Theme 21 of 43
Corruption, Administrative Misconduct & Public Oversight
Until 2021, less than half the amount committed to parliamentary earmarks was disbursed in the same year. In 2025, R$ 7 billion went out as special transfers, the "Pix earmarks" that reach municipalities with no defined purpose. Today 14 thousand people and companies are barred from contracting with the government.
The Office of the Comptroller General (CGU) publishes every parliamentary earmark, with author, function, action and execution, and keeps the registers of punished companies and individuals (CEIS and CNEP) and the leniency agreements. The Federal Court of Accounts (TCU) lists those whose accounts were judged irregular and the companies declared unfit, and the National Council of Justice (CNJ) gathers convictions for administrative misconduct. Together, these sources show how public money is allocated, where transparency fails and how much punishment actually reaches those who diverted funds.
Budget execution: how much is left in unpaid carryovers?
Year after year, the government commits far more than it actually disburses. In 2019, only 44% of the committed amount became liquidated spending; even in more recent years, with better execution rates, between 30% and 40% of the authorized budget still goes without full execution. That gap doesn't disappear — it accumulates as "restos a pagar" (unpaid carryovers), a spending reserve that can be used in following years without going through Congress's scrutiny again.
| Year | Committed (R$ bi) | Liquidated (R$ bi) | Execution Rate |
|---|---|---|---|
| 2018 | 12.0 | 5.6 | 46% |
| 2019 | 13.9 | 6.1 | 44% |
| 2020 | 37.5 | 18.2 | 49% |
| 2021 | 33.4 | 16.0 | 48% |
| 2022 | 25.5 | 17.2 | 68% |
| 2023 | 35.4 | 22.1 | 62% |
| 2024 | 44.8 | 31.5 | 70% |
| 2025 | 47.1 | 31.7 | 67% |
Until 2021, less than half the amount committed to earmarks was disbursed in the same year. The rest becomes carryover and is spent later, outside the budget voting cycle.
Concentration of earmarks: who controls the budget?
Three parliamentary committees and the Budget's general rapporteur, together, decide the fate of more than R$ 30 billion in earmarks. The Health Committee alone, with just 10 earmarks, moves R$ 9.6 billion — nearly R$ 1 billion per earmark on average. This volume of resources concentrated in very few hands exceeds the annual budget of several entire Brazilian states.
| Author | Earmarks | Total Value (R$ bi) | Average Value |
|---|---|---|---|
| Health Committee | 10 | R$ 9.6 bi | R$ 956 mi |
| Regional Development Committee | 12 | R$ 8.6 bi | R$ 720 mi |
| General Rapporteur | 23 | R$ 8.6 bi | R$ 375 mi |
| Social Affairs Committee | 8 | R$ 3.2 bi | R$ 399 mi |
Three committees and one general rapporteur alone control R$ 30 billion in earmarks — more than the entire annual budget of several Brazilian states.
General Rapporteur's earmarks: health comes first
A single budget line — the temporary boost to Primary Care, decided by the Budget's general rapporteur — receives nearly R$ 4 billion, a value close to a full month of Bolsa Família nationwide. The other actions funded by the rapporteur, such as SUAS and integrated local development, get far smaller fractions of that amount, showing how much a single individual decision can direct colossal budget sums.
| Action | Value (R$ bi) |
|---|---|
| Temporary boost to Primary Care | R$ 3.98 bi |
| SUAS strengthening | R$ 0.96 bi |
| Integrated local development | R$ 0.17 bi |
| Basic education infrastructure | R$ 0.15 bi |
| Primary Care network | R$ 0.12 bi |
A single health action decided by the general rapporteur receives about R$ 4 billion — nearly the equivalent of a full month of Bolsa Família for the entire country.
Pix earmarks: R$ 7 billion with no defined purpose
In 2025 parliamentary earmarks committed R$ 47.1 billion. Individual earmarks with a defined purpose added up to R$ 17.3 billion, state-caucus earmarks R$ 11.6 billion and committee earmarks R$ 11.2 billion. Another R$ 7.0 billion went out as special transfers, the so-called Pix earmarks: the money lands directly in the municipal or state treasury, with no agreement and no linked project. In 2024 the Supreme Court began requiring a work plan and a dedicated account for these transfers, because the money trail was lost after the transfer.
| Earmark type | R$ billion | % |
|---|---|---|
| Individual, transfer with defined purpose | 17.3 | 36.7% |
| State caucus | 11.6 | 24.6% |
| Committee | 11.2 | 23.8% |
| Individual, special transfer ("Pix earmark") | 7.0 | 14.9% |
| Total | 47.1 | 100% |
One in every seven earmark reais in 2025 was transferred with no linked project. Oversight of that money depends on each city hall rendering accounts.
Sectoral concentration of earmarks
More than half of all parliamentary earmark value — 51.8%, or R$ 79.2 billion — goes to the Health function. Social Assistance, an area that directly serves the most vulnerable population, receives just 2.6% of the total, about fifteen times less than Special Charges. The distribution by function reveals where Congress chooses to concentrate political bargaining power, and that choice doesn't always reflect the areas of greatest social need.
| Function | % of Total | Value (R$ bi) |
|---|---|---|
| Health | 51.8% | R$ 79.2 bi |
| Special charges | 16.8% | R$ 25.6 bi |
| Urban planning | 7.6% | R$ 11.6 bi |
| Agriculture | 4.4% | R$ 6.7 bi |
| Education | 3.6% | R$ 5.6 bi |
| Social Assistance | 2.6% | R$ 3.9 bi |
| Public Security | 1.4% | R$ 2.1 bi |
More than half of all parliamentary earmarks go to health; Social Assistance, which serves the most vulnerable, receives fifteen times less than Special Charges.
Half the corporate card value is spent without saying to whom
Brazil's transparency portal publishes every federal government payment card transaction since 2013: 1,666,696 operations totalling R$ 820.1 million. Publishing, however, is not the same as disclosing. In 18.9% of transactions the recipient field reads "Classified" — and because classified transactions are far larger than average, they account for 49.5% of all value spent. That is, half the money on the federal corporate card is published without revealing who received it. The opacity is not evenly spread: the Presidency (98.6% of value classified) and the Ministry of Justice and Public Security (86.0%) account for nearly all of it, while Education, Health, Planning and Agriculture report 0.0%.
| Ministry | Total value | % of value classified |
|---|---|---|
| Presidency of the Republic | R$ 227.8m | 98.6% |
| Ministry of Justice and Public Security | R$ 205.1m | 86.0% |
| Ministry of Finance | R$ 20.4m | 22.3% |
| Ministry of Defence | R$ 71.6m | 0.2% |
| Ministry of Planning and Budget | R$ 81.6m | 0.0% |
| Ministry of Education | R$ 78.4m | 0.0% |
| Ministry of Health | R$ 20.4m | 0.0% |
The two largest corporate-card spenders account for half the value and hide the recipient in almost all of it — while ministries spending comparable sums publish 100% of their payees.
Three in four purchases skip bidding — but add up to 4% of the money
Counting public purchase processes and counting public purchase money produce opposite pictures, and using only the first leads to the wrong conclusion. Between 2020 and 2023, 60.5% of federal processes were bidding waivers and 14.6% were non-enforceability declarations — that is, three in four purchases were made without competition. The alarm dissolves in the value, however: waivers and non-enforceability together moved R$ 154.4 billion against R$ 3.69 trillion through competitive electronic bidding. Translated, 75.1% of processes correspond to 4.0% of the money. Waivers are the instrument of small recurring purchases; competitive electronic bidding is where 95.4% of spending passes. Anyone auditing risk should look at value, not count.
| Procedure | % of processes | Value | % of value |
|---|---|---|---|
| Bidding waiver | 60.5% | R$ 83.0bn | 2.1% |
| Non-enforceability | 14.6% | R$ 71.4bn | 1.8% |
| Electronic bidding — price registry | 14.6% | R$ 3,619.8bn | 93.5% |
| Electronic bidding | 9.7% | R$ 71.5bn | 1.8% |
| Open competition and others | 0.5% | R$ 25.2bn | 0.7% |
Waivers and non-enforceability dominate the process count and almost vanish in value — the big money passes through open competitive bidding.
Sanctions: 14 thousand barred from contracting with the government
In October 2026 the Register of Unfit and Suspended Companies (CEIS) held 19,151 sanctions against 7,277 companies and 6,982 individuals. Three in four are fixed-term bans on contracting, imposed mainly by states (8,931) and the federal government (6,442). The National Register of Punished Companies (CNEP), under the Anti-Corruption Law, held 1,630 sanctions and R$ 2.74 billion in fines. The register also lists 52 leniency agreements with companies that admitted wrongdoing.
| Category | Sanctions |
|---|---|
| Fixed-term ban on contracting | 14,373 |
| Suspension | 2,008 |
| Unfitness declaration, no fixed term | 1,383 |
| Unfitness declaration, fixed term | 1,147 |
| Ban on contracting, no fixed term | 152 |
| Dismissal | 61 |
| Total (incl. other categories) | 19,151 |
The most common punishment is temporary: three in four sanctions expire, and the company may bid for contracts again.
Misconduct: 53 thousand convictions, R$ 9.3 billion to repay
The CNJ's National Register of Convictions for Administrative Misconduct held 53,342 convictions of 41,558 people registered between 2009 and February 2024. Registrations grew from about 1,400 a year up to 2015 to more than 9 thousand in 2023. Of the convictions, 20,069 (38%) involved public servants, 83% of them municipal. The rulings add up to R$ 9.27 billion in repayment to the treasury, R$ 3.77 billion in fines and R$ 1.13 billion in asset forfeiture, after discarding impossible placeholder values such as R$ 888 billion.
| Registration year | Convictions |
|---|---|
| 2015 | 1,562 |
| 2017 | 3,334 |
| 2019 | 4,487 |
| 2020 | 5,207 |
| 2021 | 4,630 |
| 2022 | 8,705 |
| 2023 | 9,136 |
Eight in ten public servants convicted of misconduct are municipal. City hall is where oversight arrives last.
TCU: 48 thousand accounts judged irregular
The Federal Court of Accounts keeps 48,249 records of officials whose accounts were judged irregular, 9,340 of them with electoral implications: those on that list can have their candidacy blocked under the Clean Record Law. Another 850 are barred from holding appointed posts, and 123 companies have been declared unfit to bid for federal contracts.
| List | Records |
|---|---|
| Accounts judged irregular | 48,249 |
| Irregular accounts with electoral implications | 9,340 |
| Barred from public office | 850 |
| Unfit companies | 123 |
The TCU list with electoral implications links audit to the ballot box: 9,340 names the Electoral Court must check at every candidate registration.
Powerful crossings
- Earmarks × Execution: until 2021 less than half of the committed amount was disbursed in the year; in 2024-2025, about two-thirds.
- Rapporteur × Concentration: three committees and the general rapporteur concentrated R$ 30 billion in earmarks.
- Pix earmarks × Traceability: R$ 7.0 bn in special transfers in 2025, with no linked project.
- Corporate card × Secrecy: 18.9% of transactions are confidential, but 49.5% of the value.
- Process × Value: 75.1% of purchases skip bidding but account for 4.0% of the money.
- Sanction × Term: 75% of the 19 thousand CEIS sanctions are temporary bans.
- Misconduct × Municipality: 83% of public servants convicted of misconduct are municipal.
- TCU × Election: 9,340 officials with irregular accounts and electoral implications.
Explanatory hypotheses
The growth of parliamentary earmarks, from R$ 12 billion in 2018 to R$ 47 billion in 2025, shifted part of the spending decision from the Executive to Congress, and with it political bargaining came to revolve around funds spent in municipalities. Special transfers and rapporteur earmarks show that when the design reduces traceability, money grows faster than oversight. At the punishment end, most sanctions are temporary and misconduct convictions concentrate on municipal servants, where audit courts and prosecutors reach first, but where the sums are smallest.
Policy implications
Requiring every earmark, including special transfers, to carry a public work plan and a dedicated bank account makes it possible to follow the money to the final supplier. Automatically and openly cross-checking CEIS, CNEP and the TCU lists against contracts and candidate registrations prevents punished companies and people from contracting or running again for lack of checks. Limiting corporate card secrecy to expenses genuinely tied to security brings what is published closer to what is revealed. And publishing the CNJ register with continuous updates, since the available version ends in February 2024, would give society a current picture of who has been convicted.