Theme 20 of 43
Science, Technology, Scholarships & Academic Output
In PISA 2025, 72% of Brazilian 15-year-olds scored below the basic level in mathematics, the same average as ten years earlier. Between 2015 and 2021 the amount CNPq paid in research grants fell by half, even before adjusting for inflation.
The OECD's PISA tests 15-year-olds from nearly a hundred countries and economies in mathematics, reading and science every three or four years. CNPq, the national research council, publishes every research grant it pays, with field, institution and state. CAPES records international mobility scholarships, including those of the Science without Borders program. Together, the three sources show basic education stuck at the bottom of the international scale and research funding that shrank, concentrated in the Southeast and swung with programs that start and stop.
PISA: ten years in the same place
Brazil's average in mathematics was 377 points in 2015, 383 in 2018, 379 in 2022 and 377 in 2025. Reading and science follow the same pattern, moving by a few points, within the test's margin of error. The pandemic, which pulled down scores in many countries in 2022, barely moved Brazil's, which was already close to the floor.
| Edition | Mathematics | Reading | Science |
|---|---|---|---|
| 2015 | 377 | 407 | 401 |
| 2018 | 383 | 413 | 404 |
| 2022 | 379 | 410 | 403 |
| 2025 | 377 | 408 | 409 |
Over ten years and four editions, Brazil has not moved in mathematics.
72% below basic in mathematics, 71st out of 90
Level 2 in mathematics is the threshold the OECD considers the minimum for solving simple everyday problems. In 2025, 71.6% of Brazilian students scored below it (73.4% in 2022, 70.0% in 2015). On the mathematics average, Brazil ranked 71st out of 90 countries and economies, behind Uruguay (404), Chile (403), Mexico (388), Peru (382) and Colombia (381), and ahead of Argentina (367).
| Country | Average |
|---|---|
| Singapore | 563 |
| South Korea | 523 |
| United States | 463 |
| Portugal | 460 |
| Uruguay | 404 |
| Chile | 403 |
| Mexico | 388 |
| Peru | 382 |
| Colombia | 381 |
| Brazil | 377 |
| Argentina | 367 |
Seven in ten Brazilian 15-year-olds do not reach the minimum in mathematics. The country's science recruits its researchers from the remaining three.
CNPq grants: half the value in six years
CNPq paid R$ 2.64 billion in grants in 2015. In 2019 it paid R$ 1.37 billion, and in 2021 R$ 1.18 billion: a 55% drop in nominal terms, before adjusting for inflation over the period. The amount rose again in 2022, to R$ 1.70 billion, still 36% below 2015.
| Year | R$ billion |
|---|---|
| 2015 | 2.64 |
| 2016 | 2.07 |
| 2017 | 1.48 |
| 2018 | 1.46 |
| 2019 | 1.37 |
| 2020 | 1.30 |
| 2021 | 1.18 |
| 2022 | 1.70 |
In six years the country's main funder of research grants came to pay less than half as much, before inflation is even taken into account.
Where the money goes: 52% to the Southeast, 3.6% to the North
In 2022 institutions in the Southeast received 52.2% of CNPq grant money, against 41.8% of the population. The North, with 8.5% of the population and the country's greatest biodiversity, got 3.6%. The Northeast, with 26.9% of the population, received 15.5%. The distribution follows the location of the oldest universities and graduate programs.
| Region | R$ million | % of grants | % of population |
|---|---|---|---|
| Southeast | 886.2 | 52.2% | 41.8% |
| South | 315.5 | 18.6% | 14.7% |
| Northeast | 263.3 | 15.5% | 26.9% |
| Center-West | 130.5 | 7.7% | 8.0% |
| North | 60.8 | 3.6% | 8.5% |
Per resident, the Southeast receives three times as much CNPq grant money as the North.
By field: life sciences, exact sciences and engineering take more than half
In 2022 the biological sciences received 19.6% of grant money, exact and earth sciences 17.6% and engineering 16.9%. Adding health (13.2%) and agricultural sciences (11.1%), natural science and technology fields took 78%. Humanities, applied social sciences and languages added up to 15.8%.
| Broad field | R$ million | % |
|---|---|---|
| Biological sciences | 333.4 | 19.6% |
| Exact and earth sciences | 298.9 | 17.6% |
| Engineering | 287.5 | 16.9% |
| Health sciences | 224.7 | 13.2% |
| Agricultural sciences | 188.9 | 11.1% |
| Humanities | 152.7 | 9.0% |
| Applied social sciences | 77.1 | 4.5% |
| Linguistics, letters and arts | 38.4 | 2.3% |
Four of every five reais in CNPq grants go to natural sciences, health and engineering.
Science without Borders: from 28 thousand scholarships to 2 thousand in two years
CAPES international mobility scholarships went from 3,010 started in 2010 to 28,564 in 2014, at the peak of the Science without Borders program, and fell to 2,301 in 2016, a 92% drop in two years. From 2005 to 2019 there were 98 thousand scholarships, 56.5% of them for undergraduate exchange. The United States hosted 32.5% of the scholars, followed by France (14.4%), Germany (8.1%) and Portugal (6.7%).
| Year | Scholarships started |
|---|---|
| 2010 | 3,010 |
| 2012 | 9,398 |
| 2013 | 19,259 |
| 2014 | 28,564 |
| 2015 | 14,586 |
| 2016 | 2,301 |
| 2019 | 4,034 |
In two years the country went from nearly 29 thousand exchange scholarships to just over 2 thousand. Science policy made by program, not by career.
Powerful crossings
- PISA × stagnation: 377 points in mathematics in 2015 and in 2025.
- PISA × basic level: 71.6% below level 2 in mathematics in 2025.
- PISA × Latin America: Brazil trails Uruguay, Chile, Mexico, Peru and Colombia.
- CNPq × budget: from R$ 2.64 bn in 2015 to R$ 1.18 bn in 2021, in nominal terms.
- CNPq × region: Southeast with 52% of grants and 42% of the population; North with 3.6% and 8.5%.
- CAPES × discontinuity: mobility scholarships fell 92% between 2014 and 2016.
Explanatory hypotheses
PISA performance reflects the inequality of Brazilian basic schooling: the national average mixes state and municipal systems with very different resources, and the lack of change over ten years suggests reforms have not reached the classroom. In research funding, the spending cap approved in 2016 and the freezes that followed fell on discretionary items such as grants, which have no legal protection. Regional concentration follows the historical distribution of universities and of the best-rated programs, which attract more grants and reinforce their lead. Science without Borders shows the risk of large-scale policies without a guaranteed budget: it grew fast and was cut all at once.
Policy implications
In basic education, learning targets by school system and technical support for the worst-performing systems tackle the point where the national average gets stuck. In science funding, shielding grants from budget freezes and indexing their value to inflation would give predictability to those choosing a research career. Calls with regional quotas and support for building research groups in the North and Northeast reduce concentration without taking resources from established centers. And international mobility programs work better when they are permanent and tied to graduate programs, not to government cycles.