A second reading of the 2025 results: the size of the gain, the meaning of the rank, the level reached, how the gain was shared, and the domain the cycle left out.
1 · The gain
Between 2022 and 2025 the Philippine mean rose from 347 to 367 points in reading, from 355 to 371 in mathematics, and from 356 to 373 in science. All three changes are statistically significant. The Philippines was one of eight participating systems to improve significantly in all three domains. Its 20-point gain in reading was the largest recorded between the two cycles: the OECD’s country comparison ranks it first of 50 systems, although the full results table shows Jordan with the same rounded gain.1
The OECD’s working conversion treats about 20 points as a year of schooling. On that rule the reading gain is roughly a year of learning, achieved while the OECD average fell. The conversion is approximate, and the OECD presents it as such. The direction of the change and its significance are not in doubt.2
The share of students reaching Level 2, PISA’s baseline of proficiency, rose in every domain: from 23 to 32 per cent in science, from 16 to 21 per cent in mathematics, and from 24 to 31 per cent in reading. Teachers and school leaders produced these figures, and they belong at the head of any reading of the results.3
| Domain | 2022 | 2025 | Change | OECD 2025 | Gap 2025 |
|---|---|---|---|---|---|
| Reading | 347 | 367 | +20 | 461 | 94 |
| Mathematics | 355 | 371 | +16 | 463 | 92 |
| Science | 356 | 373 | +17 | 482 | 109 |
Source: OECD, PISA 2025 Results (Volume I), Tables I.1 and I.2.6.
| Domain | Philippines 2022 | Philippines 2025 | OECD 2025 |
|---|---|---|---|
| Reading | 24 | 31 | 69 |
| Mathematics | 16 | 21 | 65 |
| Science | 23 | 32 | 74 |
Sources: OECD, PISA 2025 Results (Volume I) and country note for the Philippines; Department of Education national launch, 8 September 2026.
2 · The rank
In 2025 the Philippines placed ahead of 18 systems in reading, 16 in mathematics and 14 in science. In 2018 it was last in reading, and in 2022 it was sixth from the bottom in reading and mathematics. Two features of the 2025 table make the rank a weaker signal than the score.4
The field grew. Seventy-nine systems took part in 2018 and 91 in 2025. Of the seven that took part for the first time, five placed below the Philippines in mathematics and six in science. Some of the climb is therefore a change in who is in the table.5
PISA’s scale is linked across cycles. Its rank tables are not, because their membership changes.
The field also fell. The OECD average in 2025 was the lowest yet recorded in all three domains. Across the 35 OECD countries with comparable data, reading fell by about 14 points from 2022 and mathematics by 9, while science did not change significantly. The reading gap between the Philippines and the OECD average therefore narrowed by about 34 points between the two cycles, and about 14 of those points came from the OECD’s decline, not from Filipino gains. Over the longer run the fall is steeper: since 2018, reading has declined in three-quarters of the systems with comparable data, and by 25 points across the OECD.6
A country can climb a rank table in three ways: by improving, by declining less than the others, or by being joined by systems that score below it. In 2025 all three happened at once. PISA’s scale is linked across cycles. Its rank tables are not, because their membership changes, so the score is the measure to quote and leaving the bottom ten is best read as a fact about the table.
3 · The level
Level 2 is a modest standard. In mathematics it describes a student who can see, without being told how, that a simple situation can be represented mathematically: comparing the total length of two routes, say, or converting a price into another currency. In 2025, 21 per cent of Filipino students reached it, against 65 per cent across the OECD. In reading the figures were 31 and 69 per cent, and in science 32 and 74 per cent. Put the other way, about four in five Filipino students were below baseline in mathematics, and about seven in ten in reading and science.7
The distance to the OECD average is 92 points in mathematics, 94 in reading and 109 in science. The 20-point rule would put that at roughly five years of schooling, but the OECD cautions against converting differences between countries this way, so the figure gives a sense of scale and no more.8
At the top of the scale the change cannot yet be seen. The OECD’s country profile puts the share of Filipino students at Level 5 or 6 in science at 0.2 per cent, against 7.2 per cent across the OECD, and press accounts of the national results put the share at 0.3 per cent. Applied to the roughly 1.75 million 15-year-olds the sample represents, either figure means a few thousand students. The arithmetic rests on a small proportion with a wide sampling error, and it should be quoted only as an order of magnitude.9
4 · The distribution
Every socio-economic group improved in reading, but not equally. Among the most advantaged quarter of Filipino students the mean reading score rose 41 points, from 376 to 417. Among the least advantaged quarter it rose 15. The OECD reports that the socio-economic gap in reading narrowed in 37 countries and economies between 2022 and 2025, held steady in the rest, and widened in one: the Philippines.10
This does not diminish the gain at the bottom, which is real. It changes what the national mean is evidence of. A mean that rises because the advantaged quarter pulls away describes a different school system from a mean that rises because the floor comes up, even when the two means are the same number.
The same data carry a counter-signal. About 11 per cent of disadvantaged Filipino students scored in the top quarter of national performance in science, the group PISA calls academically resilient. Background did not decide their outcome. The schools that produce such students are the natural place for the next round of study to begin.11
A mean that rises because the advantaged quarter pulls away describes a different school system from a mean that rises because the floor comes up.
5 · The sample
PISA tests 15-year-olds enrolled in Grade 7 or above. The 2025 Philippine sample of 8,072 students in 208 schools represents about 1.75 million young people, roughly 82 per cent of the country’s 15-year-olds. In 2022 the figure was about 83 per cent, so the gain is not the product of a narrower sample. Nor does it say anything about the fifth of the age group outside the frame, those out of school or held back below Grade 7, whose learning went unmeasured in both cycles.12
Preparation was unusually intense. The Department of Education formed a task force for the 2025 cycle, and the Secretary likened the run-up to preparing for the bar examination. Familiarity with the platform, with the item formats and with the experience of sitting a long computer-based test can raise scores without any matching change in what students can do outside the test room. Preparation of this kind is legitimate, and other systems do it. The point is narrower. The 2025 data cannot separate gains in learning from gains in test familiarity. Later cycles can, and so can assessments that nobody prepared for.13
6 · The unmeasured domain
PISA 2025 assessed reading, mathematics and science, with Learning in the Digital World as its innovative domain. It did not assess creative thinking. The only international measure of that capability comes from the 2022 cycle, which scored students’ ability to generate, evaluate and improve ideas in written, visual, scientific and social tasks on a scale of 0 to 60. Filipino students averaged 14 points. The OECD average was 33, and Singapore’s 41. The Philippines placed among the lowest four of the 64 systems that took part.14
Two details from 2022 bear on what schools can do. The first is that creative-thinking scores track performance in the core domains: on average across OECD countries, the within-country correlation was 0.67 with mathematics and 0.66 with reading. The two deficits are therefore related, though they are not the same deficit, and the OECD notes that a student can excel in creative thinking without excelling academically. The second concerns curiosity. Filipino students who reported being curious scored four to five points higher in creative thinking than those who did not, against about three points across the OECD, and the Philippine difference was the largest among participants. The OECD treats three points on that scale as a large difference. PISA 2025 did not test creative thinking, but the OECD reports that Filipino students’ self-reported curiosity declined between 2022 and 2025.15
The 2025 innovative domain is related to creative thinking but is not the same thing. It measures how well students learn and solve problems with computational tools. On its computational problem-solving scale the Philippine mean of 434 was below the OECD average of 500, and the Department of Education reports that 34 per cent of students reached baseline. Yet Filipino students did better here than their science results predicted. The Philippines was one of six systems in which students scored more than 30 points above that expectation. The OECD will publish further analysis of the domain in 2027. Nothing in the 2025 results tells us whether the creative-thinking gap has moved since 2022.16
7 · The input ledger
Three weeks after the PISA release, the 2026 Global Innovation Index placed the Philippines 52nd of 139 economies, down from 50th. The fall came from the input side. The innovation-input rank dropped from 59th to 69th while the output rank held at 49th. The weakest of the index’s seven pillars was human capital and research, at 100th, down from 90th.17
The two reports measure different things, but they meet at one point. The index’s education sub-pillar includes PISA scores, and the 2026 edition carries a PISA average of 352.5 points, consistent with the 2022 results rather than the 2025 ones. The 2025 gains should register in later editions. They will not move the pillar far on their own, because the same pillar records a secondary pupil–teacher ratio ranked 112th, research spending of 0.28 per cent of GDP in the latest year available, 2020, and an output of scientific and technical articles ranked 133rd of 139.18
Read together, the indices describe an economy whose measured outputs run ahead of its formation of capability. WIPO’s commentary on the Philippines and Viet Nam says that their international production strengths still exceed their domestic capacity to generate knowledge and invent. PISA records the beginning of that pipeline, and the index records its end.19
8 · What would settle it
A gain produced by sustained change in classrooms and a gain produced by a well-run campaign look the same in a single cycle. They diverge afterwards. Four observations in PISA 2029, and in the national assessments around it, would favour the structural reading.
The opposite observations would favour the campaign reading. That would be less a failure than a measurement of what a campaign can do. Either result is worth having.
9 · The Institute’s questions
The Institute’s science, Creatology, treats creativity and innovation as one continuous phenomenon, from the forming of an idea to its use. The PISA results bear on that work at two points: the floor of foundational literacy on which inventive work depends, and the unmeasured domain above it.
LICI’s measurement instruments, the Substrate Cognition Inventory, the Throughput Operations Assessment, the Pedagogical Innovation Readiness Scale and the Organisational Creative Capacity Index, are in active development and have no psychometric record yet. Nothing in this note rests on them. They are designed to give profiles for cultivation, not scores for ranking students. When they are ready, the first questions will be the ones the 2025 data leave open: how creative capability is distributed across the socio-economic quarters, and whether the schools that produce academically resilient students also produce inventive ones.
The Second Congressional Commission on Education (EDCOM 2) has said it will examine how different groups of learners are progressing. The questions above are offered as a contribution to that work.20
Figures are taken from the sources below, and responsibility for the reading rests with the team.
Notes
Sources
Creation as discipline · Innovation as practice
