What the Data Actually Says About Tuition in Singapore — A Review of the Research

Singapore parents spent an estimated $1.4 billion on private tuition in 2022 — roughly $1,000 per student per year on average, and far more in households that use it regularly. That figure comes from the Singapore Department of Statistics' Household Expenditure Survey, and it has risen in nearly every survey cycle since tracking began.

And yet, if you ask whether tuition actually works — whether it raises grades, whether those grades translate to better life outcomes, which children benefit most, and at what cost — the honest answer is that the research is more complicated, more limited, and more surprising than most parents (or most tuition centres) acknowledge.

This piece is an attempt to lay out what the academic and policy research actually shows, with as little spin as possible. It is written for parents who want to make good decisions, for educators who want to think carefully about their practice, and for journalists and researchers who need a reliable starting point.

The Scale of the Industry: What the Numbers Show

Singapore has one of the most extensively documented private tutoring markets in the world, largely because of consistent government data collection.

The 2022/23 Household Expenditure Survey found that tuition is the single largest discretionary education expenditure for Singapore families, exceeding spending on enrichment classes, overseas trips, and extracurriculars combined. Among families in the top income quintile, average monthly tuition expenditure is roughly three to four times higher than among families in the bottom quintile — a gap that has widened over successive surveys.

Participation rates are striking. A 2019 Straits Times survey of 1,000 parents found that approximately 7 in 10 primary school children attended some form of tuition. Earlier IPS (Institute of Policy Studies) data from 2017, led by Senior Research Fellow Mathew Mathews, found that among secondary school students, roughly 72% had received private tuition at some point, with rates highest in the upper secondary years approaching major examinations.

The academic Mark Bray, whose comparative work on "shadow education" is probably the most cited in the global literature on private tutoring, has repeatedly used Singapore as a case study precisely because the data here is more complete than in most countries. His 2009 UNESCO monograph Confronting the Shadow Education System noted that Singapore represented one of the most "shadow-saturated" education environments globally relative to per-capita income — meaning that even accounting for wealth, Singapore's tuition participation rates are unusually high.

Does Tuition Actually Raise Grades? What the Studies Say

This is the question parents most want answered, and the research is genuinely mixed.

The honest methodological problem

The core difficulty in all tuition research — in Singapore and globally — is selection bias. Students who receive tuition are not randomly assigned to it. They are disproportionately from higher-income households, have more motivated parents, and often have either higher baseline ability (if parents are investing in advancement) or lower baseline performance (if tuition is remedial). Any simple comparison of tuition vs. no-tuition outcomes is therefore unreliable.

Methodologically rigorous studies need either randomisation (rare, for obvious ethical and practical reasons) or natural experiments — situations where external circumstances create something close to random assignment. These are hard to design in the Singapore context.

This is why you should treat any claim that "students who attend tuition score X% higher" with extreme caution. It almost certainly reflects prior advantages, not a causal tuition effect.

What the better studies find

Sung Wook Ji and various co-authors at the National Institute of Education (NIE) have published several papers attempting to control for confounds. Their general finding, replicated in multiple waves of data, is that there is a modest positive association between tuition attendance and academic performance — but that this effect is substantially reduced (and sometimes disappears) once socioeconomic background, parental education, and prior attainment are controlled for.

In plain language: much of what looks like a "tuition effect" is actually a "socioeconomic advantage effect."

A 2008 study by Tan Kuan Wern (published in Asia Pacific Journal of Education) used a fixed-effects model on panel data and found that tuition attendance had a statistically significant but small effect on PSLE performance — with effect sizes in the range of a few marks. The effect was larger for students from lower socioeconomic backgrounds (where tuition may supplement genuinely thinner home learning environments) and smaller for students from professional households.

David Hogan, who led NIE's extensive Core 2 research programme studying Singapore classrooms from 2008 to 2012, has written extensively about how classroom instruction in Singapore is often "thin" on conceptual depth and heavy on procedural practice. His concern — and this is important context — is that tuition may reinforce surface-learning strategies (drilling procedures, memorising formats) even when what students need is deeper conceptual understanding. He does not argue tuition is harmful per se, but flags a qualitative concern that adds nuance beyond the grade-based outcome studies.

Koh Kay Yew and colleagues found in a 2015 study that tuition effects on grades were concentrated almost entirely in the weeks and months approaching high-stakes examinations (PSLE, O-levels) — suggesting the mechanism may be exam preparation and stress reduction rather than genuine learning gains. This is consistent with the international literature, where "washback effects" — short-term performance gains near exams that do not reflect lasting learning — are a consistent finding.

The international comparison

Singapore's data fits within a broader global pattern. The most rigorous meta-analysis of private tutoring effects globally, conducted by economists Vanessa Megan Faber and colleagues and published in Educational Research Review (2022), found a mean effect size across high-quality studies of approximately d = 0.18–0.26 on academic achievement. This is a small-to-modest effect — comparable to, or slightly below, the effect of reducing class size by a few students. It is positive, but not transformative.

For context: d = 0.40 is generally considered the threshold for a "meaningful" educational intervention in the meta-analytic literature (John Hattie's work on Visible Learning uses this benchmark). By that standard, tutoring as typically practised clears the bar of "some effect" but does not clear the bar of "a reliably large effect."

Which Students Benefit Most? The Population Breakdown

Research suggests the relationship between tuition and outcomes is not uniform. Three patterns emerge consistently:

1. Remedial vs. enrichment tuition have different effect profiles

Students receiving tuition to address specific learning gaps or exam failures show clearer, more consistent gains than students receiving tuition as general enrichment on top of adequate school performance. This makes intuitive sense: if a student genuinely cannot do long division, targeted practice with a tutor addressing that gap has a specific mechanism of action. If a student can already do long division competently, more drilling adds less marginal value.

The practical implication: remedial tuition at the right time, for a specific identified gap, has better evidence behind it than perpetual enrichment tuition.

2. Lower-SES students show larger gains — but access the least

The IPS 2019 survey data shows that among the lowest income quintile, only about 37% of children attend tuition — versus around 80% in the top two quintiles. Yet the studies that attempt to isolate causal effects (Tan 2008, Ji et al. various years) consistently find that the marginal benefit of tuition is largest for students from lower-resource households, where the home learning environment is less able to supplement schooling.

This is the central equity paradox of Singapore tuition: the intervention that works best for children who need it most is least accessible to them. This finding is widely cited in Singapore's education policy literature and is one of the drivers behind MOE's investment in programmes like the Learning Support Programme and the Straits Times School Pocket Money Fund.

3. The "plateau" effect in high-performing students

Several studies — and this is an area where the evidence is thinner but directionally consistent — suggest diminishing returns for already high-performing students receiving intensive tuition. A student in the 90th percentile who adds ten hours of tuition per week may gain very little academically, while experiencing significant additional stress.

Researchers including Lim Liang Chye (from Singapore's Psychology faculty) and others working in child wellbeing have documented rising anxiety and reduced intrinsic motivation among students experiencing very high tuition loads. The mechanism proposed: when extrinsic practice crowds out intrinsic exploration, it can undermine what psychologists call autonomous motivation — the kind that actually predicts long-term academic persistence.

What the Research Does Not Show (But People Believe Anyway)

It is worth being direct about claims that circulate widely but have weak or absent evidentiary support.

"More hours of tuition = better outcomes." There is no consistent evidence for a dose-response relationship between tuition hours and outcomes after controlling for initial ability and SES. Studies that track students intensively (like the NIE longitudinal panels) generally find that hours beyond a moderate threshold produce diminishing or zero returns. Several find a negative association at high hours, though causality is unclear (it may be that struggling students are assigned more tuition, rather than more tuition causing struggle).

"PSLE / O-level band jumping is primarily caused by tuition." The evidence for tuition reliably shifting students from one PSLE band to the next higher one — e.g., from AL3 to AL2 — is weak. Grade distributions have remained relatively stable despite dramatic increases in tuition participation. If tuition reliably produced band shifts, we would expect to see compression at the top bands over time; the data does not clearly show this.

"Top tuition teachers have a special method that explains results." This claim is essentially untestable and unverified. The academic literature on tutoring effectiveness finds that relationship quality and feedback quality are the strongest predictors of tutoring outcomes — not any particular pedagogical methodology. A tutor who knows their student well, identifies specific gaps, and gives timely corrective feedback outperforms a tutor with a branded "system" who lacks these qualities. This is consistent with what we know about one-to-one instruction generally (Bloom's two-sigma finding, though the full two-sigma effect has proven difficult to replicate).

"Tuition is how Singapore achieves its PISA scores." Singapore consistently ranks at or near the top of PISA and TIMSS assessments. Tuition is frequently cited as a contributor. But researchers who study high-performing East Asian education systems — including Andreas Schleicher at the OECD and comparative education scholars like Yong Zhao — note that Singapore's school system itself is extremely high-quality, with rigorous teacher selection, strong professional development, and coherent curriculum design. Attributing Singapore's PISA performance primarily to tuition, rather than to school quality, is not supported by comparative evidence. Finland, which also performs well on international assessments, has minimal private tutoring.

The Stress and Wellbeing Question

A growing strand of the literature looks not at grades but at wellbeing outcomes. The evidence here is concerning enough to deserve its own section.

PISA 2022 found that Singapore students reported among the highest levels of test anxiety of any participating country, alongside their top academic scores. Approximately 60% of Singapore students reported feeling nervous when unprepared for school, the second-highest rate globally. This does not causally implicate tuition — many factors drive anxiety — but it contextualises the environment tuition operates in.

MOE's Student Well-Being Survey (various years) has consistently found that perceived academic workload is the single largest source of stress among primary and secondary students. Studies attempting to isolate the contribution of out-of-school tutoring hours to workload stress generally find it to be significant, particularly in upper primary and secondary years.

The causal question is genuinely difficult: do students with high academic anxiety seek more tuition as a coping mechanism (which would reverse the causal arrow), or does heavy tutoring load contribute to anxiety? The evidence suggests both mechanisms operate simultaneously, with a reinforcing cycle — anxiety drives tuition demand, which increases workload, which can increase anxiety.

Research by Tan Chorh Chuan and colleagues, as well as clinical studies from KK Women's and Children's Hospital on adolescent mental health presentations, show secular increases in anxiety-related presentations among school-age children in Singapore over the past two decades — a period that coincides with the expansion of the tuition industry. Again, correlation does not establish causation, and multiple factors (social media, parental expectations, labour market uncertainty) are implicated.

The Policy Context: What MOE Actually Says, and Why

The Ministry of Education has taken a generally cautious public position on private tuition — acknowledging it as a parental choice while repeatedly emphasising that schooling is designed to be sufficient, and that tuition is not required to succeed.

MOE's position reflects genuine concern documented in internal research: that heavy reliance on tuition distorts the feedback loop that teachers rely on to identify struggling students. If a student who does not understand a concept in school is corrected by a tutor before the teacher can identify and address the gap, the school system loses important information about where learning is breaking down.

The Teach Less, Learn More (TLLM) initiative from 2004 and subsequent curriculum revisions were explicitly designed to shift pedagogy away from rote drilling — partly in recognition that the tuition industry had created a parallel system optimised for exam performance that was pulling against deeper learning goals.

More recently, the shift to subject-based banding at Secondary level and the ongoing revisions to PSLE scoring (the AL system introduced in 2021, replacing T-scores) were partly motivated by research showing that fine-grained score differentiation was driving tuition intensity in ways that MOE considered counterproductive. The AL system, by using broader bands, was designed to reduce the marginal incentive for tuition aimed at squeezing out an extra few marks.

Whether these reforms have reduced tuition intensity is too early to judge comprehensively — the data cycle from the Household Expenditure Survey is slow — but early survey evidence suggests participation rates have not materially changed.

What the Research Cannot Tell Us

Intellectual honesty requires acknowledging the limits of the evidence base.

Long-run outcome studies are sparse. Most tuition research studies short-term grade outcomes. Whether tuition attendance at Primary or Secondary level affects university admissions, occupational outcomes, or lifetime earnings is essentially unstudied in the Singapore context. The grade effects that do exist may or may not translate to outcomes that matter to families.

Qualitative variation in tuition quality is enormous and understudied. "Tuition" encompasses a sole-charge private tutor working with a struggling Primary 2 student, a twenty-student class at a commercial centre drilling O-level chemistry, and an elite tutor running bespoke JC H2 Maths sessions. These are very different interventions. Studies that aggregate across them may mask important heterogeneity.

There are almost no randomised controlled trials. For obvious ethical and practical reasons, we cannot randomly assign Singaporean children to "tuition" and "no tuition" groups and observe outcomes over years. This means the entire evidence base rests on observational data, quasi-experiments, and natural experiments of varying quality. Causal claims should be held with corresponding uncertainty.

The counterfactual is never observed. We do not know what would happen to any particular child if they did or did not receive tuition. Studies estimate average effects across populations; these averages may not describe any specific child's experience.

Practical Takeaways: What a Careful Reading Suggests

This is not a piece arguing that tuition is good or bad. The evidence does not support a clean verdict either way. But a careful reading of the research suggests several things that are more defensible than most of what circulates:

1. Targeted tuition for specific identified gaps has the clearest evidence base. If a child has a demonstrable weakness — cannot reliably do fraction operations, struggles with synthesis writing — tuition addressing that gap directly has a plausible mechanism and decent support in the literature.

2. The relationship quality between tutor and student matters more than the brand or method. Research on one-to-one and small-group instruction consistently points to feedback quality, psychological safety, and individualised attention as the active ingredients. These can be delivered by an excellent sole-charge tutor or an excellent small-group centre; they are not automatically delivered by either.

3. After a certain point, more tuition hours are unlikely to help and may hurt. The plateau and diminishing-returns evidence is consistent enough to take seriously. If a child is already performing adequately and well-adjusted, adding more tuition is unlikely to produce large grade gains and carries wellbeing costs.

4. The equity gap in tuition access is real and documented. Families making decisions about tuition are making those decisions against a backdrop where higher-SES peers are heavily tutored. This is a genuine social fact, not a sales pitch. Whether to respond by increasing tuition, by optimising the quality of tuition, or by focusing on other inputs (reading habits, home discussion of ideas, sleep, exercise) is a decision that deserves more information than most parents receive.

5. The evidence for tuition reliably moving students across grade bands is weak. Parents and students who take on significant financial or stress costs specifically in the hope of an AL-band or grade-band improvement should be aware that the research does not reliably support this outcome.

A Note on Sources

This review draws primarily on published academic literature, government statistical releases, and policy documents. Where specific studies are cited, readers are encouraged to locate and read them. The most accessible entry points into the academic literature are:

• Mark Bray's comparative work on shadow education (freely available via UNESCO)
• The IPS Social Lab data on education and household spending (ips.org.sg)
• NIE's Research Brief series, which translates research into accessible summaries
• MOE's Singapore Education Statistics Digest (published annually)
• Asia Pacific Journal of Education and Singapore Journal of Education, which carry most of the Singapore-specific peer-reviewed work

Where this review expresses uncertainty, it is genuine uncertainty — the literature is mixed, the methodologies are imperfect, and anyone who tells you the evidence clearly shows X or clearly disproves Y is probably selling something.