1. A Peking University study found China’s campaign to rein in private tutoring sharply reduced participation but did not erase inequalities in family spending or access to educational services.[para. 1] From 2017 to 2025, the share of students receiving academic tutoring fell from 38.1% to 20.1%, and participation in nonacademic programs fell from 22.4% to 16.7%.[para. 2] Yet average tutoring spending across all students fell only 11.8%, from 3,257 yuan to 2,873 yuan.[para. 2] Spending also became highly concentrated: the top 10% of students by expenditure accounted for 75.8% of total tutoring spending, 89.6% of academic-tutoring spending, and 91.1% of nonacademic-tutoring spending in 2025.[para. 3] The study said that without high-quality public alternatives for disadvantaged families, lower participation does not necessarily mean educational opportunities have become more equitable.[para. 4][para. 5]
2. The research, led by Wei Yi of Peking University’s China Institute for Educational Finance Research, used five rounds of household surveys from 2017 through 2025 covering primary, middle, and general high school students.[para. 6] China began cleaning up private tutoring institutions in 2018 and introduced the “Double Reduction” policy in 2021, limiting academic tutoring in compulsory education.[para. 7] Among students who received tutoring, average academic spending reached 10,066 yuan a year in 2025, and nonacademic spending reached 6,923 yuan, both substantially higher than in 2017.[para. 9] The top-spending households’ shares of academic and nonacademic tutoring expenditures fell to 56.4% and 56%, respectively, from 63.3% and 65.5%, but large disparities remained.[para. 10][para. 11] Students from low-spending households were far less likely to receive remedial, arts, sports, or individualized instruction.[para. 12]
3. A mother’s education level remained a strong predictor of tutoring participation: in 2025, 30.3% of students whose mothers held college or graduate degrees received academic tutoring, versus 8.7% of students whose mothers had little or no schooling.[para. 13] The gap was even larger for nonacademic programs, at 32.3% versus 3.8%.[para. 14] Urban students were about 2.5 times as likely as rural students to receive academic tutoring and roughly four times as likely to take nonacademic classes.[para. 15] The study said the market’s contraction narrowed some absolute gaps but did not remove the channels through which household resources and urban-rural differences shape educational opportunity.[para. 16][para. 17]
4. Academic-tutoring spending became increasingly concentrated in middle school and general high school.[para. 19] Average spending among participants rose from 5,816 yuan in lower primary grades to 17,540 yuan in general high school, roughly triple the lower-primary level, compared with a ratio of 1.85 in 2017.[para. 20] Course choices shifted by grade: foreign-language tutoring was most common in primary school, chosen by 71.2% of lower-grade and 72.2% of upper-grade participants, while math and science dominated at higher levels.[para. 22][para. 23] Mathematics was chosen by 78.8% of middle-school and 83.1% of general-high-school tutoring participants, and 49.7% of general-high-school participants chose physics, chemistry, or biology.[para. 23] The share of students combining academic and nonacademic tutoring fell from 12.7% in 2017 to 4.9% in 2025, and the average number of academic subjects taken dropped from 2.12 to 1.80.[para. 25] Nonacademic tutoring remained concentrated among younger children, with arts courses accounting for 75% of lower-primary nonacademic tutoring and sports for 40.1%.[para. 26]
5. Families still seeking tutoring used one-on-one instruction, commercial institutions, schoolteachers, live online lessons, and digital products, but access was uneven.[para. 28] Live online tutoring use fell from roughly one-quarter of language and math participants in 2021 to about 13% by 2025, while one-on-one math tutoring rose to 17.4% from 11.9%.[para. 29][para. 30] Digital products such as question banks and AI tools kept about one-quarter use rates, with annual user spending between 840 and 1,048 yuan.[para. 31] Regional gaps persisted: 17.7% of foreign-language tutoring participants in eastern China used live online instruction versus 5.4% of rural students, and 23.1% of private-school students used one-on-one foreign-language tutoring versus 2.3% of rural students.[para. 33] Commercial institutions remained the main provider for 56.8% to 66.4% of subject tutoring participants, while schoolteachers served 29.5% to 39.5%.[para. 34] In top-tier cities, 75.1% of foreign-language tutoring participants used commercial institutions versus 42.9% in rural areas, where 55.9% relied on schoolteachers versus 20.5% in the largest cities.[para. 35] Higher-spending households were more likely to use one-on-one tutoring, while lower-spending households relied more on schoolteachers.[para. 36]
6. The study said the shrinking private-tutoring market coincided with expanded after-school services in public schools, but the two did not create a consistent substitution pattern.[para. 37] It recommended that public policy move beyond simply expanding after-school coverage and focus on improving quality, especially for struggling students, students transitioning between school stages, and those who cannot afford market services.[para. 38] County-level and rural areas need more stable access to qualified teachers, after-school academic help, learning-difficulty support, digital resources, and arts and sports programs.[para. 39]
AI generated, for reference only
Sign in with Google
Sign in with Facebook
Sign in with 财新