also moves the total-spending estimate close to zero, while some child-composition specifications raise the per-student estimate toward +0.11 (p < 0.001). The continuous per-student fiscal-capacity interaction remains positive, but split-sample first stages are weak in two terciles. The spending–enrollment relationship is an accounting decomposition, not evidence about why enrollment changes. These estimates are conditional on the identifying assumptions; the design does not identify statutory protection or migrant-child exclusion, and it does not establish causal budget reallocation.