Fiscal Transparency Evaluation

This report assesses fiscal transparency practices in Brazil in relation to the requirements of the IMF Fiscal Transparency Code. Brazil’s practices meet many of the principles of the code at good or advanced levels. Fiscal statistics encompass the general government sector and recognize most of its assets and liabilities. Fiscal reports are published frequently and in a timely manner, and annual financial statements are audited by the Federal Court of Accounts. The transparency of fiscal forecasting and budgeting benefits in some areas from good or advanced practices. The institutional scope of budget documentation is comprehensive, and the key sequences of the budget process are timely.


This report assesses fiscal transparency practices in Brazil in relation to the requirements of the IMF Fiscal Transparency Code. Brazil’s practices meet many of the principles of the code at good or advanced levels. Fiscal statistics encompass the general government sector and recognize most of its assets and liabilities. Fiscal reports are published frequently and in a timely manner, and annual financial statements are audited by the Federal Court of Accounts. The transparency of fiscal forecasting and budgeting benefits in some areas from good or advanced practices. The institutional scope of budget documentation is comprehensive, and the key sequences of the budget process are timely.

I. Fiscal Reporting

1. Fiscal reports should provide a comprehensive, timely, reliable, comparable, and accessible summary of the government’s financial performance, financial position, and cash flows. This chapter assesses the quality of Brazil’s fiscal reporting practices against the standards set by the IMF’s Fiscal Transparency Code for the following dimensions:

  1. coverage of public sector institutions, stocks, and flows;

  2. frequency and timeliness of reporting;

  3. quality, accessibility, and comparability of fiscal reports; and

  4. reliability and integrity of reported fiscal data.

2. In the past decades, Brazil has made substantial improvements in government accounting and fiscal statistics compilation. With the introduction of the FRL, fiscal reporting became a central input to fiscal policy discussions, and numerous initiatives were launched to increase the transparency and reliability of fiscal data. In the public accounting area, most of these initiatives have been related to the gradual implementation of IPSAS-based accounting standards since 2008, which allowed Brazil’s government units to produce increasingly more comprehensive accrual-based financial statements. In parallel, fiscal statistics evolved from pure cash-based reports, covering only central government units, to a report consolidating all general government subsectors and presenting comprehensive information on revenue, expenditure and financing on both cash and noncash3 bases, as well as stock positions and other economic flows of the most relevant financial assets and liabilities. Following the recommendations of the G20 Data Gaps Initiative, the Treasury started reporting general government data on a quarterly basis in April 2016, placing Brazil in the group of most advanced countries in GFS reporting. Many of the advances have yet to be reflected in the main reports used by the public and policy makers (e.g., Treasury report and Central Bank statistics on the public sector).

3. Brazil publishes a large volume of fiscal data through a number of different documents, resulting in a fragmented reporting framework. In part, this results from the need to comply with legal requirements and adhere to international dissemination standards. Brazil’s various fiscal reports cover different institutions, include different flows and stocks, are prepared on different accounting bases, and are classified according to different standards. Brazil’s main summary fiscal reports, presented in Table 1.1, comprise:

  • monthly financial statistics, including the Treasury’s data on the pure cash revenue and primary expenditure of the federal government and the Central Bank’s data on the financing of the nonfinancial public sector (excluding Petrobras and Eletrobras) and associated stock positions;

  • monthly debt management reports, published by the STN’s Debt Management Unit, comprising detailed information on stocks and flows (issuance/redemption) of Treasury securities, including breakdowns by counterpart residency, type of interest rate, currency, maturity, and average cost; they include data on both the primary and secondary markets, including main groups of investors;

  • budget execution reports, including a bi-monthly update of the annual budget estimates, and a four-monthly assessment of compliance with the fiscal target;

  • revenue collection reports, published by the Federal Revenue Authority, containing data on the monthly collection of taxes and social contributions and annual estimates of tax burden and revenue loss from tax expenditures;

  • annual government finance statistics, jointly published by the Treasury and IBGE, which provide the most comprehensive data on stock positions of financial assets and liabilities and cash/accrual flows of general government and its subsectors. These data are compiled and disseminated in the GFSM 2001/2014 and 2008SNA frameworks, and include a disclosure of other economic flows for the financial instruments reported in the balance sheet A preliminary version of these data is made available by the Treasury on a quarterly basis;

  • annual report on public corporations, published by the Ministry of Planning’s Department for Coordination and Governance of State Enterprises (DEST), comprising the budget execution and annual financial statements (individual and aggregated) of the non-dependent financial and nonfinancial corporations of the federal level;

  • annual financial statements, comprising the Federal Government’s Financial Statements (Balanço Geral da União, BGU) and the consolidated4 financial statements of the Federal, State, and Municipal governments (Balanço do Setor Público Nacional, BSPN).

Table 1.1.

Brazil: List of Fiscal Reports

article image
Note: Fin = Financing, R =Revenue, E = Expenditure, A = Assets, L = Liabilities, M = Monthly, Q = Quarterly, An = Annual.

1.1. Coverage of Fiscal Reports

1.1.1. Coverage of Institutions (Good)

4. In 2014, Brazil’s public sector comprised 7,231 separate institutional entities of various legal forms. These can be broken down into the following subsectors (Table 1.2):5

  • budgetary central government, which comprises 459 legislative, judiciary, and executive bodies including 109 direct administration units (ministries and departments), 214 agencies, 28 non-commercial enterprises, 47 foundations, and 61 funds. The latter include the two social security funds: the Regime Próprio de Previdência Social (RPPS), covering public sector employees and the Regime Geral de Previdência Social (RGPS), covering private sector workers;

  • extrabudgetary central government, which comprises 5 constitutional funds, 9 nonprofit institutions controlled by government, 23 professional councils, 3 public corporations providing services primarily to government units, 2 public corporations operating on a nonmarket basis, and 1 defeasance structure (EMGEA);

  • subnational governments, which comprise the legislative, judiciary and executive bodies of 27 states (including the Federal District) and 5,570 municipalities;

  • public corporations. These are mainly nonfinancial corporations, which comprise Petrobras, Eletrobras, and another 59 federal corporations, 394 state corporations, and 664 municipal corporations. In addition, there are 12 financial corporations, including Banco do Brasil, BNDES, Caixa Económica Federal, and 3 other federal public banks, 6 regional development banks controlled by States, and the Brazilian Central Bank.

Table 1.2.

Brazil: Public Sector Institutions and Finances, 2014

(in percent of GDP, unless otherwise stated)

article image
Source: IMF staff estimates.

5. Brazil’s public sector includes a sizeable presence of subnational governments and public corporations (Table 1.2).6 Since Brazil is a decentralized federation, it is not surprising that the central government represents less than half of the general government expenditures (net of intra-government transfers) in 2014. State governments expenditures account for around ⅓ of total general government, while local governments account for around 20 percent. At both state and local levels, the estimates exclude expenditure of non-dependent enterprises. Brazil’s public sector also includes a strong presence for nonfinancial corporations, to a large extent reflecting the two largest companies Petrobras and Eletrobras.

6. Brazil’s most comprehensive fiscal reports cover most of the consolidated general government, as defined in international statistical standards (GFSM 2001/2014). The units covered in the Treasury’s annual general government financial statistics accounted for 63 percent of public sector expenditure in 2014. The Treasury is gradually introducing in the statistics the units comprising the extrabudgetary central government subsector; so far, only Fundo de Garantia do Tempo do Serviço (FGTS) and the Fundo Remanescente PIS/PASEP are covered. The missing units (due to source data unavailability) are small in size, accounting for less than 1 percent of the central government expenditure (Figure 1.1.).

Figure 1.1.
Figure 1.1.

Brazil: Coverage of Public Sector Institutions in Fiscal Reports

(in percent of expenditure)

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Source: IMF staff estimates.Note: “Not Reported” refers to expenditures of units not consolidated in summary fiscal reports.

7. Expanding the institutional coverage of Brazil’s fiscal reports from the general government to the entire public sector (including financial corporations) results in a slightly higher deficit in 2014. Including nonfinancial public corporations would increase the deficit by 1.3 percent of GDP, mainly reflecting Petrobras and Eletrobras. While adding both nonfinancial and financial corporations (including the Central Bank) results in a public sector deficit of 5.8 percent of GDP.

1.1.2. Coverage of Stocks (Good)

8. The federal government’s annual financial statements (BGU) includes a comprehensive coverage of financial assets and liabilities, but only a partial assessment of nonfinancial assets. Brazil’s balance sheet coverage goes beyond that of most emerging and many advanced countries, by including the value of both financial assets and liabilities (including part of civil servants’ pension liabilities) and parts of the nonfinancial assets in the BGU. The 2014 BGU balance sheet provides a detailed breakdown of the federal government assets and liabilities, amounting to:

  • 13.5 percent of GDP in nonfinancial assets including 13.1 percent of GDP in buildings, structures, and land; however, it does not include estimates of subsoil assets (see below).

  • 46.1 percent of GDP in financial assets including 5.4 percent of GDP in shares and equity of corporations, 18 percent of GDP in currency and deposits, and 19 percent of GDP in loans granted mainly to states and municipalities;

  • 86.5 percent of GDP in liabilities including 59.1 percent of GDP in debt securities, 21.3 percent of GDP in pension entitlements of civil servants, and 5.9 percent of GDP in accounts payable;7

  • an overall net worth and net financial worth of 26.9 and 40.4 percent of GDP, respectively.

Figure 1.2.
Figure 1.2.

Brazil: Public Sector Gross Liabilities in Selected Countries

(in percent of GDP)

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Source: IMF staff estimates.

9. The financial statements of the general government (BSPN) understate the value of some assets and liabilities. On an annual basis, the STN consolidates the BGU with the financial statements of states and municipalities. The balance sheet resulting from this exercise expands the coverage of stocks, but still undervalues or omits material assets and liabilities of the public sector. As shown in Table 0.2 and Figure 1.3, consolidated public sector asset holdings are estimated to be at least 216 percent of GDP and its liabilities are estimated to be around 217 percent of GDP in 2014. The differences between these figures and those reported in fiscal reports, such as the financial statements or the general government financial statistics, reflect:

  • Brazil’s large reserves of natural resources are not recognized. In particular, the central government has unrecognized subsoil assets (petroleum, natural gas, iron ore, copper, bauxite, niobium, gold, and nickel) of around 87 percent of GDP, based on staff estimates, as discussed further in Section 3.2.6;

  • states and municipalities undervalue their holdings of land and other nonfinancial assets by at least 2.8 percent of GDP by virtue of the fact that (i) the units still do not fully reflect those assets in their balance sheets, and (ii) the valuation of those assets is not based on market values;

  • general government has unreported liabilities in the form of civil service pension entitlements (related to the pension systems of the personnel of the armed forces, states, and municipalities) accrued to date of around 58.8 percent of GDP;8

  • the general government has growing assets and liabilities under public private partnerships (PPPs) and concession contracts estimated by 2014 at 4.6 percent of GDP, split between central government (2.2 percent of GDP) and subnational governments (2.4 percent of GDP);

  • public corporations (including the Central Bank) have 107.2 percent of GDP in non-equity liabilities and 119 percent of GDP in assets.

Figure 1.3.
Figure 1.3.

Brazil: Public Sector Balance Sheet Coverage in Fiscal Reports, 2014

(in percent of GDP)

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Source: IMF staff estimates.

10. Expanding the coverage of the public sector balance sheet in accordance with international statistical standards would improve Brazil’s estimated financial position. Initial estimates of the public sector balance sheet suggest a net worth in 2014 of about -0.9 percent of GDP, 33 percent of GDP higher than currently reported. This increase comes mainly from the inclusion of unreported subsoil assets and revaluation of land in the balance sheet, which are larger than the additional liabilities from civil service pension entitlements. As in other resource-rich countries, the estimates for Brazil’s public balance assets are particularly difficult to calculate and subject to large year-to-year fluctuations reflecting the volatility of commodity prices. Brazil’s overall public sector is comparable to the average of other countries for which net worth estimates are available, but significantly lower than other resource-rich countries (Figure 1.4).

Figure 1.4.
Figure 1.4.

Brazil: Public Sector Net Worth in Selected Countries

(in percent of GDP)

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Source: IMF staff estimates.

11. If private sector employees’ pension entitlements were included in public sector liabilities, the overall public sector net worth would decrease to -209 percent of GDP. The treatment of pensions in the government’s accounts raises difficult issues in Brazil, as in many countries. At present, the financial assets of RGPS are included on the statistical balance sheet of general government, while the system’s liabilities are not. Although this asymmetric treatment follows international statistical standards, it presents the government’s fiscal position as better than it really is.

1.1.3. Coverage of Flows (Basic)

12. The coverage of flows in Brazil’s summary fiscal data varies significantly across reports. The monthly Central Government Primary Balance Bulletin, widely used for fiscal policy discussions, covers most cash inflows and outflows of the federal government. One notable exception is the flows resulting from the relationship between the Treasury and the Central Bank, as monetary authority, which are only disclosed as memorandum items (the recording of this relationship in the various reports is described in Annex 2, along with a discussion on its impact in fiscal transparency). The implementation of IPSAS contributed to a much wider coverage of flows with annual financial statements covering not only full cash flows but also most accrual data, including accounts payable/receivable, depreciation, revaluations of assets (including equity holdings) and liabilities, or accrual of pension entitlements of civil servants. The financing reports compiled by the Central Bank present comprehensive cash data, but accrual information is limited to accrual of interest and a subset of other economic flows whose classification is not always in line with the 2014 GFSM. Fiscal statistics disseminated in the IMF’s statistical databases on Brazil (Government Finance Statistics Yearbook and International Financial Statistics) have recently improved with the adoption of a modified-cash basis for the reporting of expenditure data. Revenue, however, continues to be reported only on a cash basis.

13. Nonetheless, significant accrued flows remain outside the most comprehensive fiscal reports. Most of these missing flows are associated with the limitations in coverage. The most relevant is the unreported annual net accrual of pension liabilities.9 Staff estimates, based on limited data, indicate that these transactions were 4.7 percent of GDP in 2014, but could vary significantly year-to-year.10 There are also annual investments in PPPs and concessions, which are estimated at 1.1 percent of GDP. Unlike the unitary cash payments to private suppliers associated with completed PPP projects, these accrued expenses are not reflected in statistics or accounts, given that IPSAS 32 has not yet been adopted. Finally, reports also exclude the depletion of subsoil assets, which would neutralize the increase in net worth deriving from the royalties received from extracting corporations.

14. The overall net fiscal impact of recognizing these additional accrued revenues and expenses would increase Brazil’s reported public sector deficit (Table 1.3). If the estimated public sector net lending/borrowing in 2014 included the accrual adjustments discussed above and other adjustments (mainly broader coverage) it would be around -11.6 percent of GDP. The nonfinancial public sector (NFPS) deficit would be 5.8 percentage points higher than currently reported in official documents. The key drivers of the difference are the accrued pension entitlements of public servants and accrual of PPP investments.11 In addition, including the deficit of the two largest nonfinancial public corporations also adds to the reported deficit by about 1⅓ percent of GDP. The financial public sector has a positive effect on the balance.

Table 1.3.

Brazil: Cash to Accrual Adjustments

(in percent of GDP, unless otherwise stated)

article image

Based on authorities reports (RTN, Nota Imprensa Bacen).

15. The more comprehensive coverage of accrued liabilities is a recent development. Financial statements of the federal government (BGU) prior to fiscal year (FY) 2015 did not reflect significant accounts payable to public financial corporations and use of extra-budgetary funds accumulated over time as a result of the overdue compensation for quasi-fiscal activities (the so-called “pedaladas”, see Box 1). Following the recommendation of the TCU on the 2014 financial statements,12 the federal government recognized the associated liabilities and subsequent payments in 2015, but accounts for previous years were not restated (the 2014 stock positions of assets and liabilities were revised only in the disclosure of the opening balance sheet for FY 2015). The joint Treasury/IBGE publication of GFSM 2014/2008SNA -based data has retroactively recorded the accrual of the missing expenditure. The BCB is also now regularly compiling the operations between the Treasury and extra-budgetary funds.13

Relations Between the Treasury and Public Banks: Unreported Liabilities

In 2015, in response to a demand by the Court of Accounts (TCU), the government settled a series of unreported liabilities—the so called “pedaladas.” These involved the three largest public banks (BNDES, Banco do Brasil, and Caixa Econômica Federal) and FGTS.

Although some of these unreported liabilities arose more than a decade ago, most have been incurred since 2013. The pedaladas included:

  • arrears on payments to BNDES for subsidies related to the BNDES PSI (Program to Sustain Investment) to allow for subsidized loans. The Treasury has extended about R$ 500 billion in loans to BNDES since 2008 to boost its lending capacity and also committed to subsidize loans to companies under the PSI. While the subsidy expense from BNDES was immediate, the government decided to delay by up to 24 months the recognition of its liabilities to BNDES, but reversed this decision following the TCU ruling.

  • arrears to Banco do Brasil to compensate for subsidies related to agricultural/rural credit lines, especially to support subsidized interest payments related to the Safra Agrícola (harvests). The TCU found there were no clear procedures to recognize or address the payment delays.

  • arrears with Caixa Econômica Federal and FGTS (managed by Caixa). The FGTS extended cash advances to the Union to cover expenses in a housing program (Minha Casa Minha Vida) and covered delays on social contributions. There were also delays in compensating for social benefits paid by Caixa as the agent of the government (e.g., Bolsa Família), which were registered in “below the line” statistics in 2014.

In economic terms (and according to the TCU) these delays functioned as a source of financing for activities of the central government by public banks. As noted by the TCU’s assessment of the financial statements of the government, these transactions did not follow the principles under the fiscal responsibility legislation and were not properly reflected in the fiscal and debt statistics. They lacked transparency as in most cases there was little public information on the fiscal impact of the operations. They also had economic consequences as the public banks’ capacity to lend to the private sector was reduced.


Stock of unrecorded Liabilities (Pedaladas)

(end of year, million reais)

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Source: Central Bank of Brazil

These liabilities were settled by the Treasury at the end of 2015, amounting to payments of R$ 72.4 billion or 1¼ percent of GDP (included interest). The Central Bank has released historic time series from 2002 onwards on the accrual and clearance of these delays, improving ex-post transparency. Following a request from the TCU, the Treasury is now producing bi-monthly reports on future payments due to BNDES.

16. One area of concern is the large increase in “Restos a Pagar” (RAP, unpaid commitments) between 2007–14, including some representing liabilities from more than a decade ago.14 RAP grew from around 5 percent of the central government budgetary expenditures to 13.4 percent at end of 2014 (close to 4 percent of GDP). In 2015, despite an effort to reduce those, they still remained high (12.3 percent of budgetary expenditures). The large size and the fact that some of the RAP refers to liabilities accrued more than a decade ago suggest the need for more transparency on these liabilities. This would help ensure appropriate classification between accrued and non-accrued liabilities and greater disclosure on how they are being addressed (e.g., explaining long delays in processing RAP, which has sometimes taken many years).

1.1.4. Coverage of Tax Expenditures (Good)

17. Every year, the government attaches to the PLDO and the PLOA a detailed report on the estimated central government revenue foregone from tax expenditures.15 In accordance with Article 14 of the FRL, the government releases an annex to the PLDO (Annex IV.11) which include a detailed list of tax expenditures for the next budget year and for two forward years, as well as different aggregates (by geographical area, sector, and tax). Ex post estimations of the individual impacts of central government tax expenditures are published on the Federal Revenue Authority website. Central government budget documents do not cover state tax expenditures—these are covered in the states’ respective LDOs.16

18. There are no limits on tax expenditures, although they need to follow legal rules. Article 14 of the FRL states that any tax expenditure should fulfill one of the following conditions: either it is proven that it does not affect the targets set in the LDO, or it should be accompanied with a compensatory tax measure. An annex to the PLDO (Annex IV.12) is devoted to reporting the tax measures that were implemented to compensate for the creation of tax expenditures in the previous budget. However, this compensatory system has had only a partial effect in containing the growth of central government tax expenditures, which expanded at the same time that the fiscal deficit deteriorated and the fiscal target was missed. In total, central government tax expenditures averaged 4.1 percent of GDP over the period 2011–2016 (Figure 1.5). Subnational tax expenditures could represent as much as 1 percent of GDP, though consolidated data is not available. Taking both central and subnational tax expenditures into account, Brazil ranks at a similar level as the regional average (Figure 1.6) and below advanced economies such as the United Kingdom or the United States.

Figure 1.5.
Figure 1.5.

Brazil: Revenue Loss from Tax Expenditures

(in percent of GDP)

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Source: Federal Revenue Authority.Note: Simulation of the impact of central government tax expenditures based on 2013 tax base.
Figure 1.6.
Figure 1.6.

Brazil: Revenue Loss from Tax Expenditures, Regional Comparison

(2012, in percent of GDP)

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Source: Inter American Center of Tax Administrations (2014), IADB and Federal Revenue Authority. It includes estimate for subnational tax expenditures.

1.2. Frequency and Timeliness of Fiscal Reporting

1.2.1. Frequency of In-Year Fiscal Reporting (Advanced)

19. Brazil’s federal government financial statistics are produced monthly with a 30-day lag and general government statistics are available on a quarterly basis. The cash-based Central Government Primary Balance Bulletins, which cover the budgetary central government (around 45 percent of total general government expenditure, excluding intragovernmental interest and grants), are produced on a monthly basis and within 30 days of the end of each month. Since April 2016, Treasury has been publishing quarterly general government data, on a modified cash basis (revenue on cash basis and expenditure in both cash and modified-cash bases) in the GFSM 2014 format for publication in the IMF’s International Financial Statistics. These data are disseminated with a lag of 90 days after the end of the quarter.

1.2.2. Timeliness of Annual Financial Statements (Advanced)

20. The opinion of the TCU on the “Accounts of the President” (PCPR) is usually published within six months of the end of the fiscal year. Article 84 of the Constitution provides that the PCPR should be submitted to Congress no later than 60 days after the beginning of the first Congressional session.17 Congress then transmits the PCPR to the TCU, which has 60 days to provide an opinion on whether the accounts present a fair view of the financial position of the government and comply with the constitutional and legal principles governing the federal public administration. This opinion is written and published by the TCU in a document titled “Report and Preliminary Opinion on the Accounts of the Government of the Republic.” This opinion aims at providing to Congress a technical analysis to support its final decision on the accounts approval.

Table 1.4.

Brazil: Timetable of PCPR Submission and Audit

article image
Source: Mission, Congress website18, and TCU website.

21. The TCU’s opinion on the 2014 PCPR was delayed due to the need to undertake additional investigations. On June 2015, the TCU requested clarifications on the 2014 accounts, to be provided within 30 days. The deadline was extended twice, after new findings were added to the process. The Presidency presented counterarguments, but they were not deemed sufficient to justify the irregularities. As a consequence, the Court recommended to reject the accounts on October 7, 2015—the first time this has occurred since 1937. Similar reasons explain the delay regarding the 2015 accounts, which the TCU also recommended to reject in October 2016.

22. The final approval of the accounts by Congress is not timely. This approval has a political nature and is not bound by the preliminary opinion of the TCU. The last accounts approved by both the Chamber of Deputies and the Senate are those from 2001.19

1.3. Quality of Fiscal Reports

1.3.1. Classification (Basic)

23. Brazil’s fiscal reports include administrative, economic, program, and functional classifications, but the latter are not consistent with international standards. National classifications are applied to all levels of general government, which is a notable achievement for a large country with a high degree of fiscal decentralization. Classifications for revenue and expenditure do not follow the structure of the GFSM 2001/2014 economic classifications, but are sufficiently detailed to allow a bridging to those standards.20 On the contrary, for the functional classification, there are several areas (e.g., research and development and foreign aid) where a breakdown is not available which would allow a direct conversion to the first level of the United Nations’ Classification of Functions of Government (COFOG). The Treasury intends to disclose expenditure by COFOG in the near future, and has been developing a compilation procedure that addresses the classification inconsistencies. This is being done primarily through a combination of the functional and administrative classifications.

1.3.2. Internal Consistency (Advanced)

24. The key internal consistency checks called for under the FTC are provided on a regular basis in the public fiscal reports. The above-the-line central government fiscal balance is computed by the Treasury, while the below-the-line calculation is provided by the Central Bank. The Treasury makes sure that both calculations are reconciled in the primary balance tables it publishes every month. The resulting discrepancy is broken down between a methodological adjustment (amortization of specific contracts) and a statistical component. In addition, the monthly debt reports published by the Treasury contain consistent information on public debt transactions and variations in public debt holdings from a unique database on securities held by the Central Bank. Last but not least, on a monthly basis, the Central Bank updates the breakdown of the explanatory factors behind the change in the general government gross debt stock.

25. Discrepancies have been contained over the past few years. Differences between above- and below-the-line calculations of the central government primary balance have been small, with an average absolute deviation of 0.1 percent of GDP (Figure 1.7). Stock-flow adjustments have averaged around 0.2 percent of GDP over the 2007–2015 period, mostly driven by the effects of exchange rate movements (including related to FX swaps) on the stock of debt (Figure 1.8).

Figure 1.7.
Figure 1.7.

Brazil: Deviation Between Below- and Above-the-Line Central Government Primary Balances, 1997–2015

(in percent of GDP)

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Source: Treasury, Resultado do Tesouro Nacional.
Figure 1.8.
Figure 1.8.

Brazil: Stock-Flow Adjustment, 2007–2015

(in percent of GDP)

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Source: Central Bank and IMF staff calculations.

1.3.3. Historical Revisions (Good)

26. All revisions to fiscal statistics and reports are applied retroactively, with changes qualitatively explained. Online databases of the monthly reports produced by both the Treasury and the Central Bank are updated retroactively whenever any change occurs to sources and methods. Nevertheless, an archive of old published reports remains available at those institutions’ websites. The same applies for fiscal statistics disseminated for publication in the IMF’s Statistical Department databases, which always contain an entire time series, fully updated to reflect the latest compilation procedures. All major changes are explained in the metadata accompanying the first release of revised data.

27. Information on major revisions does not usually include bridging tables between the old and new time series. The only case where a revision of methodology in fiscal statistics led to a disclosure of a bridging table, together with data to allow the conversion of the new time series into the old methodology, was the change in the Central Bank’s compilation of the General Government Gross Debt. For broader revisions, no bridging table was published (e.g. the recent change in the sources and methods for compilation of GFSM data, disclosed in the new joint Treasury/IBGE publication).

1.4. Integrity of Fiscal Reports

1.4.1. Statistical Integrity (Basic)

28. Fiscal statistics have been disseminated according to international dissemination standards since 2001. Brazil subscribed to the SDDS on March 14, 2001 and met all SDDS requirements at the time of subscription. Until 2008, Brazil’s in-year and year-end fiscal statistics covered only budgetary central government cash revenues, expenditures, borrowing, and debt, based on the GFSM 86 methodology. Since 2009, the Treasury—the agency in charge of GFS compilation—has gradually improved its compilation procedures, allowing it to start disclosing quarterly general government operations on both cash and modified-cash bases and a financial balance sheet, in the GFSM 2014 reporting format. With this, Brazil complied with the dissemination requirements set up under the G20 Data Gaps Initiative.

29. However, there is room for improvement in the implementation of methodological standards. Treasury’s compilation procedures are still significantly bound to the source data, which constrains its ability to perform reclassifications for alignment with the GFSM 2014 methodology (e.g., for classification of units or the statistical treatment of operations). The Treasury’s mandate as compiler of fiscal statistics, requiring it to align fiscal statistics to both international best practices and local requirements, results in a fragmentation of the reporting framework. Currently, there is no systematic analysis of consistency with the methodological guidelines. Statistical classifications follow the legal form rather than economic substance.

30. There is also room for improvement in the verification of fiscal statistics against external sources. The current compilation procedures are essentially based on the conversion of accounting data into the GFSM format. The resulting numbers could be checked against data that are compiled from different sources but have linkages with the government sector—in particular, external sector and monetary statistics, results from TCU audits, and financial statements of public corporations. Efforts in this direction were taken in the past—particularly with the setting up of a temporary inter-agency working group (Treasury, IBGE, and Central Bank) for the improvement and harmonization of macroeconomic statistics—but they have never been fully institutionalized.

1.4.2. External Audit (Good)

31. The TCU is the Supreme Audit Institution in charge of auditing the federal government’s financial statements. Based on the judicial model, it benefits from the constitutional protections and guarantees aiming at ensuring the independence of judges.

32. The TCU opinions, on the last three fiscal years’ accounts, include major qualifications. The “Report and Preliminary Opinion on the Accounts of the Government of the Republic” includes an audit of the BGU, which is part of PCPR. Regarding the fiscal year 2014, it expressed a qualified opinion based on 16 identified irregularities and 2 limitations on the scope of its audit.

Table 1.5.

Brazil: TCU’s Preliminary Opinions on the Accounts of the Government of the Republic

(fiscal years 2012–14)

article image
Source: “Report and Preliminary Opinion on the Accounts of the Government of the Republic,” 2012, 2013, and 2014.

33. The TCU is currently further strengthening its audit methodology through the implementation of international audit standards. The “Brazilian Standards for the Audit of the Public Sector – Level 1” (NBASP), based on International Auditing Standards of Supreme Audit Institutions (ISSAI) Levels 1 and 2, have been published in 2015. A “Manual of the Financial Auditor,” based on international standards, has been prepared by the TCU and was subject to public consultation in June 2016.

1.4.3. Comparability of Fiscal Data (Basic)

34. The breakdown table of the primary balance of the central government is the key presentation format that is provided and monitored across budget documents, in-year financial statistics, and budget execution reports. Forecasts and outturns according to this presentation can be found in the PLOA and in budget execution reports published by the Treasury. Recent efforts have been made to improve comparability between documents: the format of the Central Government Primary Balance Bulletin (“Resultado do Tesouro Nacional” (RTN)) has been revamped in 2015 to allow proper comparability with budget documents. Other FRL-mandated types of presentation, which distinguish between current and capital revenue and expenditure of the central government, are also consistently used in budget documents and the final accounts published by the Presidency. It can, however, be difficult for the public to understand differences across documents, as bridging tables that could explain differences in the scope or coverage of different documents are not provided.

35. Budget outturns, which have been slightly different from one report to another in recent years, are not reconciled with fiscal statistics or final accounts. There are no bridging tables which explain differences between the RTN, GFS, or BGU formats. Information on fiscal outturns can also slightly vary from one document to another, complicating any efforts at reconciliation. Such differences could be due to recent changes in the formatting of the documents or in the classification of the different revenue and expenditure items, but they are not quantified or mentioned in the budget documents and fiscal reports. Even though the discrepancies are quite limited (Table 1.6), they should be accounted for in fiscal reports to allow for a transparent understanding of outturns.

Table 1.6.

Brazil: Discrepancies Between Outturn Figures

(in R$ million)

article image

1.5. Conclusions and Recommendations

36. Brazil’s fiscal reporting meets good and advanced practices in most areas. The assessment against the FTC, summarized in Table 1.6, shows that fiscal statistics cover the general government sector and recognize most of its assets and liabilities. It also shows that fiscal reports are prepared frequently and in a timely manner, and that annual financial statements are subject to an audit by an independent supreme audit institution, whose opinion has contributed to improvements in the government’s accounting practices.

37. However, there remains scope to enhance the coverage and integrity of fiscal reporting in key areas. Brazil’s fiscal reports do not consolidate the large public corporations sector (the two largest nonfinancial corporations are excluded), whose expenditure accounts for 22.1 percent of GDP. The general government’s balance sheet excludes important elements, such as government employees’ pension entitlements of around 59 percent of GDP or subsoil assets of approximately 89.7 percent of GDP. The assessment also shows that the compilation of fiscal statistics is often constrained by classification decisions that do not always comply with international statistical standards (e.g., recording of the relationship between the Treasury and the BCB, as described in Annex 2).

38. Based on the above assessment, the evaluation highlights the following priorities for improving the transparency of fiscal reporting:

  • Recommendation 1.1: Expand the institutional coverage of fiscal reports by incorporating all public corporations and public banks to provide an overview of the fiscal performance of the entire public sector. This would provide a more comprehensive picture of the extent of government-directed activity in the economy. In particular, it would allow for a more accurate accounting of the cost of quasi-fiscal operations undertaken by public corporations and public banks.

  • Recommendation 1.2: Expand the coverage of (a) balance sheets to reflect the full market value of government infrastructure, subsoil assets, and pension entitlements of all public employees; and (b) flow statements to capture the associated expenses. This would provide a comprehensive view of the government’s overall net worth and the costs associated with transactions in nonfinancial assets and the accrual of pensions obligations.

  • Recommendation 1.3: Enhance the integrity of fiscal statistics by:

    • a. Setting up a permanent inter-agency committee for harmonized classifications in macroeconomic statistics. This would help ensure that economic substance prevails over legal form in the classification of institutional units and statistical treatment of public sector units’ operations;

    • b. Recording Treasury-Central Bank transactions according to the most recent international statistical standards. This would help better reflect transactions according to their true economic nature. It would also allow for separate reporting of transactions and the balance sheets of the BCB and the central government (a complex task at present), which are two separate units, classified in different parts of the public sector.

    • c. Conducting regular consistency checks between different macroeconomic datasets. This would help detect misclassifications or inconsistent recording of government operations with other sectors of the economy; and

    • d. Including reconciliations of key fiscal aggregates within and between the different fiscal reports. These would explain the reasons behind apparent inconsistencies between related key fiscal aggregates, such as stock flow adjustments or differences in coverage or accounting basis. This will become increasingly more important as the authorities gradually converge to GFSM 2014 standards.

Table 1.7.

Brazil: Summary Assessment of Fiscal Reporting

article image

II. Fiscal Forecasting and Budgeting

39. This chapter assesses the quality of fiscal forecasting and budgeting practices relative to the standards set by the FTC. It focuses on four main areas:

  • Comprehensiveness of the budget and associated documentation;

  • Orderliness and timeliness of the budget process;

  • Policy orientation of budget documentation; and

  • Credibility of the fiscal forecasts and budget proposals.

The assessment is based on information publicly available (Table 2.1). In addition to the multi-annual plan and budget documentation, the assessment takes into account freely available online databases that provide detailed information on budget authorizations and execution, including performance information.

Table 2.1.

Brazil: Documents and Data on Fiscal Forecasting and Budget

article image

2.1. Comprehensiveness

2.1.1. Budget Unity (Good)

40. The budget documentation presents gross revenue, expenditure, and financing for the central government, budgetary funds, and social security. The LOA includes the operations of central government and no expenditure can be made without explicit budgetary appropriation. Budgetary funds are created by law and are funded by earmarked revenue. Nevertheless, they are included in the LOA and are submitted to the common rules regarding budget preparation and execution. The administrative budget21 of the Central Bank is also presented in the LOA. Nevertheless, as the budget does not include extensive information related to FGTS, a large extra budgetary fund submitted to specific governance arrangements, the practice related to budget unity is not meeting the advanced standard.22

2.1.2. Macroeconomic Forecasts (Basic)

41. Three-year forecasts for key macroeconomic indicators are provided in several budget documents and execution reports, but explanations of the forecasts are limited. The PLDO and the PLOA (and the PPA before them) include, at a minimum, forecasts of real GDP growth, inflation, the real/dollar exchange rate, and short-term interest rates, covering the budget year and between one and three years forward. However, the scope of the forecasts presented can vary significantly across documents (Table 2.2). The PLOA complements the annual forecasts with monthly projections of imports, total wages, price indexes, and oil prices for the budget year. Though the analysis of the most recent outturns is fairly detailed in the Presidential message attached to the PLOA, only a short explanation of the macroeconomic scenario is provided to support the forecasts. Forward year forecasts are indicative and rely on extrapolation methods, rather than on detailed analysis. There is only a very brief mention of the expected impact of global economic developments on the Brazilian economy. Furthermore, the forecasts of the different components of GDP are not published (the Presidential message attached to the PLOA does contain both a demand and supply breakdown of GDP over the past few years).

Table 2.2.

Brazil: Scope of the Macroeconomic Forecasts Contained in Recent Budget Documents

article image
Source: Ministry of Planning.

42. More extensive discussion of the assumptions underlying the macroeconomic projections could improve their credibility, as they have shown an optimism bias in recent years. While the average deviation between one-year-ahead GDP forecasts and outturns was commensurable with independent forecast practice over the 2000s,23 it has been systematically negative since 2011, reaching -3.9 percentage points over the period 2011–2015 (Figure 2.1.). In part, this reflects the unexpectedly difficult macroeconomic environment in the last two years, but also reveals an optimism bias relatively to market projections for the main economic variables.24 Explaining the assumptions and choices which led to the forecast presented in the budget could help inform the public. Furthermore, the assumptions regarding new fiscal policy measures (which can have a significant impact on the economy, especially tax measures) should be communicated.

Figure 2.1.
Figure 2.1.

Brazil: Real GDP Growth: Deviation Between One-Year-Ahead Forecast and Outturn, 2001–2015

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Source: Ministry of Planning; WEO.

2.1.3. Medium-Term Budget Framework (Basic)

43. The Brazilian budget system does not include a rolling medium-term budgetary framework. The PPA is prepared by the Executive during the first year of a government and defines the main strategic targets and programs of the Federal Government for the next four years, including the first year of the next government. It includes the main economic assumptions and fiscal targets for the primary budget balance, debt levels, and nominal deficit. The PPA covers all activities and expenditures of the central government and provides a broad estimate of the resource allocated to each activity or project during the four-year period. De facto, it is a strategic planning tool relying on optimistic assumptions regarding both resource availability and implementation capacity. As a consequence, it does not effectively provide a policy anchor for the annual budget process. Annual budget preparation is informed by the LDO submitted every year to Congress. It includes economic assumptions, fiscal targets for the primary budget balance, debt levels, and revenue forecasts for the year of the budget and two years forward. Nevertheless, neither document presents expenditure estimates at the program or ministry level.

2.1.4. Investment Projects (Basic)

44. Brazil continues to face challenges in executing its public investment budget. On average, about two thirds of the investment budget has been executed over the past three years. Though due in part to the “Lava Jato” investigations in the recent years (Box 2), these shortfalls are macroeconomically important, as Brazil’s public investment is lagging behind other emerging market economies, both in quantitative and qualitative terms (Figure 2.2). Brazil’s public investment rate has been consistently below the average of emerging market economies, leading to a relatively low level of capital stock. The perceived quality of infrastructure is also inferior to the average for emerging economies and for countries in the Latin America and Caribbean region.

Transparency and Fiscal Risks: The Case of Petrobras and Eletrobras

Petrobras and Eletrobras are the two largest nonfinancial public corporations, but they have been excluded from the fiscal targets since 2009. The objective was to allow the companies, considered to pose low fiscal risk, more flexibility to conduct their large investment plans. They were also removed from fiscal statistics of the public sector (deficit, debt) weakening fiscal transparency. In addition, Petrobras has been exempted from several of the public sector procurement, borrowing, and personnel rules. The fiscal performance of the two companies has deteriorated since 2009. If they had been included in the fiscal accounts, they would have caused a further deterioration of the primary surplus of around ½ percent of GDP yearly on average during 2009–14.


Fiscal contribution of Petrobas/Eletrobas deteriorated after 2008 (Overall balance, yearly average)

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Both companies’ operations and financial health have been affected by corruption scandals:

  • In 2009, the federal police started an investigation of money laundering in several states, called operation “Lava Jato” (Car Wash). From 2014, the investigation focused on irregularities involving constructing companies and suppliers of Petrobras. A group of companies colluded to overcharge Petrobras with some of the funds being directed to illegally finance political parties, among others. The contracts affected had a value of about R$ 200 billion. Petrobras estimated that the overcharging represented about 3 percent of the value of the contracts (R$ 6.2 billion). Petrobras’ shares have suffered with the scandal and the company faces the risk of losses arising from lawsuits.

  • Several large projects by Eletrobras (and companies in the group) are being investigated by Brazilian prosecutors after executives in the projects were involved in corruption scandals (including “Lava Jato”). The company’s ADRs were suspended on the New York Stock Exchange and could be delisted for failing to submit audited financial statements to the Securities and Exchange Commission—the auditing firm has not signed the reports due to a lack of information on losses from corruption in its large projects (e.g., Belo Monte hydroelectric plant and the Angra 3 thermonuclear plant). The quality of the company’s governance has been downgraded (e.g., “weak” by S&P).

  • Both companies have adopted measures to improve procurement practices and internal controls, including an anticorruption plan at Eletrobras. The government has also approved a new law for public enterprises (2016) and has issued a series of directives in recent years to improve governance (including strengthening Boards), procurement rules, and controls of SOEs.


Petrobras: Rising indebtness in recent years

(Billion Reais)

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Source: Petrobras

The two companies represent significant fiscal risks to the rest of the public sector:

  • The large fall in international oil prices and high leverage—net debt almost quadrupled since 2011 to 6½ percent of GDP—has put Petrobras under financial distress. Eletrobras may also need further government financial support.

  • Neither company paid dividends in 2015 (after paying as much as 0.3 percent of GDP in 2009) and their problems contributed to the deepening economic recession (and fall in tax revenues) as they curtailed their investment plans. They also represent a risk for public banks that have granted them considerable loans and have equity participation in the companies.

Figure 2.2.
Figure 2.2.

Brazil: Selected Indicators of Public Investment

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

45. Detailed information on major investment projects is available online, but the total amounts of multiyear commitments attached to them are not published.

  • The PPA, published every four years, contains a list of ongoing and pre-selected investment projects, with a projection of their overall costs, as well as their planned beginning and end dates, classified by program and ministry.

  • The PLOA provides both an aggregate view and individual information on planned investment spending, though only amounts to be appropriated for the budget year are disclosed—information on multiyear commitments is not available in the budget documents. Though there is a clear separation between investment project expenditure and maintenance and operational costs, the latter are not presented in a project-specific manner.

  • Projects under the Programa de Aceleração do Crescimento (PAC), which account for approximately two-thirds of the federal government’s investment spending, are subject to deeper scrutiny, with the publication of total overall costs, monthly disbursement reports and regular information on expected and realized outputs, and sometimes, outcomes.

46. Project appraisal, selection, and procurement processes still fall short of best practices in terms of transparency. Investment projects do not systematically go through cost- benefit analyses before being selected, and when they do, these analyses are not made public and do not follow a common standard methodology. Public procurement procedures are regulated by a 1993 law25 which covers the federal government as well as states and municipalities. Open and competitive public tender is a core principle of this law, and complements the Constitution, which already states that public tender “should assure equal conditions for all bidders” (Article 37). Most tenders are available online via the government’s Portal da Transparência. However, ongoing judicial investigations, in the “Lava Jato” case for instance, indicate that there have been weaknesses in the procurement process in the recent past. Consequently, cost overruns have been frequent. The legal framework and public procurement processes are being strengthened to address some of these weaknesses: for instance, the new Public Companies General Law (June 30, 2016) clarifies the rules and procedures for public procurement driven by public enterprises.

2.2. Orderliness

2.2.1. Fiscal Legislation (Advanced)

47. The Federal Constitution, Law No. 4320 of 1964,26 and the FRL clearly set the budget timetable and its content. Article 165 of the Constitution requires that the executive branch prepares the draft PPA, LDO, and LOA. Article 35 of the Temporary Constitutional Provisions Act sets the deadlines for their transmission to Congress and to return it to the executive branch (Table 2.3). The Constitution and the FRL define the format and the content of the documentation supporting the budget process.

Table 2.3.

Brazil: Timetable for Budget Preparation and Approval

article image
Source: Constitution of the Federative Republic of Brazil.

48. The legal framework clearly defines Congress’ power to amend the executive branch’s budget proposals. Article 166 of the Constitution provides that amendments to the budget proposals should be compatible with the PPA and the LDO and specify their funding from either the cancellation of other expenditure or higher revenue than initially forecast. Nevertheless, this obligation is not always implemented. Some policy changes have been adopted while their financial cost or funding was not identified. In the case of the modification of the social security factor (fator previdenciário) in 2015 both the cost and offsetting measures were not identified. In the case of the increase of net revenue allocated to health expenditure, offsetting measures were not identified.

49. The fragmentation of the legal framework limits its accessibility to the public. The Constitution provides that a supplementary law must be enacted to detail the rules and procedures related to public finances. In effect, this would have provided an organic budget law, updating and replacing the 1964 law. As this new law was never adopted, the legal framework for the budget comprises not only the 1964 law (which is outdated in some aspects) and the FRL but also the LDOs, which include detailed legal provisions related to the public finances. As a result, the legal framework is excessively complex. Consolidating the legal framework into a single modern organic law would increase the transparency of the rules related to the budget and reduce legal uncertainties resulting from the extensive use of the LDOs to set up new procedures and rules every year.

2.2.2. Timeliness of Budget Documents (Basic)

50. The enacted budget has been published after the beginning of the fiscal year for the last two years. The last three budgets have been submitted to Congress four months before the beginning of the fiscal year. The 2014 and 2016 budgets were approved by Parliament in December. The 2015 budget approval and publication have been delayed, due to the October 2014 general elections. The 2016 budget was published in January 2016.

Table 2.4.

Brazil: Dates of Budget Submission, Approval, and Publication, 2014–2016

article image
Sources: Transmission to Congress: date of the presidential message; Approval by Congress: Senate website; Publication: Date of publication in the Diário Oficial da União.

2.3. Policy Orientation

2.3.1. Fiscal Policy Objectives (Good)

51. The PPA and the LDO are the two main tools used by the Brazilian government to set numerical fiscal objectives and to regularly report on its performance against them.

  • The PPA27 contains macroeconomic forecasts (Table 2.1.), as well as indicative fiscal objectives for the primary balance, the nominal balance, and debt levels. Every year, the Ministry of Planning releases an evaluation report, which discusses progress made and updates the forecasts and targets.

  • Brazil does not have a fully developed medium-term budgetary framework. However, it is bound by the guidelines set every year in the LDO.28 Brazil’s fiscal objectives for year t are stated every year in an annex (Annex IV.1) to the LDO drafted in April t-1. These are also compared with the outturns in another annex to the LDO (Annex IV.3), drafted in April t+1. The key fiscal target is the primary balance of the non-financial public sector (excluding Petrobras and Eletrobras), which drives the debates on fiscal policy. Targets for primary revenue and expenditure, the nominal balance and net debt are also set, but they are secondary in practice. The LDO also provides objectives for two forward years, but they are mostly rough indicative objectives and are not backed by robust economic and fiscal projections in published documents.

52. Brazil’s budget documents do not provide a clear picture of the fiscal strategy and how policy measures will help achieve the government’s objectives for fiscal policy. The description of the policy intentions of the government for the budget year and the medium-term, as well as the explanations accompanying the targets, remain limited. Furthermore, in-year large revisions to the targets through supplementary revisions, together with large adjustors to the main fiscal target, undermine the role of the LDO as providing clear guidance to the fiscal strategy.

53. The three last LDOs have been approved by Congress more than six months after the release of the drafts, highlighting the limits of the framework in a fast-changing macroeconomic environment. While the LDO is supposed to be approved by Congress before the draft budget (PLOA) is released in August, the LDO was approved just before the LOA for the last three budgets. Each time, the adopted LDO has been substantially different from the PLDO, with targets set to match the appropriations and forecasts of the LOA. As such, the LDO has thus not acted as a guide for the preparation of the budget, but rather as a legal formality which has not constrained the level of spending.

2.3.2. Performance Information (Good)

54. For more than a decade, the Brazilian government has been developing performance indicators and evaluations through its planning process and the PPA. The PPA includes for each of its programs high-level performance indicators, objectives, and output targets. All programs in the PPA are required to be evaluated each year. Program managers report on progress against PPA targets, using an online system (SIOP). The evaluations and indicator results feed into the annual PPA progress report, which is sent to the Executive and Congress. Since 2011, the PPA has been significantly revised and streamlined. The number of programs in the plan was reduced from over 300 to 54 thematic programs, covering policy areas. Outcome targets have been de-emphasized in favor of more output and activity based targets. In addition, there have been efforts to improve monitoring and reporting on output indicators included in budget documents. Despite the progress made with reporting performance information, its use in the budget decision-making process is reportedly still limited.

55. Budget documentation includes targets for the outputs to be delivered under each major government policy. The PLOA presents the outputs expected for each governmental programs at the detailed policy measure (“action”) level and links it with the budget allocation. An online register of the actions, the Cadastro de Ações, aims at providing qualitative information on the actions planned in the budget of each year. It comprises 54 thematic programs, 42 management and support programs, and 12 special operations programs.

2.3.3. Public Participation (Good)

56. Since 2010, the government has published a Citizens Budget. It is published alongside the PLOA. It presents, in a non-technical and visual manner, the core information expected from a citizens’ budget, as defined by the International Budget Partnership.29 It also includes information about the budget preparation and execution process. Moreover, it provides information on key inputs, outputs, or outcomes of policies per sector and clarifies the funding source of these policies. The implications of the budget proposal are presented from the point of view of a typical citizen, but the proposal does not provide detailed information for different demographic groups (for example, low-income households or indigenous populations).

57. Brazil fosters the public’s knowledge of budgetary matters through a number of other initiatives. An e-learning system, Escola Virtual SOF, provides on-line training and information to all interested citizens for free. In addition, a comic book, “Sofinha and her gang,” presents in a very accessible language the key information on the budget institutions and concepts. Furthermore, a radio program, Budget Time, provides information to the listener in a simple and direct language.

58. Brazil is a leading country in providing citizens with a formal voice in budget deliberations. The Intercouncil Forum (Forum Interconselho) comprises elected representatives of National Councils and committees and representatives of civil society. It aims at formally incorporating policy experts in the preparation of the PPA. The budget preparation benefits from the expertise of the National Councils, composed of civil society organizations (CSOs) and government representatives, that are in charge of formulating the budget within their policy area. Moreover, CSOs and citizens can contribute to the legislative process through their participation in public hearings.

2.4. Credibility

2.4.1. Independent Evaluation (Not Met)

59. There is no independent evaluation of the government’s macroeconomic and fiscal forecasts and performance. There is no independent fiscal institution30 that examines the accuracy of fiscal projections. Furthermore, no budget document provides a comparison of the government’s forecasts with those of independent or consensus forecasts. Greater comparison of the authorities’ macroeconomic assumptions and fiscal projections with those done outside of government could help improve the quality of fiscal forecasts.

2.4.2. Supplementary Budget (Basic)

60. The legal framework clearly mandates that substantial modifications of the budget or increases in the total level of expenditure needs to be approved by Congress. The legislation foresees three options to modify the enacted budget during the fiscal year:31

  • Supplementary budget appropriations (créditos orçamentários suplementares) can be enacted through presidential decrees during the fiscal year, as long as the expenditure increase is backed up by identified funding and does not exceed a percentage approved in the LOA (a supplementary budget proposal must be approved by Congress to exceed this limit). During fiscal year 2015, 21 decrees increased budgetary credits for a total of R$ 116 billion.

  • Special budget appropriations (créditos orçamentários especiais) are approved by Congress, for new expenditures which are not included in the enacted budget. They must also identify the funding source. During fiscal year 2015, Congress approved 28 laws, increasing budgetary credits by R$ 40 billion.

  • Extraordinary budget appropriations (créditos orçamentários extraordinários) cover unforeseen and urgent expenditures. These may be enacted by a provisional executive measure and should be approved by Congress ex post. During fiscal year 2015, six provisional executive measures made available additional budgetary credits for a total of R$ 116 billion.

61. Over the last several years, budget modifications have been approved after, rather than before, their implementation. The practice has deviated from the legal requirements in two key areas.32 Regarding the investment budget for federal public enterprises, in several cases public corporations have modified significantly their investment plans from the budget before authorization from Congress. This is usually resolved before the end of the year, but not always. Regarding the “fiscal” budget (central government), the key issue has been the need to change the fiscal target before implementing changes to budget allocations (as required by law). This has become a crucial issue in recent years, as the initial fiscal targets were based on projections that were too optimistic. In some cases, the authorities decided to execute modifications of the budget prior to approval of Congress of the change in the fiscal target. This happened as recently as 2015, when the requested modifications were approved by Congress before the end of the fiscal year. The TCU recently stressed that these practices were not according to the law, and as a result stricter adherence to the legal framework for supplementary appropriations is expected.

2.4.3. Forecast Reconciliation (Basic)

62. Macroeconomic and fiscal forecasts are revised from one budget stage to another, with little explanation for the reasons for the changes. Changes in one year ahead real GDP forecasts from one vintage to another have been especially significant in the past few years, due to the slowing growth of the Brazilian economy (Figure 2.3). Consequently, total revenue projections have also been significantly revised at every new official step. Nevertheless, forecast revisions from one LDO to the next or between a LDO and a LOA are not explained in budget documents. In particular, tables presenting the new forecasts do not contain a reference to the previous forecasts.

Figure 2.3.
Figure 2.3.

Brazil: Successive Official Forecasts of Real GDP Growth

(in percent)

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Sources: Ministry of Planning and IMF staff calculations. For year N, the graph covers official forecasts published from the PLDO released in year t-3 to the PLOA released in August of year t (the successive stages for the revision of the forecast for year 2014 are provided on the graph as an example).

63. In contrast, in-year forecast revisions introduced in the bimonthly execution reports of the Treasury are better documented. Article 9 of the FRL states that if after a period of two months, the government is off track to meet its fiscal targets, it should take measures to meet the LDO’s targets. Consequently, the FRL has made it compulsory to publish bimonthly reports which update the macroeconomic and fiscal forecasts, if deemed necessary by outturns in the first months of the year or by the macroeconomic environment. Tables compare the fiscal forecasts (primary revenue and expenditure, detailed by tax or spending type, and primary revenue) from the previous bimonthly report and the new ones. The reports also provide details of the updated underlying assumptions to each forecast. However, they do not provide a comparison with the assumptions that had been used in the previous report.

2.5. Conclusions and Recommendations

64. Brazil’s budgeting and forecasting practices are basic or good for most principles set by the FTC, but critical weaknesses remain. Areas of particular strength are the unity of the budget and public participation in the budget process (Table 2.5). A key weakness of the system is that the policy intentions for fiscal policy are not discussed in budget documents, nor is the outlook for key fiscal variables such as the budget deficit, primary balance, and public debt ratios. In addition, there is no scrutiny of the government’s fiscal projections by an independent fiscal institution. In this context, key reform priorities are as follows:

  • Recommendation 2.1. Publish a full-fledged medium-term fiscal policy statement. A medium-term fiscal policy statement, which could be attached to the next PLDO as an annex, would bring more clarity to the government’s fiscal intentions for the budget year and the medium term, reduce the uncertainty surrounding fiscal policy decisions, and provide a better sense of available fiscal space for new policies. It would provide, on top of the fiscal objectives and targets proposed by the government:

    • A concise (no more than a few pages) overall view of the government’s economic and fiscal strategy for the medium term;

    • Detailed information on the medium-term macroeconomic scenario, including an analysis of the international macroeconomic environment, an explanation of the assumptions underlying the GDP forecast (which could be broken down according to supply or demand), specific analyses on employment, inflation, the wage bill, and at least a few alternative scenarios to assess the level of uncertainty surrounding the forecasts;

    • A qualitative discussion of the government’s fiscal objectives, as well as a quantitative distinction between the baseline scenario and the impact of new measures;

    • An account of the revision of forecasts and objectives compared with the previous LDO, with a quantitative breakdown of the key revision factors.

Elaboration of a medium-term fiscal policy statement would also help prepare the government and the public for an eventual shift to an MTEF. Preferably indicative and limited in scope at first, an MTEF would require, among other things, a clear distinction between the baseline33 and the impact of new measures, and include an evaluation of the multiyear commitments the budget is facing due to ongoing investment projects.

  • Recommendation 2.2. Create an independent fiscal council. A fiscal council is a permanent independent entity with a statutory or executive mandate to assess government’s fiscal policies and performance, and promote sustainable public finances.

    • a. Such a council could at least express a judgment on the realism of macroeconomic and fiscal forecasts, provide ex ante and ex post evaluation of fiscal policy, and assess compliance with the fiscal rules. It could also assess how well different states are complying with their adjustment programs and requirements under the FRL In addition, the council could provide an assessment on long-term fiscal sustainability. For the fiscal council to be effective, it needs a high degree of independence and be appropriately staffed.

    • b. The government could at least compare and disseminate its forecasts with the ones published by independent analysts, and provide ex post analyses of differences between forecasts and outturns.

  • Recommendation 2.3. Initiate a full review of the existing legal framework and prepare a modern organic budget law. A consolidation of the legal framework into a single law would increase the overall transparency of the budget process. A modern organic budget law in line with best international practices, which was envisaged in the Constitution, is long overdue. It would also facilitate broader reform of the public financial management system. Many emerging market economies have focused in recent years on strengthening fiscal rules, medium-term fiscal frameworks, and management of fiscal risks. Most advanced economies have adopted these reforms after revising their basic budgeting framework, including a modern organic budget law.

Table 2.5.

Brazil: Summary Assessment of Fiscal Forecasting and Budgeting

article image

III. Fiscal Risks

65. This section assesses the government’s analysis, reporting, and management of fiscal risks against the practices of the Fiscal Transparency Code. It looks at three dimensions:

  • General arrangements for the disclosure and analysis of fiscal risks;

  • The reporting and management of risks arising from specific sources, such as government guarantees, public-private partnerships, and the financial sector; and

  • Coordination of fiscal decision making between central government, subnational governments, and public corporations.

Table 3.1 lists key government reports that provide information on fiscal risks.

Table 3.1.

Brazil: Selected Government Reports Relevant to Fiscal Risks

article image

General social security system (RGPS) (Projeções Atuariais para o Regime Geral de Previdência Social), civil servants’ pension scheme (RPPS) (Relatório da Avaliação Atuarial do Regime Próprio de Previdência Social), and of the military pensions (Avaliação Atuarial do Regime Próprio de Previdência Social dos Militares da União).

3.1. Disclosure and Analysis

3.1.1. Macroeconomic Risks (Basic)

66. The Brazilian economy is exposed to significant macroeconomic uncertainty, which, in turn, creates high uncertainty on fiscal outturns. It is crucial that budget documents discuss how macroeconomic developments could influence the public finances. This is critical at the current juncture where macroeconomic uncertainty has increased in light of less favorable and more volatile external conditions, a sharp decline in economic activity, and political turbulence. With the timing of the recovery uncertain, government revenues and the budget balance are likely to remain volatile, creating new forecasting challenges for the government. The forecasting is also complicated by the large number of revenue earmarking and mandated spending and indexation of key variables (like minimum pensions).

67. The statement of fiscal risks published every year provides some analysis of the sensitivity of revenue, expenditure, and debt to key macroeconomic indicators. Annex V of the PLDO on fiscal risks devotes a few paragraphs to macroeconomic risks. It provides a series of sensitivity analyses (impact of a 1 percentage point change in GDP, inflation, the exchange rate, policy interest rate, and the wage bill) on total tax revenue and total social security revenue; and the impact of a 1 percent change in the exchange rate, inflation rate, and policy interest rate on expenditure and debt. Moreover, a new feature of the PLDO 2017 report (published in April 2016) is the presentation of an alternative macroeconomic scenario (or stress test) for the years 2016 and 2017, with its consequences for the projections of selected revenues and expenditures. The sensitivity analysis does not, however, cover all key variables of interest in assessing fiscal policy—the budget balance and gross financing needs, notably. An example of this type of sensitivity analysis is provided in Figure 3.1, which is drawn from the IMF’s debt sustainability framework for market access economies and the 2016 Article IV Staff Report for Brazil.

Figure 3.1.
Figure 3.1.

Brazil: Example of Sensitivity and Debt-Sustainability Analysis

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Source: IMF (2016), Article IV Staff Report for Brazil.

3.1.2. Specific Fiscal Risks (Basic)

68. Many specific fiscal risks are disclosed, though the disclosure is not comprehensive and the relationship of the risks to the fiscal forecasts is sometimes unclear. The annual report on fiscal risks mentioned above has been published since the early 2000s—making Brazil one of the pioneers in transparency about fiscal risks. The most recent edition of the statement discusses, in addition to macroeconomic risks, specific fiscal risks relating to the government’s financial assets, public banks, public debt, and lawsuits brought by and against the government. Risks that are not discussed include those arising from subnational governments, public corporations, government guarantees, private banks, natural disasters, and concessions and public-private partnerships. Another important source of fiscal risks not discussed are linked to the volatility of commodity prices, although these are more relevant to some subnational governments. When risks are discussed, it is sometimes unclear how they might affect the achievement of the fiscal target and the forecasts for the following years. In addition, the statement does not address the fact that many risks are positively correlated—for example, periods of low economic growth are likely to be associated with the realization of fiscal risks from the financial system and state-owned enterprises.34

3.1.3. Long-Term Sustainability of Public Finances (Not Met)

69. There is no information on the long term, or even medium term, sustainability of key fiscal aggregates such as gross debt in budget documents. Projections of debt levels in the LDO only cover the budget year and, only indicatively, two forward years. There is no analysis on the possible impact of medium- and longer-term economic trends, such as demographic developments, on the fiscal accounts. There is no assessment of the available fiscal space in the coming years, nor a mention of legal constraints—earmarked revenues, mandated spending—that tend to rigidify fiscal balances and complicate medium- to long-term projections. Such analyses could give policymakers a better sense of the fiscal challenges that lie ahead.

70. Long-term projections for social security and pensions are available and provide a valuable building block for developing long-term projections of fiscal aggregates. These reports, which are required by the FRL, are produced by the Ministry of Social Security (and by the Ministry of Defense, in the case of military pensions). The report on the general social security scheme (RGPS) projects its annual balance over the next 45 years, relying on a set of detailed assumptions on demographics, the labor market, and labor productivity growth. The civil servants’ pension scheme is projected over the long term (75 years) in another report, as are the military pensions in separate reports. These reports indicate sizeable fiscal problems—for example, under present policies, spending related to the RGPS is projected to increase by almost 9 percentage points of GDP between 2014 and 2060. International comparisons also show Brazil has a much larger challenge than other countries (Table 3.2). Though these reports could be made more useful by including sensitivity analyses or alternative scenarios, they provide key inputs for long-term projections. Nonetheless, it should be noted that there remain important gaps in the coverage of long-term spending pressures, including pension funds at the subnational level and for public enterprises, and public healthcare spending.

Table 3.2.

Long-Term Sustainability of Pension Spending

article image
Source: Fiscal Monitor database, IMF.

3.2. Risk Management

3.2.1. Budgetary Contingencies (Good)

71. The budget allocation for contingencies can be used at the government’s discretion to meet unforeseen expenditures. Article 5 of the FRL provides that the LOA must contain a contingency reserve. While its use and amount must be defined in the LDO, the FRL provides that it aims at meeting contingent liabilities and other unforeseen events. The LDO for 2016 provided that the budgetary contingency would at least represent, respectively, 2 and 1 percent of the net revenue in the PLOA and the LOA. The difference of 1 percentage point aims to provide resources to fund amendments during the examination in Congress of the LOA. The contingency reserve has not been used during the fiscal years 2013 to 2015.35

72. In addition, at the beginning of each fiscal year, a presidential decree sets the ceilings for ministries’ expenditure. These are always below the amounts voted by Congress to contain public expenditure. Subsequent presidential decrees usually increase expenditures, resource permitting. Moreover, every two months, the Federal Budget Secretariat and the National Treasury Secretariat issue limits for commitments and cash disbursements. The requirement to meet the primary surplus target set by the LDO exacerbates this very short-term management of the budget. This implies an active control of the budget execution and leaves little room for a strategic management of the budget and its medium-term consequences.

3.2.2. Management of Assets and Liabilities (Good)

73. The government has large liabilities and also valuable financial assets. At the end of March 2016, the gross debt of the general government was 73.6 percent of GDP.36 At the same time, the financial assets of general government were worth 34.1 percent of GDP, the largest of the assets being deposits at the Central Bank (16.5 percent of GDP) and loans to BNDES (8.5 percent of GDP). In recent years, both the assets and liabilities have grown (Figure 3.2).37 The value of the government’s assets and liabilities, and the cash flows they give rise to, vary with inflation, interest rates, the exchange rate, and the performance of the companies the government owns or has lent to. Among the most salient risks now are the possibilities that the government will have to pay higher interest rates on its debt and that BNDES’s creditworthiness will decline, reducing the value of the government’s loans to the bank.

Figure 3.2.
Figure 3.2.

Brazil: Debt and Financial Assets of General Government, January 2007 to March 2016

(percent of GDP)

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Sources: Central Bank, Série Histórica da Dívida Líquida e Bruta do Governo Geral.Notes: The negative of debt and net debt are shown. Debt includes all government debt held by the Central Bank.

74. Government borrowing is authorized by law, and risks related to the government’s debt and financial assets are disclosed. The statement of fiscal risks discusses risks related to the debt portfolio and the foreign reserves that are counted in the government’s measure of net debt.38 The credit risk of another portfolio of financial assets is reported in a separate section of the same document. Although these disclosures are useful, some of the details more properly belong in the notes to the government’s financial statements or reports on debt management.

3.2.3. Guarantees (Good)

75. At the end of 2014, the total stock of recorded government guarantees was R$ 224 billion (3.9 percent of 2014 GDP).39 Much of this amount relates to borrowing by states, municipalities, and state-owned enterprises that is already included in headline measures of the debt of the public sector (as defined by the government); it does not represent an additional obligation. Guarantees given to parties whose debts are not included in the headline measures include export-credit guarantees and guarantees given to Eletrobras, Petrobras, and public banks.

76. Guarantees are disclosed, and their issuance is controlled by law. New guarantees, their beneficiaries, the stock of outstanding guarantees, and guarantee calls are disclosed in the annual Accounts of the President (Sections 1.9.2–5 in the 2014 accounts). Information on guarantees is also available in the quarterly Reports on Fiscal Management. Guarantees granted above the limits set by the Senate are null and void (FRL Art. 40(5)); for the federal government, the limit is 60 percent of net current revenues.40

3.2.4. Public-Private Partnerships, Including Concessions (Not Met)

77. Public-private partnerships are widely used, especially concessions. In this report, the term public-private partnership is used to include concessions, in which users pay for the service, as well as projects in which the government pays; in Brazil, the term is often used more narrowly to refer only to the second kind of project. The World Bank’s Private Participation in Infrastructure database includes 494 projects in Brazil’s transport and water and sewerage sectors in the 20 years beginning in 1995.41 The total estimated investment in these projects is $133 billion. The largest projects include the Rio de Janeiro, Guarulhos, and Brasilia airports; the Carajás railway; the Sao Paulo and Rio de Janeiro metros; and several toll roads. Brazil’s concessions transfer most ordinary risks to the concessionaire, though the government bears risks related to changes in government policy, force majeure, and extraordinary economic changes.42 The government also bears risk indirectly through its ownership of BNDES, which provides much of the projects’ financing. Projects in which governments pay for the service typically involve fairly predictable payments, but the debt-like obligation to make these payments reduces the governments’ flexibility to deal with adverse economic shocks.

78. Public-private-partnership contracts are published and government payments are restricted by law, but no report summarizes the government’s total obligations. In some respects, there is considerable transparency about public-private partnerships: in contrast to the case of many countries, the contracts themselves are generally available on the websites of the contracting agencies.43 Yet, summary information on governments’ rights and obligations in all public-private partnerships is unavailable, and the projects are not recorded on the contracting governments’ balance sheets. The off-balance-sheet assets and liabilities are estimated to be about 5 percent of GDP.44 Total government payments in public-private partnerships are limited by law to 1 percent of government revenue in the case of the federal government, and 5 percent in the case of subnational governments.45

3.2.5. Financial Sector (Not Met)

79. The risks created by banks are, as in most countries, among the most important. In March 2016, the aggregate liabilities of all financial institutions supervised by the Central Bank were R$ 7,673 billion (130 percent of 2015 GDP) (Table 3.3).46 This amount includes the abovementioned 8.5 percent of GDP that BNDES owes the government. Private banks’ liabilities are not guaranteed by the government, but the liabilities may benefit from some degree of implicit support.47 Up to certain limits, bank deposits are guaranteed by the Credit Guarantee Fund (Fundo Garantidor de Crédito), which is not government-guaranteed, but is too small to deal with the failure of a large bank or a systemic banking crisis. In addition, the treatment of banks’ deferred-tax assets creates an explicit contingent liability for the government: if a bank is unprofitable, is liquidated, or becomes bankrupt, its deferred-tax assets are converted into tax credits, and, if the value of these credits exceeds the taxes owed by the bank, the government pays the bank the difference.48 On December 31, 2015, the six largest banks had deferred-tax assets of R$ 233 billion (3.8 percent of 2015 GDP).

Table 3.3.

Brazil: Liabilities of Financial Institutions, March 31, 2016

article image
Sources: Central Bank, Selected Information on Supervised Institutions; Credit Guarantee Fund (Fundo Garantidor de Créditos), Annual Report, 2015, p. 13 (guaranteed deposits); IMF WEO database, April 2016, for GDP.Notes: The figures for deposits are for December 31, 2015.

80. The Central Bank thoroughly analyzes and monitors the banking sector. Bank supervision is well regarded, and banks’ capital exceeds regulatory minimums. The IMF Staff Report for the Article IV Consultation completed in March 2015 noted that “banking system soundness indicators” were “encouraging,” and the Central Bank recently concluded that “the solvency of the banking system remained at a high level in the second half of 2015.”49 Banks’ capital ratios exceed regulatory requirements for both private and public banks.50 Nevertheless, regulatory minimums do not eliminate fiscal risks, and the continuing recession is creating increasingly severe problems for many bank borrowers.51

81. Fiscal risks from public banks need to be closely monitored given their large size and exposure of the federal government. They have aggregate liabilities of R$ 3,641 billion (61.7 percent of 2015 GDP), 95 percent of which is owed by Banco do Brasil (BB), Caixa Econômica Federal (Caixa), and BNDES, respectively the first, second, and fourth largest banks in Brazil by liabilities.52 As noted above, in the case of BNDES, much of the liabilities are owed to the government, which has made substantial investments in the bank in recent years to allow it to expand. The federal government explicitly stands behind the liabilities of both BNDES and Caixa,53 which unlike BB are wholly government owned. The government has also established a bad bank, Empresa Gestora de Ativos (EMGEA), which bought bad loans from Caixa with a face value of about RS 7 billion from it as recently as 2014.54

82. Using the public banks to carry out government policy has generated fiscal risks. They have all reported profits in recent years, and some of their lending is less risky than that of the private banks. Their nonperforming loans are low, albeit increasing.55 But they also carry out public policy and make some high-risk loans, for example to municipalities and the rural sector. Because the interest rates they charge do not reflect these risks, some form of federal government support is probably necessary for them to achieve their policy objectives.56 The public banks are also more leveraged than their private counterparts (Figure 3.3), in part because the government has asked them to expand their operations while also taking dividends from them to increase the reported primary surplus.57 Overall, the combination of large liabilities, low equity, and rapid public-policy-driven expansion in a weak economy suggests important fiscal risks.

Figure 3.3.
Figure 3.3.

Brazil: Leverage of 20 Largest Banks by Liabilities, March 2016

(in billion reais)

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Source: Central Bank database, “Dados Selecionados de Entidades Supervisionadas –,” available at (accessed June 4, 2016).Note: public banks are shown in blue.

83. The central government does not disclose its contingent liabilities related to BNDES, Caixa, and deferred-tax assets. The Central Bank’s comprehensive, twice-yearly reports on financial stability present numerous indicators of the soundness of the financial sector, including the results of stress tests that estimate the effect of various shocks on banks’ finances.58 The Central Bank’s website includes comprehensive, easily accessed data on banks’ finances, which is also very timely. The most recent statement of fiscal risks also briefly discusses the risk that the government will need to recapitalize public banks. Management of the fiscal risks created by banks, including mechanisms for the resolution of any crisis, may warrant closer coordinated attention by the Central Bank and the Treasury—recognizing that the Central Bank has the expertise to monitor the risks but the Treasury must manage the fiscal consequences of possible problems.

3.2.6. Natural Resources (Not Met)

84. Natural resources have recently become more important for the public finances. In 2005, the World Bank estimated that Brazil had about $15,000 in natural capital per capita (the 26th highest out of 152 countries), most of it in pasture, crops, and forests.59 Since then, the volume of known oil and gas reserves has increased by 42 percent.60 In 2014, it was reported to be 15 billion barrels, or 0.9 percent of the world’s total (by comparison, Brazil’s share of the world population is 2.8 percent). The value of these resources has of course declined with the recent fall in the price of oil. The federal government has a direct stake in the government’s natural resources through its ownership of Petrobras and its receipt of royalties, but the fiscal importance of natural resources arises mainly indirectly, through general tax revenues. For some state governments, including Rio de Janeiro, commodity-linked royalties and indirect revenues are a particularly important source of revenue.

85. The government publishes a great deal of information on natural resources, though not the value of reserves of oil or mineral reserves. In the oil and gas sector, for example, ANP (the National Agency of Petroleum, Natural Gas, and Biofuels) publishes an annual statistical report that details the country’s oil and gas reserves and production, prices, and other information on the sector. It also publishes extensive information on royalties and other government revenues linked to the sector.61 However, it does not provide an assessment of the value of hydrocarbon reserves. Similarly, Brazil also provides information on other mining activities, including data on production and reserves for key products, as well as royalties. The information on prices is more limited.

3.2.7. Environmental Risks (Basic)

86. Droughts and other environmental problems frequently affect public finances. Most recently, the 2014–16 drought caused losses for Eletrobras and thus the government, as well more generally hurting the economy. In 2015, the collapse of the Samarco dam devastated some of the surrounding areas. Floods are also common. Diseases create additional risks, as the recent outbreak of the Zika virus has underscored. Earthquakes, tsunamis, and volcanic eruptions, by contrast, are rare. Although natural disasters can have a severe effect on particular areas, the size of the country means that their effect on public finances at the national level is mitigated. The 2015 edition of the World Risk Report ranked Brazil 123rd out of 171 countries in terms of exposure to risk to natural disasters (where a high ranking indicates high risk).

87. The government has published extensive information on natural disasters, though not much on their possible fiscal costs. The statement of fiscal risks does not mention environmental risks. Nor does any other government report routinely discuss their possible fiscal implications. The government has, however, published extensive information on natural disasters and industrial accidents, and their human and economic costs.62

3.3. Fiscal Coordination

3.3.1. Subnational Governments (Good)

88. Several states and municipalities are financially troubled. A few are no longer servicing debts to the federal government that were contracted after they were bailed out following an earlier crisis in the late 1990s. Some are not fully paying wages and salaries. Rio de Janeiro missed payments to the Agence Française de Développement and the Inter-American Development Bank in early 2016.63 The underlying causes of the states’ problems include falling revenue, especially in commodity-dependent states like Rio, and rising salary and pension costs.

89. Congress is debating a proposal to restructure states’ debt with the federal government. Subnational debt is about 13 percent of GDP. For states the amount was 11 percent of GDP at the end of 2015,64 and for municipalities it was 1.8 percent of GDP at the end of September 2015. Three-quarters of municipal net debt is owed by the city of Sao Paulo.65 Most subnational debt is owed to the federal government; debt to parties outside general government is 4.1 percent of GDP.66 States have asked for some debt relief from the federal government. The option being considered in Congress involves an extension in maturities and some relief in debt service over the next years.

90. Information on the finances of the larger subnationals is available, some of it monthly or quarterly, and subnational borrowing is limited by law. The FRL requires subnational governments to publish quarterly information on their finances (Articles 54–55) and some also publish monthly data. Summary information, including some aggregate or consolidated data, is available from the Central Bank and the National Treasury.67 States are required to keep the ratio of their net debt to their net current revenue below 2, and municipalities are required to keep the ratio below 1.2. Subnational spending on personnel is limited to 60 percent of net current revenue.68 The adjustment programs between the federal government and states impose additional, generally tighter, limits.

91. Yet there are serious problems with the management of subnational risks, including weaknesses in the quality of fiscal reporting and monitoring.

  • The debt limits in the adjustment programs have not always been enforced, partly because the parallel existence of the looser limits under the FRL has undermined them in practice. In addition, some states were allowed to increase their borrowing, with guarantees of the federal government, despite technical assessments that highlighted the risks.

  • Weaknesses in fiscal reporting also undermine the ability to assess the fiscal position and risks. Not all states and municipalities comply with their reporting obligations, and information on subnational finances is generally less timely and comprehensive than information on the central governments. The lack of a fiscal council, as envisaged in the FRL, makes it more difficult to ensure that governments implement appropriate reporting standards and that there is effective monitoring of fiscal statistics at all levels of government69

  • A key weakness is the reporting on the wage bill, the most important spending item. Personnel spending is measured in different ways in different states (Annex 3), allowing some to circumvent the limit on such spending.70 At the end of 2015, six states had in any case breached the limit (Rio Grande do Sul, Mato Grosso do Sul, Minas Gerais, Paraíba, Goiás, and Rio de Janeiro). Rio Grande do Sul had also breached the debt limit, and Rio de Janeiro and Minas Gerais were close to doing so. Moreover, even when these limits are observed, they have not prevented serious financial problems. Arrears have been reported in Amapá, Amazonas, and Roraima, which, like other states in the north, had very low reported debt (Figure 3.4).71 In some cases, borrowing rules have exacerbated liquidity problems by limiting borrowing even for government with low debts.

Figure 3.4.
Figure 3.4.

Brazil: Debt-to-Revenue Ratio by State, 2015

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Sources: National Treasury Data are for December 2015, except for Paraná, Rio Grande do Sul, and Sergipe (August 2015) and Mato Grosso do Sul (April 2015).

3.3.2. Public Corporations (Good)

92. Public corporations create significant risks. The three federal public banks discussed in Subsection 3.2.5 are the ones with the largest liabilities (Figure 3.5), some of which, as noted above, are owed to the central government. The smaller nonfinancial corporations are consolidated in the accounts of the public sector (as defined in Brazil), so the risks associated with their borrowing are monitored as part of the budget process. The Post Office has nevertheless experienced financial problems, partly because of its pension scheme.72 The biggest risks associated with nonfinancial corporations, however, relate to Eletrobras and Petrobras, which are not consolidated in the fiscal accounts.73 They have gross liabilities of 1.9 percent and 11.4 percent of GDP, respectively. Petrobras—where net debt almost quadrupled since 2011—is financially stressed, in part because of corruption, but mainly because of its quasi-fiscal activities (principally, selling petrol at less than the market price)74 and heavy borrowing during the oil price boom to pay for the development of the pre-salt oil fields (Box 2). Eletrobras is also financially stressed, partly because of its obligations to provide subsidized public services without compensation (that is, quasi-fiscal activities), though it also suffered from the above-mentioned drought and possibly also corruption. Both companies are highly leveraged, and both are viewed by credit-rating agencies as benefitting from implicit government guarantees.75 Another area of concern is the high degree of exposure of public banks to these companies.76

Figure 3.5.
Figure 3.5.

Brazil: Liabilities of 20 Largest Public Corporations Outside the Fiscal Accounts, March 2016

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Sources: For Petrobras and Eletrobras, the company’s financial reports for the first quarter of 2016; for the financial corporations, the Central Bank’s database Dados Selecionadas de Entidades Supervisionadas.

93. Transfers between the government and public corporations are disclosed, and the government publishes an annual report on the corporations’ finances. The annual report and accompanying website produced by the Ministry of Planning and Budget, include information not only on the companies the ministry supervises but also on those supervised by the Treasury. The report contains a discussion of the sector, data on the finances of each of the corporations, and aggregate data on the finances of all nonfinancial public corporations and all financial public corporations. The report is not very timely, but the corporations themselves publish their own financial statements in generally a timely manner. The government has published a series of directives setting out aspects of its ownership policy.77

94. Yet there is no disclosure of the quasi-fiscal activities of public corporations. Given the size of these activities and their complexity in the case of public banks, this is a significant shortcoming. Moreover, although the financial reports of the major companies are generally detailed, informative, and timely, investigations of corruption at Eletrobras have delayed the publication of its financial statements. The audit opinion on BNDES’ most recent accounts was qualified because BNDES did not recognize in its income statement its full loss on its shares in Petrobras. Furthermore, Petrobras’ report states that the company’s problems with internal financial controls are not yet solved, which also caused the delay of its previous annual report.

3.4. Conclusions and Recommendations

95. Table 3.3 summarizes the assessment of Brazil’s practices in the area of fiscal risks and reveals a number of strengths. It shows that Brazil meets the standard of good or basic practice on 7 of the 12 principles. In particular, good information is available on government guarantees, the governments’ financial assets and liabilities, and the finances of subnational governments and public corporations. Statistics on subnational government and the coverage of the statement of fiscal risks have both improved in the past year, and Brazil’s performance against the code could be improved fairly easily by further increasing the scope of the statement. In the areas of environmental risks, natural resources, and public-private partnerships (including concessions), Brazil could meet at least the standard of basic practice by adding to the statement brief discussions of the relevant risks.

96. Nevertheless, the disclosure and management of fiscal risks are the weakest area of fiscal transparency in Brazil. The statement of fiscal risks should focus on large risks and their possible effects on the deficit, the debt, and other fiscal aggregates. To achieve this objective, it needs to be transformed. At present, it is a collection of largely unrelated and sometimes excessively detailed discussions of various fiscal issues, some of them relatively minor, which reflects the absence of centralized monitoring and analysis of the risks that the government faces. The statement should become a report that explains clearly and concisely how and why fiscal outcomes could differ from fiscal forecasts. Producing a shorter, simpler, and better integrated statement would require more analysis and more synthesis of existing information. The necessary work could be led by a risk group that would also assess whether the government was managing fiscal risks adequately.

97. Dissemination of long-term fiscal projections in a separate report is a high priority. Building on the existing projections for pensions and social security, this report could project the government’s primary balance, nominal balance, and debt over a period of at least 10 years—and ideally much longer. It would assume the continuation of current tax and spending policies, and its purpose would not be to predict the future, but to give the best possible estimate of whether the government’s policies were sustainable in the long term. This would help the public better assess the sustainability of Brazil’s fiscal position, and it would help policymakers prepare for indispensable reforms. Brazil’s high level of public debt and large expected increases in age-related spending make it urgent, from a transparency perspective, to undertake such projections and to disseminate them. To complement these projections, the government should also take advantage of its fiscal policy statement to indicate the steps it intends to take in the medium term to slow and eventually reverse the growth of public debt (Section 2).

98. The fiscal significance of federal public banks deserves additional disclosures. The relationships between the banks, general government, and the rest of the public sector are large enough to warrant not only additional discussion in the statement of fiscal risks, but also a new chapter in the annual report on public corporations (or a separate publication). This chapter or publication would detail the financial relationships between the government and each of these banks and between the banks and the rest of the public sector. The government should also disclose and quantify the quasi-fiscal activities of the banks, Eletrobras, and Petrobras.

99. Based on the above assessment, the evaluation highlights the following priorities for improving the transparency of fiscal risk disclosure and management:

  • Recommendation 3.1. Create a risk group to provide a centralized overview of fiscal risk management (identify, monitor, and mitigate).

  • Recommendation 3.2. Improve the disclosure of fiscal risks in budget documents:

    • a. Present in the statement of fiscal risks a comparison of past forecasts with fiscal outcomes (April 2017) and then develop fan charts based on stochastic analysis (2018 or 2019)—in both cases to give an indication of the extent of fiscal uncertainty;

    • b. Ensure that sensitivity analyses show the effect of the changes in the underlying assumptions on total government spending and revenue, the primary and nominal balances, and government debt (April 2017) and, in the longer term, present more fully worked out alternative scenarios (April 2018);

    • c. Add disclosures of the government’s explicit support of the financial sector and of the risks related to public corporations and subnational governments (April 2017);

    • d. Ensure that all sections of the statement make clear how the risks under discussion relate to the budget and medium-term fiscal forecast (April 2017); and

    • e. Include in the annual report on public corporations produced by the Ministry of Planning and Budget, in collaboration with the National Treasury, a chapter on the financial relationships of the government with each of the federally-owned banks and on the quasi-fiscal activities carried out by the banks and other public corporations (2017).

  • Recommendation 3.3. Improve reporting on the long-term sustainability of public finances:

    • a. Publish a periodic report on the long-term sustainability of public finances that includes projections of the government’s primary and nominal balances and debt, under current government policy, over the next 20 years at least;

    • b. Expand long-term sustainability analyses of the social security and public pension schemes provided in the PLDO with further sensitivity analyses and alternative scenarios; expanding the methodology to include pension funds not yet covered, such as pension funds of public enterprises; and including projections of expected increases in health care expenditure; and

    • c. Complement the fiscal policy statement (Section II) with a section on the policy changes envisaged in the medium term by the government to improve Brazil’s long-term fiscal sustainability.

Table 3.4

Brazil. Summary Evaluation: Fiscal Risks

article image

Annex 1. Fiscal Transparency Action Plan

article image
article image
article image
article image
article image

Annex 2. Transactions Between the Treasury and Central Bank

The institutional relation between the Central Bank of Brazil (BCB) and the National Treasury is characterized by a number of distinctive features. Over the past decade, this particular setup interacted with external and domestic economic developments (e.g., exchange rate variations, fiscal deficits), resulting in sizable impacts on the balance sheets and fiscal flows of the two institutions. These include: (i) more than 20 percent of public debt now being issued for monetary policy purposes; (ii) the high balances of the government deposits at the Central Bank, which partly reflect large unrealized profits of the BCB; and (iii) the interest bill of the central government, which reflects costs of the BCB interventions in the foreign exchange markets.

This complex relationship (Figure A2.1) and, in some cases, the accounting and statistical treatment, could be strengthened to improve transparency. This annex describes in more detail the operations and how they are recorded. While there is an ongoing debate on reforming the relationship between the Treasury and the BCB, policy issues are beyond the scope of this report.

Annex 2. Figure A2.1.
Annex 2. Figure A2.1.

How Fiscal, Monetary and Exchange Rate Policies Affect Public Debt

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

* Scenario where FX swaps corresponds to 1/3 of international reserves** Considers onlyliquidity effects

The Relationship between the Treasury and Central Bank

Monetary and exchange rate policy and fiscal policy are the key drivers of the level and dynamics of public debt:

  • The FRL prevents the BCB from issuing own debt securities, and the BCB has been using securities issued directly by the National Treasury for liquidity management. At end 2015, about 20 percent of public debt was used as collateral for repurchase agreements, or repos (compromissadas). The key drivers of the fast increase have been the sizeable reserve accumulation since 2006 and the high interest rates (BCB has to issue repos to absorb liquidity when it pays interests on the repos).

  • The public sector debt statistics of the BCB consider the Central Bank to be part of the central government, a feature dating back to 1991—reflecting the legal framework in Brazil. Public debt securities held by the Central Bank under outright ownership are not considered as central government (or general government, GG) debt since 2008; only those that are used as collaterals for repos (compromissadas) are included in debt statistics. This contrasts with the treatment in the GFSM, where the Central Bank is considered a public financial institution, and, as such, its entire Treasury securities holdings are considered a liability of the GG.


Brazil: GIR and Repo operations

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Sources: Central Bank of Brazil.

The gains and losses of the BCB have an asymmetric treatment. Losses are covered through bond issuances by the Treasury, while profits are credited into the TSA. The law also provides for the possibility to retain up to 25 percent of profits, which has been done occasionally (e.g., in 2013). The profits received by the Treasury are earmarked and can only be used to pay debt, preferably the debt held by the BCB. In addition, the results of the BCB are separated into those originating from foreign exchange operations and those resulting from other operations (Law 11.803, 2008), and the two are not consolidated.78 This implies that in the same period, the Treasury may need to issue bonds to cover losses under one criteria and, at the same time, have profits credited in the Treasury Single Account (TSA).


Central Bank of Brazil: Profit/Losses

(R$ billion and percent of GDP)

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Sources: Central Bank of Brazil.

With the large foreign exchange (FX) reserves accumulation, the BCB’s results have become more volatile over time. Since 2010, profits averaged about 1½ percent of GDP per year, a great part of it composed of volatile unrealized gains from the revaluation of FX assets. The large increase in government deposits at the Central Bank stems mainly from these accumulated (but unrealized) profits of the BCB. Moreover, government deposits, regardless of their origin, accrue interest remuneration (different from the remuneration of FX reserves) to the government, which is not earmarked.


Balances of the Single Treasury Account

(RS billion and percent of GDP)

Citation: IMF Staff Country Reports 2017, 104; 10.5089/9781475597448.002.A001

Sources: Central Bank of Brazil.

The treatment of the transactions between the BCB and the Treasury also makes it difficult to assess the fiscal position and debt dynamics of the central government (CG). It could also hamper the BCB’s capital position when unrealized profits are distributed.

  • As discussed in more detailed below, many of the effects of the operations of the BCB are immediately reflected in the interest bill of the central government. This accounting of the BCB operations in Brazil makes it difficult to assess the individual fiscal situation of two very different units of the public sector, the CG and the Central Bank. For example, the BCB intervenes in foreign exchange markets, including through FX derivatives (commonly known as “FX swaps”). The cost (or gains) from the FX swaps are reflected monthly as net interest expenditure in the interest bill of the central government This complicates the assessment of debt dynamics and the cost of public debt79

  • The distribution of unrealized BCB profits also hinders the assessment of net debt dynamics of the central government For example, net debt of the CG may fall sharply as cash deposits increase, thanks to the distribution of unrealized gains of the BCB. These operations thus result in fluctuations in the assets and liabilities of the CG that are not linked to changes in CG fiscal balances. This treatment of unrealized profits could also come at the expense of negatively affecting the capital of the BCB.80,81 Furthermore, the asymmetry between the instruments to compensate for losses (issuance of debt securities) and the absorption of gains (deposits in the TSA) have resulted in an increasing stock of gross government debt.

  • The authorities are considering changes to the relationship between the Treasury and the Central Bank.

Transparency in the Fiscal Statistics

While the relationship between the Treasury and BCB is reflected in the different public reports, the treatment varies and tends to differ from international practices:

  • The fiscal statistics in the Treasury reports (RTN) only present limited information on the impact of the transactions (e.g., interests paid by BCB on government deposits and transfers related to gains/losses of the FX derivatives). However, a memorandum table in the report presents more details on the accrual and transfers of BCB profits.

  • The fiscal statistics reported by the BCB reflect the fact that the BCB is considered part of the CG. Most BCB operations are immediately reflected in the central government interest bill and not as non-interest transactions, according to international statistical standards (Table Annex 2.1). In addition, the transfer of unrealized gains/losses from exchange rate valuation associated with international reserves is treated as other economic flows.

  • The Balance Sheet of the Union (BGU) records all the flows between the Treasury and the BCB, as the monetary authority, as cash flows (in both the budgetary and cash flow statements) and as accrued changes in net worth (in the statement of changes in net assets/equity); however, the accounting classification structure does not permit disclosing the split between the revenue and financing components of those transactions.

  • Data submitted for the GFS yearbook database reflects some of the relationship between the two institutions (e.g., interests paid on the government deposits, and interests paid by the Treasury on securities held by the BCB); however, it does not reflect the stocks or flows related to profits or losses of the Central Bank in accordance with international statistics standards. Under the GFSM 2014, the profits due to current operations of central banks transferred to government units should be recorded as dividends, and the compensation of losses due to current operations of central banks should be recorded as subsidies to public corporations. However, a different treatment applies to the transfer of unrealized gains/losses due to valuation effects on the FX reserves. The GFSM 2014 prescribes that one-off payments based on holding gains should be recorded as withdrawals of equity (an analogy can be made for compensations for holding losses, which should be recorded as equity injections).

Annex 2. Table A2.1.

Treatments of BCB Results in Fiscal Statistics

article image
Source: Central Bank of Brazil (Notas de Imprensa de Política Fiscal).

The complex relationship between the two institutions requires greater transparency to better inform policymakers and the public in general. To a large extent the differences with international practices arise from the fact that in Brazil the Central Bank is considered part of the central government and not as a separate unit of the public sector. The present reporting makes it difficult to identify the benefits/costs and risks associated with policies. It also complicates international comparisons on key fiscal aggregates, like public debt and the primary/overall fiscal balances.

Annex 3. Differences Across States in Complying with the Fiscal Responsibility Law

The FRL determines different ceilings for the spending on personnel for the three branches of power (executive, legislative and judicial) and all levels of government (federal, states and municipalities). Total personnel expenditure is defined as the sum of expenditures incurred for both active and inactive workers, and thus includes expenditures for pensions, in-kind benefits, and social security contributions.

Annex 3. Table A3.1.

Limits for the Spending on Personnel

(in percent of net current revenue)

article image
Source: Government of Brazil.Note: Legislative includes the Court of Accounts. Other includes the Prosecutor Office for the federal government and states. In the case of the federal government it also includes spending with the DF and ex-territories (Amapá and Roraima).

In the case that any of these limits are reached, subnational governments (SNGs) have a period of eight months to bring spending back within the limit Otherwise, they lose all of their voluntary federal transfers and permission to contract credit from either the private or public sector. Two exceptions to this rule are credit operations to refinance existing securities debt and credit operations that assist in reducing personnel expenditures.

In practice, it has been difficult to evaluate whether states are complying with the FRL’s limits. The reason is that the interpretation of the law by the states’ courts of accounts has differed. Decisions of the court of accounts have, in effect, allowed SNGs to circumvent the limit on personnel spending. The main court decisions have addressed the following issues which affect compliance with the personnel spending limit are as follows:

  • Removal of the income tax in workers’ wages as an expense/revenue. As an employer, the SNGs pay their employees a gross salary, of which a part is retained for the payment of the income tax. Yet, some states do not take this amount into account, either as an expense or as revenue.

  • Exclusion of social security and pensions as an expense. Some states do not compute the expenditure with social security and pensions as personal spending, under the argument that this type of expenditure cannot be restricted by the government. In the case of the state of Rio de Janeiro, the government directs the revenue from royalties to the state pension fund. As the fund covers the state’s payments of social security and pensions, the amount is not recorded as spending on personnel, yet the payments of the royalties are recorded as revenue. In some states the spending on social security and pensions of the legislative and judicial branches are classified as part of nonwage spending for the executive branch, distorting the reporting on the spending with personnel.

  • Exclusion of in-kind benefits to employees and outsourcing. Some states, in addition to the exclusion of some in-kind benefits (meals and housing, among others), do not record the spending with outsourced workers and contracts with individual service providers.

Table A3.2 below shows that different interpretations regarding the reporting on personnel spending create sizeable differences in assessing SNG compliance with the FRL The table indicates that the fiscal reports released by the states used to verify their compliance with the FRL, and the report released by the Treasury. Table A3.2 shows that in some states the difference in 2015 was more than 140 percent.

Annex 3. Table A3.2.

Total Spending on Personnel by the States

(in reais 2015)

article image

Annex 4: Technical Note on Table 0.2

Sources and methods

Methodological framework

As described in the footnote to Table 0.2, this table presents estimates compiled in accordance with the GFSM 2014, and adopts the accrual basis of recording for transactions and market valuation for stocks to the extent possible.

These estimates attempt to present a broad overview of Brazil’s public sector finances, by complementing data published in the various fiscal reports with IMF staff estimates for the most material data gaps, in reference to the GFSM 2014 framework. Given the limitations in data sources and time available for their compilation, this exercise results in just an approximate picture of the public sector finances.

In the GFSM presentation, transactions that increase net worth are recorded as revenue, while transactions that decrease net worth are recorded as expense. The net operating balance, which provides a measure of the sustainability of the policies of each group of units, corresponds to the difference between revenue and expense, and, as such, excludes transactions in nonfinancial assets.

The net investment in nonfinancial assets (acquisitions less disposals less consumption of fixed capital) does not change the net worth of the public sector, but affects the financial resources available to it, so it must be subtracted from the net operating balance when calculating the net lending/borrowing (also referred to as fiscal balance). This represents the amount that the government has available to lend or must borrow to finance its nonfinancial operations. Total expenditure can be derived as the sum of expense and net investment in nonfinancial assets.

Primary sources of information

Estimates for transactions of general government and its subsectors were based on the noncash data disclosed in the joint STN/IBGE publication “Government Finance Statistics and Government Sector Account, 2014.” For coverage completion, transactions and stock positions pertaining to the small “Sistema S” units and professional councils were added, based on publically available accounting information.

Estimates for stocks of assets and liabilities of general government and its subsectors were based on two sources of information:

  • STN’s data submission for the 2014 edition of the IMF’s Government Finance Statistics Yearbook, for data on assets in currency and deposits, debt securities and loans and on liabilities in debts securities and loans (these data are derived from BCB’s Press Release on Fiscal Policy);

  • Financial statements of federal government (BGU) and general government (BSPN), for all other assets (including nonfinancial) and liabilities; and

  • Financial statements of FGTS, “Sistema S” units and professional councils, for assets and liabilities of the extrabudgetary central government subsector.

Estimates for transactions and stocks of public corporations of the central level (excluding the Central Bank) were based on the corporations’ accounting information available in DEST’s database. Data for the Central Bank was sourced from its financial statements.

Estimates for transactions and stock positions of subnational public corporations were based on the financial statements of the largest corporations.

Coverage adjustments

The institutional coverage of the public sector and its subsectors in Table 0.2 differs from that in the main fiscal reports (STN’s Central Government Primary Balance and BCB’s Press Release on Fiscal Policy), in the following respects:

  • The administrative functions (“autarquia” component) of the Central Bank was excluded from the central government; in accordance with the institutional unit approach of 2008 SNA and GFSM 2014, the Central Bank as a whole was included in the public corporations’ subsector;

  • FGTS, “Fundo Remanescente do PIS/PASEP”, “Sistema S” units and Professional Councils were added to the extrabudgetary central government subsector, because these nonprofit institutions are considered to be controlled by the government, following the control criteria of 2008 SNA and GFSM 2014;

  • Three public corporations (Casa da Moeda do Brasil, DATAPREV, SERPRO) were reclassified into central government, because they provide most of their output to the government; two other corporations (CODERN, TELEBRAS) were also reclassified into central government due to their nonmarket orientation (over a sustained period, their operating revenue covered less than half of their operating costs); in addition, EMGEA was reclassified because it acts with strong public financial support and effectively on behalf of the government when handling the “bad” assets it acquires from public banks.

  • Petrobras and Eletrobras were added to the nonfinancial public corporations’ subsector, because they are considered to be controlled by the government following the control criteria of 2008 SNA and GFSM 2014;

  • Finally, public financial corporations, both central subnational, were added to complete the coverage of the public sector.

Sources and methods for staff estimates
1. Subsoil assets

The estimation of the value of the reserves of non-renewable natural resources was based on a set of the most relevant resources for which there was available information on volumes of reserves and price benchmarks. The value of the oil and gas reserves were computed by multiplying the volumes of reserves by the reference prices (both based on data provided in ANP reports) and an estimated government share of 60 percent (past studies for Brazil suggested a government share around this magnitude).

The same methodology was adopted for the most important metals for which there was public data (iron ore, copper, bauxite, gold, and niobium). The reserve data was from DNPM (Sumário Mineral 2015), the prices data were from WEO, Bloomberg, and USGS. As international experience shows that the government take for metals is lower for oil and gas, we assume it to be at 40 percent.

2. Pension entitlements and related transactions

The primary data source for the stock of civil servants’ pension entitlements was the “Prestacão de Contas da Presidenta da República.” This document includes the entitlements of civil servants in both the federal level (a provision in BGU’s balance sheet) and subnational level (a note in the section related to the outcomes of Program 2061 – Social Security).

The pension entitlements for military personnel of the armed forces were estimated by IMF staff using data from the actuarial projections published in the annexes to the PLDO.

The accrual of pension entitlements (transaction) of 2014 was derived according to the following formula:

  • (1) Accrual of pension entitlements in 2014


  • (2) Change in stocks of pension entitlements between 2013 and 2014


  • (3) Benefits paid in 2014


  • (4) Actual contributions received in 2014


  • (5) Actuarial gains and losses in 2014

The stocks of 2013 and 2014 (item 2) were derived from the data sources described above. The 2014 actual flows of benefits paid and contributions received (from employee and employer) in each level of government (items 3 and 4), were provided to the mission by STN. The actuarial gains and losses in 2014 were estimated by IMF staff, taking into consideration the changes in life expectancy tables published by IBGE. Due to unavailability of data sources, the mission assumed that there were no other actuarial gains and losses in 2014.

In summary, the treatment of “unfunded employment-related defined benefit pension schemes” prescribed by the GFSM 2014 (see the manual’s Appendix 2 for details) requires that:

  • Actual contributions received by the scheme, be treated as an increase in pension entitlement liabilities (a below-the-line transaction);

  • Actual benefits paid by the scheme, be treated as a decrease in pension entitlement liabilities (a below-the-line transaction);

  • The “service” costs (i.e., the increase in pension entitlements associated with the wages and salaries earned in the current period), be treated as compensation of employees; this is broken down into two sub-components:

    • Actual employer social contributions, paid by the employer unit to the scheme; and

    • Imputed social contributions (calculated as a residual, i.e., the total service costs net of actual contributions);

  • The “financing” costs (i.e., the increase in pension entitlements due to the fact that the benefits are one period closer to settlement), be treated as property expense.

The accrued pension entitlements calculated above (item 1) correspond to the sum of the “imputed social contributions” and “property expense” components, described in the previous paragraph.

As discussed in the data sources section of this annex, the starting point for transactions of general government and its subsectors was the STN/IBGE publication “Government Finance Statistics and Government Sector Account, 2014”. The GFS tables in this publication follow the format of the GFS Yearbook Questionnaire, which break down the flows of social contributions and social benefits between social security and employment-related (civil servant specific) pension schemes. Since STN and IBGE still have not adopted the 2008 SNA and GFSM 2014 recommended treatment of employment-related pension schemes, the current recording reflects mostly cash flows. The only exception is the imputation of social contributions (both in revenue and expense), corresponding to the difference between benefits paid and actual contributions received (this treatment was only applied to central government).

To convert this recording into a GFSM 2014-compliant recording, the mission performed the following adjustments:

  • Exclude actual contributions from revenue;

  • Exclude payments of social benefits from expense;

  • Add to expense the difference between the accrued pension entitlements calculated above (item 1) and the value currently recorded as social contributions (which includes three components: the employee’s contributions, a component of the recorded wages and salaries; the employer’s contributions; and the imputed contributions).

3. Government-controlled PPP assets and related liabilities and transactions

Estimates on the value of PPP assets/liabilities and related transactions were obtained from the World Bank’s Private Participation in Infrastructure database (as of May 31, 2016). This database includes data on total annual commitments and the value of physical assets. Due to unavailability of data, no estimates were done for consumption of fixed capital or revaluations related to these assets.

As discussed in Section 3.2.4, only projects described as concessions and greenfield projects were considered. They were deemed to be classified as government assets primarily because the government is bearing a substantial financial risk, as owner of BNDES, which provides much of the projects’ financing. Other supporting factors include: (i) the assets are supposed to revert to government ownership at the end of the contract; (ii) the government also bears risks related to changes in government policy, force majeure, and extraordinary economic changes.

According to the GFSM 2014, if the government bears the majority of the risks related to a PPP contract, it should record in its accounts the investment in assets. At the same time, a liability of the same size of the asset needs to be imputed, to account for the fact that the government has acquired an asset without immediately paying for it (i.e., there were no immediate cash implications).

For more detailed description on the statistical and accounting treatment of investment undertaken via PPP arrangements, consult Appendix 4 of GFSM 2014 and the appendix of the PPP Fiscal Risk Assessment Model (P-FRAM) user manual.

4. Relations between the Treasury and the Central Bank

The explanatory notes to the BGU and