Government Finance Statistics to Strengthen Fiscal Analysis

This paper seeks Board approval of a phased migration strategy for implementing the GFSM 2001 as the standard for Fund fiscal data. The financial crisis has underscored the need for better and more comparable fiscal data, including on government assets and liabilities. In 2005, the Board endorsed the Government Finance Statistics Manual 2001 (GFSM 2001) and asked staff to propose a migration strategy for the Fund, based on lessons to be drawn from conducting pilot studies. The pilot studies and other tests confirmed that some aspects of the GFSM 2001 methodology (primarily relating to presentation) can be implemented by staff with relatively modest resource needs, while other aspects can proceed meaningfully only after member countries have made sufficient progress in their own fiscal reporting—including in detailed balance sheet breakdowns, and in strengthening underlying accounting systems.

Abstract

This paper seeks Board approval of a phased migration strategy for implementing the GFSM 2001 as the standard for Fund fiscal data. The financial crisis has underscored the need for better and more comparable fiscal data, including on government assets and liabilities. In 2005, the Board endorsed the Government Finance Statistics Manual 2001 (GFSM 2001) and asked staff to propose a migration strategy for the Fund, based on lessons to be drawn from conducting pilot studies. The pilot studies and other tests confirmed that some aspects of the GFSM 2001 methodology (primarily relating to presentation) can be implemented by staff with relatively modest resource needs, while other aspects can proceed meaningfully only after member countries have made sufficient progress in their own fiscal reporting—including in detailed balance sheet breakdowns, and in strengthening underlying accounting systems.

I. Introduction

1. The Fund recognized the importance of a more rigorous and comprehensive framework for recording government finances over a decade ago. This led to the preparation of the Government Finance Statistics Manual 2001 (GFSM 2001). The GFSM 2001 was adopted by the international statistical community as an international standard in 2001; and in 2005, the Executive Board considered that “GFSM 2001 provides a comprehensive analytical framework that will strengthen fiscal policy analysis and reporting in Fund surveillance and program work, through its summary fiscal tables—the operations statement (on a cash and/or accrual basis) and the balance sheet—and the core indicators derived from these tables.” Directors also noted that use of the GFSM 2001 framework will lead to greater transparency and consistency in the presentation of country fiscal data in staff reports. They recognized the GFSM 2001 as an appropriate framework for handling new and complex fiscal operations that pose challenges to fiscal reporting and analysis, and noted its usefulness for assessing sustainability.1

2. The GFSM 2001 also addresses many of the fiscal data concerns arising from the financial crisis.2 While the crisis originated from the private sector, it is having major implications for fiscal accounts throughout the world—thus highlighting the importance of reliable and comprehensive government sector data.3

  • Efforts to track the fiscal stimulus, and in particular, to compare country contributions to it, put a spotlight on the importance of consistent, comparable, and comprehensive government finance statistics.

  • The fiscal response to the crisis confirmed the need for information on government balance sheets as a fiscal policy tool. Most government support for the financial system was provided through balance sheet operations (lending, asset purchases, equity injections) or operations that could have implications for the government balance sheet (guarantees and other contingent support, use of special purpose vehicles), rather than by direct government spending. This means that, in many cases, the traditional government operations table—which shows only flows of revenue and expenditure, and how these are financed—is an incomplete depiction of the fiscal response to the crisis, even when complemented by data on gross debt stocks. The adoption of GFSM 2001 will also support MCM’s sovereign risk assessment.4

3. However, in 2005, Directors also cautioned that national authorities will require time to fully implement the GFSM 2001. They recognized the likely need for a phased introduction. Directors considered that changes in presentation would be easiest to implement across the Fund’s membership, while improvements in the quality of data reporting would take more intensive technical assistance, and the strengthening of underlying fiscal accounting systems would take longest.

4. Directors asked staff to conduct pilot studies and to report back with a proposal for a migration path to fully implement the GFSM 2001 methodology. The pilot studies were considered necessary to test the system, including assessing the resource implications of the required reforms, to arrive at a determination of how best to adopt this format for the Fund’s published fiscal data. This paper describes the findings of the pilot studies and other progress in the worldwide adoption of the GFSM 2001, and proposes a migration strategy for Board endorsement.

II. What is the GFSM 2001 Integrated Framework?

5. The figure below is a stylized depiction of the GFSM 2001 framework. It incorporates some important innovations compared with the previous Fund standard, GFSM 1986.

  • Unlike GFSM 1986, which tracks the public finances only with a government operations (flow) table, GFSM 2001 nests the government operations table between opening and closing balance sheets. This means that the system tabulates how changes in the government’s net worth position from year to year are explained partly by government operations and partly by “other flows” (notably valuation changes, debt write-offs, etc.) The benefits of doing so are: (1) explicit tracking and hence a better understanding of the impact of valuation changes, debt operations, etc., on the health of the public finances; and (2) a potential check on the consistency of flow and stock data.

  • While the GFSM 2001 operations statement remains close to the government operations table used in most staff reports, the statement reflects several differences in order to align it as closely as possible with other statistical systems and private sector accounting standards.

    • GFSM 2001 envisages government operations being measured on an accrual (or resource-accounting) basis—in the same way as the national accounts, balance of payments, and the operations of private firms. A shift to accrual accounting would systematize many of the ad hoc amendments made by Fund economists to the cash data that were the basis for GFSM 1986 in order to capture a better picture of the economic impact of government operations (inclusion of in-kind transfers, recognition of interest due, addition of expenditure arrears, adjustments for complementary periods, etc.).5 The operations statement is complemented by a cash-flow statement, in recognition of its continued relevance in assessing government’s operations.

    • The other differences are changes in classification and in some recommended breakdowns of aggregates. These are shown in Appendix I. The most relevant one is the focus, as a bottom line summary of the fiscal stance, on the “Net lending/borrowing,” rather than on the “Overall Balance,” —the former being obtained from the latter by moving below the line the government’s lending operations.

Figure 1.
Figure 1.

The Integrated GFSM 2001 Framework

Citation: Policy Papers 2010, 012; 10.5089/9781498337847.007.A001

6. For ease of presentation, the GFSM 2001 approach could be summarized through two tables, whose use was tested in the pilot studies. The first one is the operations statement (or the cash-flow statement), and the second one is a balance sheet showing how the initial balance sheet position is linked to the final position through transactions and other changes.

Figure 2.

Proposed Fiscal Presentation

Citation: Policy Papers 2010, 012; 10.5089/9781498337847.007.A001

    Table 1.

    Government Operations Table

    article image
    Table 2.

    Government Integrated Balance Sheet

    article image

    III. The Pilot Studies—and Related Experience

    7. Staff conducted seven pilot studies. Box 1 shows an illustrative extract. The experience of European Union member countries in implementing the accrual-based European System of Accounts (ESA95) also provides important lessons—since, by design, the GFSM 2001 is closely aligned with ESA95.6 The main conclusions that can be drawn from the pilots are the following:

    • Conversion of existing data into the GFSM 2001 presentation is relatively easy, at least for the operations table and assuming some key data series are available. Ideally, fiscal data in line with GFSM 2001 would be compiled from fully compatible government accounting systems—with accrual accounting and charts of accounts aligned with the GFSM 2001 classification. However, as in those EU member states which still rely predominantly on cash accounting, the alternative is to make adjustments to national data (for instance, adding significant noncash items and adjusting for lags in payment)—so that the presentation is in line with GFSM 2001 variables even if the underlying systems remain to be aligned. The pilots show that the type of bridge tables widely used in Europe to determine compliance with Maastricht criteria can also be used satisfactorily in other Fund member countries.

    • The GFSM 2001 presentation can be produced for countries at various degrees of statistical sophistication. The pilot studies included countries with purely cash data, accrual-based data, and a mixture of both. While the GFSM 2001 standard for coverage is general government (as was the case for the GFSM 1986), the presentation was found to be easily applied also to narrower or broader definitions of government (e.g., budgetary central government or the non-financial public sector). Likewise, even in countries without official government balance sheets, it was almost always possible to construct broadbrush-but-nonetheless-informative aggregate financial balance sheets, based on monetary surveys and debt data. However, only a few advanced countries have comprehensively valued government non-financial assets, so the preparation of full balance sheets remains a long-term goal.

    • Resource needs for introducing the GFSM 2001 presentation are reasonable, assuming one does not aim at perfection and that some key series are available. The pilots and unofficial exercises showed that, while there is an initial learning curve as economists become accustomed to the new presentation, in general the resources required for applying the presentation within the Fund are likely to be reasonable—where the presentation is defined to include the amended government operations table and a financial balance sheet (not including real assets). Resource needs were found to be modest for Fund economists in the case of countries already using the GFSM 2001 and in other cases amounted to about one to two days per country to recast other fiscal data in other formats, including basic backcasting.7

    • Data reporting by countries remains uneven. In many cases, the authors of the pilot studies had to identify multiple sources and construct bridge tables to create the GFSM 2001-compliant tables. Moreover the exercises highlighted varying gaps in data reported—whether under GFSM 1986 or GFSM 2001. It is clear that—although there has been significant progress—improvements in data reporting remain a core agenda item. That said, as described above, the GFSM 2001 framework can accommodate existing data without any loss of the information currently available, while showing clearly areas where data are not complete or raise consistency questions. This makes application of the framework, even in countries where data remain poor, extremely useful for guiding economists (and countries) to identify data problems and priorities.

    • Other than in advanced economies, few governments have introduced accrual reporting systems. A greater number, however, have aligned their charts of accounts with the GFSM 2001 classifications.

    The Uruguay Banking Crisis of 2002—Extract from a Case Study Using GFSM 2001

    Uruguay: Balance sheet for the Combined Public Sector based on the GFSM 2001, 2000-2002 1/

    (In percent of GDP)

    article image
    Sources: IMF Country Report No. 04/172 and STA staff estimates.

    The Combined Public Sector includes the Central Government, Extrabudgetary Funds, Social Security Funds, Local Governments, Non financial Public Enterprises, and the Central Bank.

    Other economic flows record holding gains and losses and other changes in the volume of assets and liabilities. For 2000 and 2001 information is available to break this aggregate into its two components.

    Data from IMF Country Report No. 03/247 August 2003, T able 15. Uruguay: Public Debt, plus own estimates of monetary liabilities of the BCU based in official data.

    The breakdown in domestic and foreign was not available for this case study, but it is reported by the Uruguayan authorities.

    The example of Uruguay in the years culminating in the 2002 banking crisis illustrates the impact of balance sheet operations and valuation changes on the government’s financial position.

    • The February 2003 staff report included two measures of the fiscal deficit: the deficit without one-time costs of bank restructuring, which was 4.6 percent of GDP, and the augmented balance including all bank restructuring as expenditure, which was 21.4 percent of GDP. The first indicator understated the fiscal impact of government interventions in the crisis, and the second overstated it.

    • The GFSM 2001 deficit is 7.6 percent of GDP (the change in net financial worth due to government transactions), including 3 percent of GDP in unrequited capital transfers to banks.

    • Also, the government used international reserves, IMF support, and development bank loans to on-lend to other failing banks—in the amount of 9 percent of GDP. While this support would not in itself have reduced net worth (assuming the banks would pay it back), the depreciation of the peso caused significant holding losses for government, which it was unable to pass on to the domestic banks. Other economic flows of 8.6 percent of GDP reflect assumption of mortgage bank debts.

    • Thus, public sector net worth declined by about 44 percent of 2002 GDP with another 9 percent value at risk (given the uncertainty about whether banks would pay back their loans).

    8. In sum, the pilot studies confirmed expectations that making the presentational shift to the new framework would be possible in most cases, but that improving the underlying data will take time. With respect to the latter, countries’ uneven progress in implementing the underlying reforms in public financial management systems suggests that it would not be reasonable for the Fund to set specific objectives for a shift to accrual reporting—but rather to let the development of best-practice public sector accounting move at countries’ chosen pace.

    IV. Other Progress in Implementing the GFSM 2001

    9. On a conceptual level, the GFSM 2001 framework is now well established. The Fund’s fiscal transparency code and the data quality assessment framework (DQAF) used for the data module of the ROSC advocate the GFSM 2001 and evaluate its use. By design, the European Union’s data presentation for fiscal monitoring in the context of the Excessive Deficit Procedure is closely aligned.8 A project on harmonization of GFSM 2001 variables with public sector accounting was conducted in 2003-2006 with the International Public Sector Accounting Standards Board (IPSASB); and in 2009, the Fund hosted an IPSASB meeting to support the move to global standardization of fiscal accounting.9 Other statistical manuals, such as the System of National Accounts 2008 and the sixth edition of the Balance of Payments and International Investment Position Manual (BPM6), refer to the GFSM 2001 and use the same definitions and terminology.10

    10. STA has pursued a work program of training and technical assistance. Since 1999, STA has provided training on the GFSM 2001 at 36 courses, workshops, and seminars, to 1,200 participants and delivered 260 technical assistance missions. It also introduced a section of companion materials with compilation guidance on the IMF website11 and is currently drafting the Public Sector Debt Statistics Guide.

    11. The GFSM 2001 has by now been widely adopted by countries for statistical compilation. As shown in Appendix IV, 124 countries already use the GFSM 2001 in reporting GFS Yearbook data to the Fund. Of these, 83 cover general government and its subsectors (though with data of varying quality and timeliness). Fifty-five countries compile GFSM 2001 data directly, while others convert national presentations or GFSM 1986 data to GFSM 2001 presentations for inclusion in Fund publications.12 As in other areas of statistics, countries can adopt the GFSM 2001 without being able to produce complete data on stocks and flows, but must ensure that the definitions are consistent with standards (or divergences noted), and that the framework is used for identifying areas for future progress.

    12. As part of the Fund’s response to the financial crisis, staff drafted a Guidance Note in September 2008 to provide well-defined and consistent recording of government interventions based on the GFSM 2001. This note has been widely used to record various government rescue operations and stimulus packages in staff reports and programs. Staff is also participating in the IPSASB Task Force on accounting of major government interventions related to the global financial crisis, to ensure consistency between accounting and statistical standards.

    13. The World Economic Outlook (WEO) will adopt the GFSM 2001 presentation for the government operations table starting with the Winter 2010 round. In 2009, FAD, RES, and STA collaborated to harmonize the questionnaires that are used to collect data for the WEO. The WEO’s fiscal data questionnaire is to be in line with the GFSM 2001 presentation starting with the Winter 2010 WEO round; however, reflecting data currently available, the questionnaire is limited to transaction flows and debt—and completion of all lines in the questionnaire will be expected for systemically important countries only.13 In some cases, approximations were used because of limited data. Updates have already been made in the April 2010 WEO database for early adopters of GFSM 2001 and the new data are already widely circulated in the Fund.

    14. Staff is also working to improve data on government liabilities. The Fund currently does not have a complete database for public sector debt. Staff is working with the World Bank to establish such a database for developing countries in coordination with the OECD’s database on central government debt of OECD countries. In addition, the envisaged Public Sector Debt Statistics Guide, being drafted by STA, and its supporting seminars and training, are designed to improve the availability, quality, and consistency of public sector debt statistics.

    15. Despite the progress described above, an examination of staff reports and IMF publications shows that a range of definitions is used in reporting fiscal data. Staff reports and IMF publications currently use a mix of GFSM 2001, GFSM 1986, and other systems in reporting fiscal data, and presentations and definitions vary sharply.

    Adoption of the GFSM 2001 framework would bring more consistency and clarity to the presentation of public finance statistics.

    V. Proposed Migration Strategy for Implementation of the GFSM 2001

    16. A full migration of the Fund towards GFSM 2001 would require a full adoption in member countries of the statistical methodology underlying GFSM 2001, including full accrual accounting. This will take time. But there are steps that can be taken in the short- to medium-term by staff to move the presentation of fiscal data in staff reports closer to the GFSM 2001 methodology.

    17. In designing these steps, the following considerations are critical:

    • Fund documents focused on cross-country developments and comparisons (such as the WEO) must ensure, as much as possible, comparability across countries.

    • Fund documents focused on specific countries—like staff reports for Article IV consultations—must ensure the policy dialogue with country authorities is as effective as possible; this implies that, together with the harmonized GFSM 2001 presentation, staff reports should continue to report the key fiscal aggregates according to the authorities’ definition.

    • Staff reports for surveillance and use of Fund resources are not tools for statistical reporting, per se. However, they should report all information that is needed to support the policy discussions and conclusions of the reports.

    • There is a need to ensure the consistency of fiscal data presentations throughout the duration of existing Fund-supported programs.

    18. These considerations suggest the following steps for the implementation of the GFSM 2001 presentation (see Appendix VI):

    • The fiscal data reported for WEO purposes should follow the GFSM 2001 methodology. As noted, this will be implemented with the 2010 Winter WEO round for which, for example, the concept of “net lending/borrowing” will replace the concept of “overall balance” for all countries. It should be noted that many countries were already reporting “net lending/borrowing,” so no change will be needed in many cases.

    • Staff reports issued after May 2011 should include presentations of the operations table in the GFSM 2001 format—expanded, if needed, to include key aggregates in the authorities’ presentation.14 The relevant operations table is the transactions flow fiscal table in Appendix VI, Table 1; other presentations (GFSM 1986 or ad hoc presentations) should be discontinued.15 A cash-flow statement could also be included, as appropriate (see Appendix VI, Table 2).16 17 Further breakdowns beyond those identified in Appendix VI, Tables 1 and 2, could be added, as needed. Additional indicators may also be added, as appropriate.

    • As of May 2011, information on the government’s stock of financial assets should be regularly reported together with information on gross debt; this stock information should be supplemented with information on “other flows,” if available. The inclusion of an integrated balance sheet table in the format presented in Appendix VI, Table 3 is best practice whenever large changes in the composition (amount or valuation) and/or amount of assets and liabilities take place.18 This information could be included as part of the debt sustainability analysis annex. Staff should analyze closely the link between flow operations and changes in the balance sheet position of governments and include such analysis in staff reports as needed. Gross debt should be recorded at nominal value, and also market value if available.

    • Deviations from the GFSM 2001 methodology due to data availability will be acceptable but should be flagged in footnotes. In general, staff report tables should identify any fiscal data for which “staff estimates” were constructed to account for incomplete data provided by the authorities.

    • The above information should be reported, to the extent possible, for the general government (or wider aggregates when appropriate). In any case, all Fund publications should specify the coverage of their data, according to the institutional structure defined in Appendix III.

    • For Fund-supported programs in existence prior to May 2011, the shift to GFSM 2001 can be delayed until completion of the program. However, GFSM 2001 tables could usefully be shown in an appendix of the staff report.

    • Staff would report back to the Executive Board for information on the implementation of this migration by end-2013.

    Table 1.

    Statement of Operations – [Institutional Coverage

    article image

    For financing, the GFSM 2001 allows for a classification, under each instrument, by the sector of the counterparty. For residents (domestic), these are: general government, central bank, deposit-taking financial corporations except the central bank, other financial corporations, nonfinancial corporations, and households and nonprofit institutions serving households. For nonresidents (foreign), these are: general government, international organizations, financial institutions except international organizations, and other nonresidents.

    The instrument classification is consistent with the 2008 System of National Accounts.

    Table 2.

    Statement of Sources and Uses of Cash – [Institutional Coverage]

    article image

    For financing, the GFSM 2001 allows for a classification, under each instrument, by the sector of the counterparty. For residents (domestic), these are: general government, central bank, deposit-taking financial corporations except the central bank, other financial corporations, nonfinancial corporations, and households and nonprofit institutions serving households. For nonresidents (foreign), these are: general government, international organizations, financial institutions except international organizations, and other nonresidents

    The instrument classification is consistent with the 2008 System of National Accounts.

    Table 3.

    Integrated Balance Sheet – [Institutional Coverage]

    article image
    Note: Nonfinancial assets are valued at [….], financial assets are valued at [....], and liabilities are valued at [....].

    For financing, the GFSM 2001 allows for a classification, under each instrument, by the sector of the counterparty. For residents (domestic), these are: general government, central bank, deposit-taking financial corporations except the central bank, other financial corporations, nonfinancial corporations, and households and nonprofit institutions serving households. For nonresidents (foreign), these are: general government, international organizations, financial institutions except international organizations, and other nonresidents.

    The instrument classification is consistent with the 2008 System of National Accounts.

    19. As noted, a full implementation of GFSM 2001 will require changes in the way countries collect and compile their statistics. Thus, staff will continue to work with countries to fill their data gaps and bring their coverage and data sources closer to the GFSM 2001 standards—i.e., improving countries’ ability to report statistics in line with the GFSM 2001.19 Progress in this effort will benefit from the continuation of STA’s technical assistance program. The main improvements needed are: extension of institutional coverage to general government or the public sector; expansion of transactions coverage (mainly to noncash items); establishment of an accrual basis of recording; valuation of assets and liabilities; and inclusion of a functional classification of outlays.

    20. In the long term, the goal should be that a broad section of the Fund’s membership fully adopts the accrual-based GFSM 2001 and regularly compiles and disseminates all its statistical tables accordingly.20 The regular fiscal reporting of the EU27 countries may be taken as a model. This will best be achieved by PFM reforms to update countries’ underlying accounting and reporting systems. The Fund cannot mandate full implementation of the GFSM 2001 methodology in countries, as this remains the prerogative and responsibility of national authorities. Hence, this paper does not suggest that any specific deadline be targeted for complete implementation of the GFSM 2001. The Fund will, nonetheless, promote full implementation of the GFSM 2001 framework, as it should remain a goal for both countries and the Fund to be attained in the next five to ten years.

    Resource requirements

    21. Resource requirements for area departments, FAD, and STA to implement the steps identified in paragraph 14 are expected to be undertaken within the given budget envelope. Resources will mainly be required on a one-off basis.21 Many countries already produce GFSM 2001 data, as shown in Appendix IV. For other countries, while the authorities will undertake a complete conversion based on detailed line-items, Fund staff doing recasting will need to make adjustments only at the aggregate level used for our fiscal monitoring. The full implementation of the GFSM 2001 methodology in countries is the responsibility of national authorities, so it depends on their available resources, country-specific circumstances, and needs identified. Fund staff will, nonetheless, continue promoting implementation of the GFSM 2001 framework as resources permit.

    VI. Proposed Decision

    Accordingly, the following decision, which may be adopted by a majority of the votes cast, is proposed for adoption by the Executive Board:

    1. The Fund hereby approves the migration strategy for implementation of the GFSM 2001 as set forth in Section V of Government Finance Statistics to Strengthen Fiscal Analysis, February 26, 2010.

    2. It is expected that the Fund will review the implementation of the migration strategy for implementation of the GFSM 2001 by December 31, 2013.

    Government Finance Statistics to Strengthen Fiscal Analysis
    Author: International Monetary Fund