Private Tutoring and Shadow Education: How Extra Spending Affects Equity

Private tutoring has moved from a limited form of remedial support to a large parallel market that follows the subjects, examinations, and selection pressures of formal schooling. UNESCO estimates that the global tutoring industry was worth $159 billion in 2023 and could reach $288 billion by 2030.^[a] These figures include one-to-one lessons, small groups, commercial learning centres, large examination-preparation classes, and online services. The market can extend learning time and provide focused academic help. Access, however, depends heavily on household income, location, information, available time, and the local supply of qualified tutors.

The equity question is therefore broader than whether tutoring raises a test score. It concerns who receives tutoring, how much they receive, what quality they can purchase, and whether the resulting advantages influence access to selective schools or universities. Two students may attend the same public school and study the same curriculum while experiencing very different amounts of instruction after the school day ends.

What Shadow Education Includes

Shadow education usually means fee-charging instruction in academic subjects that supplements formal schooling. The term “shadow” reflects its dependence on the formal system: when school curricula, examinations, or admission rules change, tutoring services adapt. Mathematics, science, national languages, foreign languages, and examination preparation account for much of the activity.

The category does not normally include unpaid help from parents, ordinary homework, music lessons taken only for recreation, or publicly funded remedial classes. Boundaries still vary across national surveys. Some datasets count arts, sports, childcare, online subscriptions, and self-study materials as private education, while others record only paid instruction in school subjects. Apparent differences between countries can therefore reflect measurement choices as well as real differences in participation.

• Individual tutoring: one learner works directly with a tutor, usually at the highest hourly price.
• Small-group tutoring: several learners share the cost while retaining some individual attention.
• Commercial classes: tutoring centres teach groups that may range from a few pupils to lecture-sized audiences.
• Teacher-provided tutoring: serving teachers earn additional income by teaching outside normal school hours.
• Online tutoring: live lessons, recorded courses, adaptive platforms, and subscription services reduce travel barriers but require devices and connectivity.
• Examination preparation: instruction focuses on test formats, speed, likely content, and admission strategies.

The Global Scale and the Data Problem

UNESCO’s 2025 study presents shadow education as a global activity rather than a practice confined to one region. Participation exceeds 70% in certain countries and grades. Yet the available statistics form an uneven collection of national surveys, household expenditure studies, and international assessments rather than one comparable global series.^[b]

In UNESCO’s request to 205 countries and self-governing jurisdictions, 143, or 69.8%, provided no quantitative data. Only 62 supplied any statistics, and 16 of those relied solely on publications more than a decade old.^[a] Governments may consequently know public expenditure per student while knowing little about the additional amount households purchase outside school.

Why Families Purchase Extra Instruction

Families purchase tutoring for several distinct reasons. A learner may need help with unfinished foundational skills, a quiet place to study, instruction in a language not spoken at home, or preparation for an examination. Other households use tutoring for acceleration, advanced curriculum coverage, or early access to material that schools will teach later.

Competition changes the meaning of the purchase. When only a few students receive tutoring, it may operate as optional enrichment or targeted remediation. When most classmates receive it, families may view non-participation as a disadvantage. Demand then arises partly from what other households are buying. Even parents satisfied with their child’s school may pay because admission decisions rank students against one another.

High-stakes examinations amplify this pattern. A small score difference can affect admission to an academic track, selective secondary school, scholarship, or university programme. Families with more resources can purchase longer courses, multiple subjects, individual feedback, repeated mock examinations, and advice about applications. Lower-income households may purchase a shorter course or a large-group class, even when tutoring absorbs a larger share of their disposable income.

School conditions also affect demand. Large classes, teacher shortages, interrupted attendance, language differences, and limited individualized support can lead families to seek outside instruction. Tutoring also remains widespread in several high-performing systems, which shows that demand does not disappear automatically when average school quality rises. Selection pressure and the desire for relative advantage can sustain it.

How Extra Spending Produces Unequal Access

Participation Is Only the First Divide

A basic comparison separates students who receive tutoring from those who do not. This can understate the gap. Among participants, families purchase different numbers of subjects, hours, class sizes, tutor qualifications, learning materials, and periods of enrolment. Intensity and quality may be more unequally distributed than participation itself.

Sri Lankan household data cited by UNESCO illustrate the distinction. In 2016/17, tutoring participation reached 68% among the richest households and 60% among the poorest. The eight-point participation gap appears moderate. Monthly spending by the richest quintile, however, was nearly three times spending by the poorest quintile. Wealthier families more often purchased individual or small-group instruction, while poorer families relied more heavily on large classes.^[c]

The Same Payment Can Create a Different Burden

Dollar spending alone does not measure affordability. A smaller tutoring payment may consume a larger portion of a low-income household’s budget. Sri Lankan evidence found that richer households spent more in absolute terms, but poorer households allocated a larger share of total expenditure to tutoring.^[a] This can displace spending on transport, food, healthcare, connectivity, books, or activities outside academic study.

The financial burden also varies across the school cycle. Examination years can produce short periods of concentrated expenditure. Families with savings can respond to a weak report, a changed examination, or an admission opportunity immediately. Households with little financial flexibility may have to delay support, reduce the number of subjects, or stop lessons before an examination.

Location Shapes Price and Choice

Urban markets often contain more tutoring centres, specialist teachers, and competing providers. Rural students may face fewer choices and additional transport costs. Online delivery can widen the geographic reach of tutoring, but it does not remove differences in broadband quality, device access, digital skills, payment systems, or access to a quiet room.

Online platforms can also segment the market. Recorded lessons may lower the price of basic access, while live individual instruction, detailed marking, analytics, and admission counselling remain premium services. Digital expansion can reduce one barrier while preserving another through service tiers.

Information Has Economic Value

Families do not enter tutoring markets with equal information. Parents with personal experience of selective education may know when to begin preparation, which subjects carry greater admission weight, how tutor reputations differ, and which credentials matter. Social networks circulate recommendations and warnings. Families new to an education system, or those with limited time and language access, may pay for unsuitable services or enter the market later.

What the Achievement Evidence Shows

Does higher spending always produce stronger learning? The research does not support such a simple relationship. Tutoring varies too widely in purpose and design, while families select it in response to prior achievement, motivation, examinations, and school quality. Students receiving tutoring may initially be far behind, already far ahead, or preparing for a particular test. A raw comparison between tutored and non-tutored students can therefore misstate the effect.

Structured Tutoring Can Raise Learning

A meta-analysis of experimental pre-primary through Grade 12 tutoring studies reported an average effect of 0.37 standard deviations on learning outcomes.^[d] The studies covered instruction by teachers, paraprofessionals, volunteers, and parents in one-to-one or small-group settings. This finding establishes that well-designed additional instruction can raise attainment. It does not prove that every commercial tutoring purchase produces the same effect.

The Education Endowment Foundation estimates that school-based small-group tuition produces about four additional months of progress on average over a year. Its evidence review defines a small group as two to five pupils and notes that impact tends to fall once groups grow beyond six or seven learners.^[e] Diagnostic targeting, tutor training, curriculum alignment, session frequency, and pupil attendance affect results.

A 2026 synthesis of 76 international studies found that private mathematics tutoring had a modest positive association with cognitive learning outcomes, while effects on non-cognitive mathematics outcomes were not evident. Results varied by location and student grade.^[f] The variation matters because “private tutoring” can describe anything from individualized instruction by a trained teacher to a crowded examination lecture.

Selection Bias Complicates Comparisons

Motivated families may purchase tutoring and also provide books, study routines, technology, and academic expectations. Conversely, families may purchase tutoring because a student is struggling. The first process can make tutoring appear more effective than it is; the second can make it appear less effective. Strong evaluations use random assignment, longitudinal data, matched comparisons, instrumental variables, or other designs intended to separate tutoring effects from prior differences.

A 2024 study using nationally representative longitudinal data from Chinese middle schools estimated that differences in tutoring participation and intensity accounted for up to 13% of socioeconomic achievement gaps in certain subjects and comparisons.^[g] Tutoring did not explain every gap. Its contribution grew when researchers compared disadvantaged learners with the most advantaged groups, whose students covered more subjects, spent more, and participated more intensively.

Tutoring Sits Inside a Wider Resource Gap

OECD’s PISA 2022 results show that a one-unit rise in its index of economic, social, and cultural status was associated with 39 additional mathematics points across OECD countries.^[h] The index combines parental education, occupational status, and home possessions. OECD identifies private tutoring as one financial resource available more often to advantaged students, alongside computers and books. The 39-point relationship must not be attributed to tutoring alone.

Equity effects depend on distribution as well as effectiveness. A highly effective programme offered mainly to already advantaged students can widen an achievement gap. The same instructional model, targeted toward pupils who have fallen behind and funded through schools, can narrow it. The pedagogy may be similar while the financing and allocation rules produce opposite distributional results.

The Republic of Korea as a Spending Case

The Republic of Korea maintains one of the most detailed official statistical systems for private education. Its 2024 survey covered elementary, middle, and high school students and reported an 80% participation rate, 7.6 weekly hours per student, and a 7.7% annual rise in total expenditure. Average monthly expenditure per student rose 9.3%, while expenditure per participating student rose 7.2%.^[i]

Participation reached 87.7% in elementary school, 78% in middle school, and 67.3% in high school. The survey also found that spending rose with household income. These figures show why a national participation rate alone cannot describe equity. The number of subjects, hours, lesson type, and monthly expenditure vary within the participating majority.

Korean statistics also demonstrate how definitions affect international comparisons. The national category includes a wider range of private education than some cross-national academic-tutoring measures. UNESCO’s 2025 report cites lower 2023 rates for academic tutoring by school level. Both sets can be accurate within their stated coverage. Analysts should compare survey questions and included services before ranking countries.

The country’s school structure, examination system, and private tutoring sector are closely connected. A broader profile of the South Korean education system provides context for the official spending and participation figures.^[j]

Effects Inside Regular Schools

Shadow education can alter classroom conditions even when schools do not operate the tutoring. Students who learned a topic in advance may answer more quickly, become disengaged during repetition, or encourage teachers to accelerate. Students without advance exposure may then experience ordinary lessons as if they were behind. The effective curriculum can move ahead of the published curriculum.

Private tutoring can also help regular classrooms. Remedial support may enable a learner to master missing skills and participate more confidently. Teachers may spend less class time revisiting earlier content. Higher-attaining students can explore material beyond the grade syllabus without requiring the classroom teacher to teach several levels simultaneously.

Alignment determines part of the effect. Tutoring that uses different terminology, methods, or sequencing may confuse learners. Examination drilling may improve speed without strengthening transferable understanding. Instruction that diagnoses a specific misconception and coordinates with the school curriculum is more likely to support durable learning.

Time creates another trade-off. Additional lessons extend academic learning but can reduce sleep, exercise, independent study, family time, and extracurricular participation. The effect depends on age, travel, session length, total weekly load, and whether tutoring replaces unstructured screen time or a valuable developmental activity. Expenditure data rarely capture these opportunity costs.

Teachers, Providers, and Market Quality

Tutors range from licensed teachers and subject specialists to university students, retirees, informal providers, and automated platforms. Price does not consistently indicate teaching quality. Families may observe examination results and testimonials but have limited access to information about tutor training, safeguarding, curriculum knowledge, class size, refund terms, or student progress after prior attainment is considered.

Serving teachers occupy a sensitive position when they tutor for payment. Their classroom knowledge can make instruction relevant, and supplementary income may support teacher retention where salaries are low. A conflict arises if teachers tutor their own pupils, reserve essential content for paid sessions, or create pressure to enrol. Clear rules about disclosure, student assignment, use of school premises, and instructional obligations protect both teachers and families.

Commercial providers can offer stable schedules, prepared materials, tutor training, and quality monitoring. They can also use sales practices that intensify parental anxiety or encourage long enrolment without evidence of continuing need. Basic registration, transparent pricing, truthful performance claims, safeguarding standards, and accessible complaint processes address market conduct without assuming that all providers operate in the same way.

Measuring Equity More Accurately

Household expenditure surveys remain an essential source because private tutoring payments often sit outside education ministry accounts. UNESCO’s education-finance methodology treats household spending as part of total education financing and notes that a higher private contribution can indicate a heavier household burden with consequences for access and equity.^[k]

Survey design requires care. Families may pay weekly, monthly, by term, or shortly before an examination. Payments may cover several children or combine academic instruction with transport and materials. Recall periods differ, informal cash payments may be missed, and respondents may not report expenses they regard as private. Annual totals should distinguish nominal growth from inflation-adjusted growth.

A useful equity dataset records more than whether a student attended tutoring. It separates participation, expenditure, instructional hours, subjects, class size, delivery mode, provider type, and purpose. Results need disaggregation by household income or wealth, school level, prior attainment, location, disability, language background, and other locally relevant characteristics.

Distribution should be examined at several points. The first is access to tutoring. The second is the quantity and type purchased. The third is the learning gain produced. The fourth is whether that gain changes placement, credentials, or admission. A small average score effect can carry greater consequences when it occurs near a competitive selection threshold.

Public Responses and Their Equity Effects

Education systems have used provider registration, limits on teachers tutoring their own pupils, operating-hour rules, price disclosure, tax reporting, public information, and school-based academic support. No single response fits every tutoring market. A country dominated by informal individual lessons faces different administrative needs from one dominated by large commercial chains or online platforms.

Publicly financed tutoring changes the allocation mechanism. Schools can offer individual or small-group instruction according to assessed learning needs rather than household purchasing power. Evidence from experimental tutoring studies indicates that frequent, well-aligned sessions can improve attainment. Equal access alone is not enough; disadvantaged learners need access to trained tutors, suitable schedules, consistent attendance, and instruction connected to classroom content.

Subsidies or vouchers can expand choice but may produce uneven results when provider quality is unclear or when families must make additional payments. School-based programmes reduce some information and transport barriers, although staffing and timetable capacity can limit reach. Online programmes can serve remote learners, provided that device, connectivity, accessibility, and adult-support needs are addressed.

Rules directed only at supply may not reduce demand created by selective examinations. Families can move toward informal providers, smaller groups, or online services. Measures connected to admissions, school quality, public remedial support, and transparent information address the reasons households purchase tutoring as well as the conduct of providers.

Evaluation should report who participates and who benefits. An overall rise in average scores can coexist with a wider income gap if places are captured mainly by learners who already have other advantages. Conversely, a programme with a moderate average effect may improve equity when it reaches students whose needs were previously unmet.

A More Precise Reading of Shadow Education

Private tutoring is neither uniformly beneficial nor uniformly unequal. Its distributional effect follows the interaction of price, household resources, instructional quality, market geography, examination pressure, and public provision. Paid tutoring can help a struggling learner recover missing knowledge. It can also allow already advantaged students to purchase more instructional time, smaller classes, and specialized preparation.

The clearest equity distinction is between tutoring as instruction and tutoring as a market allocation system. Research shows that additional instruction can improve learning. Markets allocate that instruction according to purchasing ability unless subsidies, school provision, community programmes, or other access arrangements intervene. The educational method may work while access to it remains unequal.

Better statistics can replace broad assumptions with testable findings. Participation must be read beside hours, spending, class size, purpose, prior attainment, and household resources. When those measures are connected to learning growth and admission outcomes, shadow education becomes visible as part of the full education-finance picture rather than an activity taking place beyond official accounts.

Sources and Data Notes

1. [a] Whose visions for what learning? Perspectives, policies and practices in private supplementary tutoring — UNESCO’s 2025 report page covering market scale, participation, data shortages, household burdens, and regulation.
2. [b] What you need to know about private supplementary tutoring — UNESCO’s explanation of tutoring forms, demand, educational effects, and social disparities.
3. [c] Shadow Education: Private Supplementary Tutoring and Its Implications for Policy Makers in Asia — Asian Development Bank study of participation, household expenditure, and inequality across Asian systems.
4. [d] The Impressive Effects of Tutoring on PreK-12 Learning: A Systematic Review and Meta-Analysis of the Experimental Evidence — Experimental evidence on learning effects from individual and small-group tutoring.
5. [e] Small group tuition — Evidence review covering group size, average learning progress, targeting, and tutor preparation.
6. [f] The association between private tutoring and mathematics learning outcomes in primary and secondary schools: A three-level meta-analysis — A 2026 synthesis of 76 studies examining mathematics outcomes.
7. [g] Inequality in the shadow: The role of private tutoring in SES achievement gaps — Longitudinal analysis of participation, intensity, expenditure, and achievement gaps among Chinese middle school students.
8. [h] Equity in education in PISA 2022: PISA 2022 Results (Volume I) — OECD evidence on socioeconomic status, educational resources, resilience, and mathematics performance.
9. [i] Private Education Expenditures Survey of Elementary, Middle and High School Students in 2024 — Official Korean statistics on expenditure, participation, weekly hours, household income, and school level.
10. [j] South Korea Education System (2026): Structure, Quality, and Performance — Country-level context for the school structure, examinations, and private tutoring sector.
11. [k] Expenditure on education — UNESCO Institute for Statistics definition covering public, international, private, and household sources of education expenditure.