Excessive Deficit Procedure Statistics Working Group

Transcription

1 EUROPEAN COMMISSION EUROSTAT Directorate D - Government Finance Statistics (GFS) and quality Unit D-1: Excessive deficit procedure and methodology Unit D-2: Excessive deficit procedure (EDP) 1 Unit D-3: Excessive deficit procedure (EDP) 2 Unit D-4: GFS. Risk and quality management Excessive Deficit Procedure Statistics Working Group 2 4 December 2015 Ampere Beginning at 13:00 p.m. Part B Item 5.a Eurostat data collection on contingent liabilities: User perspective (DG ECFIN) Fiscal risks and contingent liabilities an overview (Provisional version. Final version will be published in "Report on Public Finances in EMU 2015") Joseph BECH Building Tel.: (+352) L Luxembourg Fax: (+352)

2 It is not unusual that the general government's deficit or debt are significantly and unexpectedly affected by the materialisation of fiscal risks associated with contingent liabilities. Examples of this include calls on guarantees issued by government, or government stepping in to cover obligations of a public corporation when it experiences financial difficulties. A distinctive feature of such events is that they stem from some explicit or implicit obligations that the state has previously established (i.e. a guarantee contract or being a majority owner), but the extent and the timing of this impact cannot be determined with certainty in advance. Such liabilities are called contingent, as they result in transactions only where certain specific conditions prevail, and are not recorded in core national accounts. Nevertheless, the management of public finances cannot ignore these liabilities, because of their potential impact on public deficit and debt. The availability of comparable information is the first step and a necessary precondition to understanding and mitigating such risks. A recent publication by Eurostat of new time series can shed light on Member States' contingent liabilities, covering in particular government guarantees, liabilities of public-private partnership agreements (PPPs) that are recorded off government's balance sheets, liabilities of government-controlled entities outside government sector, and information on non-performing loans issued by government. The next sections review the new dataset, including factors that need to be taken into account when interpreting the data, explores the ways how contingent liabilities can materialise as actual losses for the government, and provides some examples of mitigating fiscal risks associated with contingent liabilities. Fiscal risk and the role of contingent liabilities Fiscal risk is a broad term capturing a variety of reasons for fiscal outcomes being different from the targeted (or forecasted) levels. The realisation of these risks is usually associated with either macroeconomic developments that have an impact on fiscal variables or with situations where governments have to make payments that were not foreseen in a budget law or other ex-ante estimate of expenditure. The current overview focuses on the narrower definition of fiscal risks, one that is associated with contingent liabilities. The analysis below does not cover fiscal risks that originate from the evolution of macroeconomic variables and concentrates instead on those liabilities of the state which under certain conditions trigger state's response, in turn resulting in changes to the anticipated deficit and debt levels. The degree of government's discretion in activating this response can vary, but a common feature is the existence of some underlying commitment, which can be formalised or not. Polackova Brixi and Mody (2002) provide a Government Fiscal Risk Matrix, where they distinguish, firstly, between explicit liabilities (government liability as recognised by a law or contract, e.g. a loan or guarantee agreement) and implicit liabilities (a moral obligation of government that reflects public and interest group pressures, e.g. future public pensions or public investment projects), and, secondly, between direct liabilities (obligation in any event, e.g. expenditure defined in existing laws) and contingent liabilities (obligation if a particular event occurs, e.g. intervention in case of a failure of a public or private entity). Implicit and contingent liabilities are thus not mutually exclusive categories but different dimensions of categorisation. The terminology is also subject to rather diverse interpretations. For example, the definition of contingent liabilities in the Fiscal Risk Matrix quoted above is macroeconomic in nature, and it differs from definitions used for national accounts (ESA2010, ) or - 2 -

3 accounting (IAS 37 / IPSAS 19) purposes. In particular, this macroeconomic framework also covers the subset of implicit contingent liabilities, for example government interventions in case of systemic bank failures (beyond deposit or unemployment insurance), where no explicit obligation exists for the state and only macroeconomic estimates of their magnitude can be made for a specified set of assumptions. The scope of implicit liabilities is difficult to define, but this does not mean that they could be ignored by policymakers. For example, the approaches to estimating cost of ageing are presented in the 2015 Ageing Report ( 1 ), while contingent liabilities linked to public support to the banking sector estimated using SYMBOL model were presented in the 2012 Fiscal Sustainability Report ( 2 ). Availability of information on potential exposure is the first step in assessing impact of a contingent liability. The availability of information on the extent of potential exposure is the necessary precondition and a starting point for the quantification. However, even when the potential exposure can be assessed (which can only be done with sufficient degree of certainty for some types of contingent liabilities, as discussed above), the full picture will also require an estimate of probability of materialisation of the risk. In most cases this information cannot be easily obtained for individual cases either. Yet, for some classes of explicit contingent liabilities, which involve large number of homogenous contracts for example export guarantees or student loan guarantees probability of guarantees being called can be established with sufficient degree of certainty at the aggregate level. This allows revealing the risk and reflecting it appropriately in the government finance statistics: under ESA2010 such standardised guarantees are recorded by analogy to non-life insurance, with fees treated similarly to insurance premiums and the transaction giving rise to financial assets and liabilities (Eurostat, 2014). The need for transparency with regard to contingent liabilities is well recognised, especially in relation to explicit contingent liabilities. ( 3 ) For example, the IMF s Fiscal Transparency Code (IMF, 2014) attributes a separate section to the management of fiscal risks, including principles that relate to disclosure and management of government's exposure to risks stemming from guarantees and Public Private Partnerships. Likewise, OECD s Best Practices for Budget Transparency (OECD, 2002) recommend that (i) all significant contingent liabilities should be disclosed in the budget, the mid-year report and the annual financial statements and (ii) where feasible, the total amount of contingent liabilities should be disclosed and classified by major category reflecting their nature; historical information on defaults for each category should be disclosed where available; in cases where contingent liabilities cannot be quantified, they should be listed and described. International Public Sector Accounting Standard (IPSAS) nr 19 on Provisions, Contingent Liabilities and Contingent Assets requires that unless the possibility of any outflow in settlement is remote, an entity should disclose for each class of contingent liability a brief description of the nature of the liability and, where practicable, an estimate of its financial effect (IPSASB, 2014). While these internationally agreed standards and guidelines provide a basis for improved availability of information on contingent liabilities, the requirements therein remain non- ( 1 ) ( 2 ) ( 3 ) This conclusion is less straightforward in case of implicit contingent liabilities due to the risk of moral hazard, see e.g. IMF(2007) - 3 -

4 binding and no precise data specifications exist, limiting comparability of nationally published data. Several countries and international organisations have also started working on the ways to mitigate fiscal risks related to contingent liabilities. A recent discussion paper by the OECD (OECD, 2013) suggests that a key objective of policy reforms mitigating the risks stemming from guarantees is to promote neutrality by ensuring that policy makers have enough information to compare their effects with those of direct budgetary appropriations, and by levying market-based fees that reflect the expected cost of such liabilities for the government. The discussion paper also advocated full and transparent disclosure of potential costs and reducing the extent of implicit liabilities by either credibly eliminating them or by converting into explicit liabilities. Some Member States have already introduced reforms in the area of contingent liabilities. For example, in the Netherlands the approach to issuance of guarantees is "no, unless" (e.g., unless risk cannot be insured on the market); guarantees are subject to a premium which is treated in the budgetary process as provision for future losses (Hofmans and van de Coevering, 2014). In Sweden the Budget Act adopted in 2011 contained clearer rules for government guarantees and new provisions regarding government lending with credit risk; the rules aim to ensure that that such commitments are self-financed in the long run, with fees corresponding to expected costs (Swedish National Debt Office, 2015). In Latvia, the Fiscal Discipline Law envisages that a fiscal risks declaration is prepared annually, including quantification of fiscal risks; the quantification of the risks in this declaration determines the size of the fiscal safety reserve in the budget (Fiscal Discipline Council of Latvia, 2015). In Portugal, a separate unit with specialised expertise has been established in order to assess and monitor public-private partnerships (PPPs) and to mitigate risks stemming from these contracts (European Commission, 2014). Improvements in public finance statistics provide another way of mitigating such risks. The statistical framework is constantly evolving, and many improvements have been introduced in recent years, notably through ESA2010 which became operational in September These innovations include qualitative aspects on control, which led to reclassification of several government-controlled entities inside the general government; new treatment of standardised guarantees in ESA2010, which takes expected losses into account already at the inception; evolving rules on determining economic owner of a PPP asset etc. As a consequence, several risks are now captured at earlier stages for example, losses of a government-controlled company classified inside general government would affect general government's net borrowing gradually as they occur, not postponing their materialisation into future when a large capital injection is needed. New data requirements supporting fiscal surveillance in the EU In the EU, analysis and reporting related to contingent liabilities have been considerably strengthened since the adoption of the so-called "six-pack" in 2011, and the subsequent adoption of the "two-pack" for the euro area. Recognising the importance of risks associated with explicit and implicit contingent liabilities, recent measures to strengthen economic governance in the EU have significantly sharpened the focus on such liabilities in the budgetary analysis. As a result, the reporting requirements related to such liabilities were strengthened as well (see Box 1). In particular, under the corrective arm of the Stability and Growth Pact the contingent liabilities are now taken into account when the Commission prepares a report - 4 -

5 Box 1. Provisions for reporting fiscal risks in the EU budgetary framework The extent of provisions in the EU budgetary surveillance framework that foresee reporting related to fiscal risks has significantly increased in last years as a result of the adoption of the so-called "sixpack" and "two-pack" packages, which aim at strengthening economic governance and which were adopted correspondingly in 2011 and Regulation (EU) No 1175/2011 amending Council regulation (EC) No 1466/97 on the strengthening of the surveillance of budgetary positions and the surveillance and coordination of economic policies ( 4 ) (the preventive arm of the pact) added a requirement that the stability and convergence programmes shall include information on implicit liabilities, such as public guarantees, with potentially large impact on the general government accounts. Regulation (EU) No 1177/2011 amending Council Regulation (EC) No 1467/97 on speeding up and clarifying the implementation of the excessive deficit procedure ( 5 ) (the corrective arm of the pact) added a requirement that the Commission, when preparing a report under Article 126(3) TFEU, shall reflect among other factors, as appropriate, guarantees, in particular those linked to the financial sector, and any implicit liabilities related to ageing and private debt, to the extent that they may represent a contingent implicit liability for the government. In order to make possible the implementation of these legal acts, the Code of Conduct governing the stability and convergence programmes was amended; in particular reporting on longterm sustainability of public finances has been enhanced and reporting on public guarantees, including those linked to the financial sector, has been added as a compulsory item. The most notable improvement in making available information on contingent liabilities was established through Directive 2011/85/EU on requirements for budgetary frameworks of the Member States ( 6 ), which also is part of the "six-pack". Article 14(3) of this Directive established new transparency requirements for the Member States: "For all sub-sectors of general government, Member States shall publish relevant information on contingent liabilities with potentially large impacts on public budgets, including government guarantees, non-performing loans, and liabilities stemming from the operation of public corporations, including the extent thereof. Member States shall also publish information on the participation of general government in the capital of private and public corporations in respect of economically significant amounts". As subsequently required by Regulation (EU) No 473/2013 on common provisions for monitoring and assessing draft budgetary plans and ensuring the correction of excessive deficit of the Member States in the euro area Member States ( 7 ), the financial risks associated with contingent liabilities with potentially large impacts on public budgets shall also be covered in a comprehensive assessment of inyear budgetary execution, to the extent that they may contribute to the existence of an excessive deficit in euro area Member States. These Member States are also encouraged to include data on contingent liabilities that could affect the medium-term government in the draft budgetary plans submitted to the Commission for monitoring purposes. The corresponding Code of Conduct reiterated this requirement with respect to additional reporting requirements that the Commission can request from a Member State in EDP, and added reporting on government guarantees, including those linked to the financial sector, as a non-compulsory indicator to be included in draft budgetary plans. ( 4 ) OJ L 306, , p. 12 ( 5 ) OJ L 306, , p. 33 ( 6 ) OJ L 306, , p. 41 ( 7 ) OJ L 140, , p

6 under article 126(3) of the TFEU; in case of euro area countries in EDP the Commission may now request additional reporting in the form of a letter, including on financial risks associated with contingent liabilities with potentially large impacts on public budgets. Under the preventive arm, contingent liabilities are mainly analysed as possible risks to the budgetary outlook. The requirements established through the Directive 2011/85/EU led to the publication by Eurostat of new data series for all EU Member States. The specific operational requirements on coverage and timeliness of data related to contingent liabilities as required by Article 14(3) of the budgetary frameworks directive were agreed between the Member States and the European Commission in 2013 and the first set of time series was submitted by Member States to Eurostat by the end of Eurostat published this information and an accompanying press release ( 8 ) in February New dataset on contingent liabilities in Eurostat's database Eurostat's publication covers three classes of contingent liabilities, as well as data on non-performing loans. Information on contingent liabilities relates to (i) guarantees, (ii) public-private partnerships recorded off balance sheet of government and (iii) liabilities of government controlled entities classified outside government, as described in further detail below. In addition, Eurostat's dataset covers information on non-performing loans issued by government. A dedicated webpage ( 9 ) provides additional information, including legal background, country-specific footnotes and reports. The summary Table 1 below gives an overview of currently available information in Eurostat's database. In most cases, updated annual time series for the previous year(s) are to be transmitted to Eurostat before the end of the year. Annual extension of these time series in the database is expected in early T+2, i.e. information for 2014 will be submitted to Eurostat by the end of 2015 and added to current time series in early Table 1: Availability of indicators related to the implementation of Directive 2011/85/EU in Eurostat's database (as of end-2015) This EU level publication is accompanied by information published at the national level. Years Sectors* Units While Eurostat publishes few time series that are (compulsory) Contingent liabilities 1. Guarantees - Total stock of government guarantees S.13 Millions of 1.A One-off guarantees - Total stock of guarantees, excluding debt assumed by government based on commonly agreed standard definitions, S.1311 T-4, T-3, T-2, T-1 national of which: guarantees to public corporations S.1312 ( ) currency; Memo item: guarantees to financial corporations S.1313 Percent of GDP 1.B Standardised guarantees - Total stock of government guarantees Member States are encouraged to publish more S.13 Millions of S.1311 T-1 national 2. Adjusted capital value of off-balance private-public partnerships (PPPs) S.1312 (2013) currency; detailed information nationally, and several S.1313 Percent of GDP 3. Total outstanding liabilities of government controlled entities classified outside general S.13 T-1 (T-2 if not Millions of government S.1311 Member States do so. In addition to providing available) national 3.A Liabilities of units involved in financial activities S.1312 (2013 or 2012 if currency; 3.B Liabilities of units involved in other activities S.1313 not available) Percent of GDP of which: liabilities of loss-making non-financial units S.1314 more details with regard to contingent liabilities, Assets that can result in losses S.13 Millions of S.1311 T-1 national national publications also cover other 4. Stock of non-performing loans provided by government S.1312 (2013) currency; S.1313 Percent of GDP S.1314 information that was required by Directive * S.13 = General government, S.1311 = Central government (excluding social security funds), S.1312 = State government (excluding social security funds), S.1313 = Local government (excluding social security funds), S.1314 = Social security funds 2011/85/EU, notably information on the participation of the general government in the Source: Commission services capital of private and public corporations, highfrequency fiscal data according to national public accounting methodology and the reconciliation table explaining the transition from this high- ( 8 ) ( 9 )

7 frequency data to ESA-compliant data standard. The national publications are often tailored to specific country situations and do not allow easy cross-country comparison; Eurostat maintains a list of relevant national websites( 10 ). At the same time, national publications improve fiscal transparency within countries, contributing thus to better oversight of fiscal policy by national monitoring institutions and by the society at large. Not all countries have so far made full use of this possibility to improve transparency, limiting their publication to few compulsory variables. Government guarantees Guarantees issued by government on the liabilities (and sometimes assets) of third parties constitute the most common form of contingent liabilities. Only explicit or formal obligations, fixed in form of a law or a contract, are covered by Eurostat's dataset. ( 11 ) For example, a debt guarantee is an arrangement in which a guarantor (government) agrees to pay a creditor if a debtor defaults; guarantees on assets entail government's commitment to cover losses related to a decrease in value of the assets. Both public and private corporations, as well as households, could benefit from government's guarantees. Usually the aim of a government guarantee is to allow the beneficiary unit to borrow at a lower interest rate, which could correspond to certain policy objectives (for example, improving access to higher education through guarantees on student loans), but they are also used to improve access to markets (for example, export guarantees or guarantees to financial institutions in times of financial distress). Eurostat's publication contains several subsets facilitating analysis of the data, however in some cases the reported data are not complete or/and exhaustive. The publication distinguishes between standardised and one-off guarantees ( 12 ) (see also Table 1 above); in most (but not all) countries the latter category is by far more important. The separate time series relate to guarantees issued to public corporations and guarantees to financial (public and private) corporations; only guarantees provided to units classified outside general government are covered by the reporting. As can be seen on Graph 1, central governments are the most frequent issuers of guarantees, although guarantees issued by state or local governments are also important in some countries. The time series currently cover four years, , and are reported at nominal value; information corresponds to total stock of outstanding guarantees, excluding the debt already assumed by government. There are, ( 10 ) ( 11 ) According to the agreed data specifications, data do not include government guarantees issued within the guarantee mechanism under the European Financial Stability Facility (EFSF - since the liabilities of the EFSF guaranteed by Member States are predominantly included in the debt of Member States issuing such guarantees directly), derivative-type guarantees (when they meet the definition of a financial derivative and are recorded correspondingly in national accounts), deposit insurance guarantees and compatible schemes (given that these are covered by separate frameworks) and guarantees issued on events which occurrence is very difficult to cover via commercial insurance (like earth quakes, large scale flooding, nuclear accidents, certain art exhibitions etc., given very large uncertainties in establishing value of such guarantees). ( 12 ) As explained in the introductory section, ESA2010 distinguishes between standardised guarantees (which are characterised by frequent repeated transactions with similar features and pooling of risks, where average loss can be estimated) and one-off guarantees (which are individual agreements where reliable loss estimates cannot be made)

8 however, gaps in the reporting: for example, information on standardised guarantees is missing for Belgium, Croatia and Portugal, and information on guarantees issued by local governments has not been provided by France, Croatia and Slovenia; other country-specific comments and remarks are included under "Country-specific footnotes". ( 13 ) Graph 1: Outstanding guarantees by level of government, 2013, % of GDP Source: Eurostat Contingent liabilities associated with guarantees will become actual costs in case guarantees are called. Guarantees constitute contingent liabilities and are therefore not recorded in core national accounts, except in cases described below. In ESA2010, when a standardised guarantee is granted, a liability should be recorded in government's balance sheet that is equal to the present value of expected calls under the guarantees, net of any recoveries. There are also other differences in the statistical treatment between standardised and one-off guarantees. Fees related to the issuance of one-off guarantees constitute revenue for the government, while in case of standardises guarantees they are recorded as a financial prepayment to cover future losses ( 14 ). When the risks materialise and a guarantee is called, call on a standardised guarantee gives rise to a financial (deficit-neutral) transaction, but in case of a one-off guarantee a partial or full call leads to the recording of a corresponding capital transfer (debt assumption), which would worsen government's budgetary position. The rules in place also foresee that the whole amount of an outstanding one-off guarantee is recorded as a capital transfer after the third partial guarantee call. The same treatment, i.e. debt assumption by the government for full amount of the one-off guarantee, applies when the likelihood of a guarantee being called is very high at inception.( 15 ) The recent financial crisis significantly affected statistics on guarantees in several Member States. The stock of outstanding guarantees related to financial corporations is particularly high in Ireland (where it was 32% of GDP in 2013, almost corresponding to the overall stock of outstanding guarantees), Spain and Austria; however, the stock of guarantees to the financial sector is now on a clear declining trend. Eurostat's reporting on financial crisis ( 13 ) ( 14 ) If government charges no fees, or fees are far from covering present value of expected calls, a capital transfer (deficit-increasing transaction) is recorded for government, reflecting the unrequited nature of the transaction ( 15 ) For more detailed overview of recording of guarantees, see Manual on Government Deficit and Debt ( Section VII.4-8 -

9 ( 16 ) provides some insight into the effect that guarantees have on public finances: for example, it illustrates that calls on guarantees to financial institutions have affected government's budgetary position in recent years in Belgium, Denmark, Spain, Latvia and Portugal. Liabilities related to public-private partnerships (PPPs) recorded off-balance sheet of government As the term implies, "public-private partnerships" (PPPs) involve two partners that belong to different sectors of the economy. These partnerships are long-term (at least 5 years and usually much longer) contractual arrangements between a government unit and a private partner, whereby the latter builds or renovates a fixed dedicated asset and uses that asset to deliver services to the government unit or directly to the public, in exchange for a periodic payment from government. ( 17 ) The key issue in national accounts is determining who is the economic owner of the asset whether the private partner or the government unit. The assets related to the PPP will be recorded on the balance sheet of the economic owner, impacting the unit's expenditure (and balance) over the construction period, and on its debt. According to ESA2010, the economic owner of an asset is the unit bearing the majority of the risks and entitled to receive the majority of the rewards related to the use of the asset. The three main risk categories considered in this respect are construction risk (related to the construction process and its costs), availability risk (related to the availability of the asset for usage after the construction phase), and demand risk (related to the demand for services related to the asset). In addition to the risk analysis, other conditions are closely analysed, such as the allocation of the assets at the end of the contract, the compensation and termination clauses, the existence of government financing and guarantees, etc. Information related to "on-balance" PPPs is directly reflected in general government's deficit and debt, but this is not the case for "off-balance" PPPs; nevertheless, the latter also imply potential liabilities for the government. A recording of the asset on the government's balance sheet means that all expenditure, revenue and financial obligations associated with the asset affect general government's deficit and debt directly, as the asset is being constructed or renovated. This is not the case for those PPPs where the private partner is deemed to be the economic owner of the asset in these cases the impact on government's statistics would be more gradual, reflecting periodic payments to the partner. However, should the private partner fail, there is a risk that government may need to take over the asset during the life of the contract, and this constitutes a contingent liability for the government. ( 16 ) ( 17 ) In the national accounts, a distinction is made between PPPs and concessions on the basis of who pays the fees to the partner: the term "PPP" is used for contracts where government is paying to a private partner all or a majority of fees associated with the use of the asset, whereas in concession agreements the majority of payments is made by final users

10 Graph 2: Off-balance PPPs, stock of adjusted capital value by the level of governments in 2013, % of GDP Source: Eurostat The dataset published by Eurostat illustrates that such "off-balance" PPPs are relevant in some countries; data reflects current value of the asset. As Graph 2 illustrates, the stock of "off-balance" PPPs exceeds 1% of GDP only in a handful of countries, and is the highest in Portugal and Cyprus. These partnerships are almost exclusively concluded by central governments, although in countries with the federal government structure state governments also play a role. Only data for 2013 were compulsory, and the coverage is overall good with only a few country-specific gaps; a great majority of countries also reported figures for the years on a voluntary basis. When interpreting the data, it should be kept in mind that they reflect for each year the current value of the project, representing an estimate of the impact on the investment expenditure and on the debt in case the government would need to take the asset over. This current value increases rapidly in the construction phase of the project, reflecting investments being made, and progressively decreases thereafter in the exploitation phase, reflecting economic depreciation of the asset. This current value does not thus reflect the amount of investment needed to finalise the asset, in case the project experiences difficulties already in the construction phase. Liabilities of government-controlled entities Governments often choose to establish corporations outside the budgetary perimeter; part of such entities would be nevertheless classified inside general government for the national accounts purposes. There is a variety of reasons why governments may decide to establish entities outside the budgetary perimeter; these entities may be established under the different legal status (e.g. limited liability companies) and combine commercial and public interest objectives. The extent to which such entities are used varies significantly across countries and there are no clear answers as to which advantages do they offer when compared to direct procurement of goods and services. As long as these units are controlled by the government, ( 18 ) they belong to the public sector. Whether a public sector unit is classified for statistical purposes as part of general government or as part of corporations' sector, depends on the nature of the unit those involved in non-market activities are classified in general ( 18 ) This control may be exercised through rights to appoint, veto or remove a majority of the governing board or key personnel; ownership of the majority of the voting interest (most commonly, ownership of >50% of shares); rights under special shares and options; rights to control via contractual arrangement, agreements or permissions to borrow, or via excessive regulation; and other forms of control

11 government and those involved in market activities are classified as public corporations. The term corporation must be understood here in a broad sense as it may include entities which do not have the legal status of a corporation. The units classified inside general government are covered by regular reporting on government deficit and debt, but this is not the case for government-controlled units outside general government their debt represents potential liability for the government. If a public sector unit is classified in the general government sector, its revenue, expenditure and debt are added to those of the rest of the government sector and thus affect government's balance and debt directly. This is, however, not the case for public corporations financial or non-financial that have a nature of market units and which thus remain outside general government. Nevertheless, participation in the capital of public corporations could represent a potential liability for the government: when the corporation is experiencing difficulties, government as a controlling entity or a majority owner may need to step in. This Graph 3: liability is contingent, i.e. it may develop into actual costs if some specific event occurs, but the intervention need cannot in most cases be quantified in advance. A government's decision to intervene may also be triggered by other reasons than financial difficulties: for example, in case of a planned privatisation of a public company, government may decide to "clean" the company's balance sheet by assuming certain obligations, including obligations of an occupational pension scheme. When such interventions have a nature of an unrequited payment (i.e. the government does not receive a payment of an equal value in return and is not Source: Eurostat expecting to earn a sufficient rate of return on its investment), they will worsen government's budgetary position. ( 19 ) Eurostat's publication on liabilities of government-controlled entities classified outside the general government is the first of its kind, but data should be interpreted with caution. Eurostat's recent publication is the first step towards establishing comparable time series on potential liabilities associated with government-controlled entities outside the general government. Nevertheless, the data reflects wide variability of models used within Member States in terms of reliance on public corporations, and this should be taken into account when interpreting it. There are also other aspects that need to be kept in mind. Firstly, data reflect gross liabilities, i.e. only debt without Liabilities of public financial institutions, by controlling government level, 2013 or 2012, % of GDP Graph 4: Liabilities of other (non-financial) public corporations by controlling government level, 2013 or 2012, %of GDP Source: Eurostat ( 19 ) The application of "the capital injection test" is described in section III.2 of the Manual on Government Deficit and Debt

12 balancing it with assets. Secondly, the reporting is non-consolidated, i.e. part of the debt could be towards another unit in the same institutional sector (for example, if two public corporations have liabilities towards each other, both amounts would be captured by the recording), and it will also cover subsidiary's debt towards its holding unit. Finally, there are some gaps in data provision in particular, information on liabilities of financial corporations is missing or incomplete in Belgium, Greece, Spain, Cyprus and Luxembourg, whereas information on corporations controlled by local governments is missing or incomplete in Greece, Ireland, Belgium and Cyprus. With these caveats in mind, the information published by Eurostat allows analysing risks associated with government's participation separately for financial and nonfinancial corporations. As Graphs 3 and 4 illustrate, gross liabilities of public financial corporations are much higher, as a rule, than liabilities of public non-financial corporations. However, interpreting these liabilities also differs in case of financial and non-financial institutions: the former tend to have large balance sheets and thus also large (gross) liabilities, but their obligations are usually counterbalanced by assets of a comparable size. Moreover, financial institutions public and private alike are covered by specific regulations like prudential supervision, deposit insurance etc., which reduce the need for owner intervention. In addition, the database available on Eurostat's website also contains, under non-financial corporations, a separate subset related to liabilities of loss-making corporations (see Table 1). Since companies experiencing occasional losses may drop in and out of reporting in any specific year, drawing conclusions on the basis of only one observation that is currently available could be misleading and therefore not illustrated here. In most cases it can also be assumed that the government would not go as far as covering all the liabilities of a corporation in difficulties, therefore the extent of reported liabilities should be understood as maximum theoretically possible loss, rather than a likely scenario. Non-performing loans Governments sometimes issue loans, although this is not part of their traditional role. Various government units may provide loans to other economic sectors or to the rest of the world, when this helps them in achieving some socio-economic or international relations policy goals, or when it helps to manage the accumulated assets in a cost-beneficial manner. Eurostat publishes quarterly information on financial accounts of general government, which provides data on the stock of government's assets in form of loans, including counterpart (i.e. debtor) sectors for central government and social security funds. The data shows, for example, that the Nordic countries tend to have relatively high share of assets of the central government (and in case of Finland social security funds) in form of loans. The stock of general governments' assets in form of loans can also indicate the presence of non-market financial institutions inside general government. It is not unusual, especially in the aftermath of the recent financial crisis, that public financial institutions cease performing traditional financial intermediation activities and are subsequently classified inside general government; in some cases inclusion of such units is a corollary to more precise definitions in ESA2010 for delimitation of the general government sector, compared to the earlier national accounts edition. Such institutions could include defeasance structures ("bad banks"), asset management companies or special purpose entities, when they are controlled by government and do not place themselves at risk. Such units would affect statistics on governments' assets, which in recent years was the case for Ireland, Spain, Latvia, the Netherlands, the UK, Austria and some other Member States

13 The loans issued by the government may become non-performing and in some cases result in losses for the government. Loans are considered non-performing when payments of interest or principal are past due by 90 days or more, or interest payments equal to 90 days have been capitalised, refinanced, or delayed by agreement, or when there are other good reasons (such as debtor filing for bankruptcy) to doubt that payments will be made in full. Such non-performing loans may lead to full or partial non-recoverability and eventually losses for the government, and these losses can take various forms. For example, a debt cancellation agreement in favour of the debtor would directly worsen the government's balance; write-off of debts after the debtor unit has been liquidated does not affect the balance directly, but implies an indirect increase in the creditor's debt level (as the borrowing needs would be higher than under the scenario when the principal and interest are fully repaid). According to Eurostat's publication, the total stock of non-performing loans issued by the government was the highest in the EU (around 11% of GDP) at the end of 2013 in Ireland, due to the inclusion of the Irish Bank Resolution Corporation, and the second-highest at around 3% in Slovenia, on account of the Bank Assets Management Company; in the majority of the Member States the amounts were, however, negligible. When the share of non-performing loans is high on account of defeasance structures being part of general government, the situation is usually expected to change in the medium term, as the amounts tend to decline rapidly reflecting winding down of positions. The coverage of the reported data remains incomplete. No data on non-performing loans has been provided by Belgium, France, Croatia, Cyprus and Slovakia, and it remains incomplete for Spain, Italy in Finland. The reporting was compulsory only for 2013, but a number of countries have provided information for on a voluntary basis. Conclusion In order to safeguard budgetary targets, fiscal risks associated with contingent liabilities need to be better understood. Fiscal risk is a broad concept capturing various reasons why fiscal outcomes deviate from targeted or forecasted levels. Among these factors, contingent liabilities are potential obligations for the state, but they materialise as actual losses and higher debt only if and when certain triggering events occur. Because of this, it is very difficult in most cases to estimate in advance the potential impact on deficit and debt and the timing of such impact. Regardless of these difficulties, such liabilities cannot be ignored, as the recent economic crisis also demonstrated. Public availability of information related to contingent liabilities is a first step and a necessary precondition for understanding and mitigating risks associated with contingent liabilities. The reporting requirements have recently been strengthened in the EU to enhance transparency of contingent liabilities. Recognising the importance of better understanding fiscal risks associated with contingent liabilities, including for the EU budgetary surveillance, several steps were undertaken in the recent years to strengthen reporting obligations in this area. In particular, as part of the implementation of Directive 2011/85/EU on budgetary frameworks, new statistical reporting was launched by Eurostat, and more detailed information is being published nationally. These new reporting obligations cover statistics on government guarantees, liabilities related to public-private partnerships (PPPs) recorded off balance sheet of government, liabilities of government controlled entities, as well as reporting on non-performing loans

14 While recognising the country specific nature of contingent liabilities, new data series allow some cross-country comparison and demonstrate that contingent liabilities tend to be high when they relate to the financial sector. The reasons why Member States use contingent liabilities instead of direct budgetary interventions often reflect historical circumstances and established institutional frameworks, which makes cross-country comparison less straightforward. Nevertheless, the data demonstrates that contingent liabilities tend to be higher in cases when they relate to the financial sector for example, liabilities of public financial corporations in some cases exceeded 50% of GDP, while they reach a maximum of ca 25% of GDP in case of non-financial corporations; guarantees to financial institutions also went as high as 30% of GDP in 2013 (which is the latest available data). These findings are not surprising, given the considerable impact that financial stabilisation measures had on deficit and debt in many countries. However, where data series cover more than one year, it can also be seen that liabilities associated with the financial sector are now on a declining trend. Contingent liabilities related to off-balance PPPs and the stock of non-performing loans are currently of macroeconomic relevance only in few countries. The ways forward include improving the coverage and comparability of the data, and finding best ways to mitigate fiscal risks associated with contingent liabilities. While the recent publication is a considerable step forward in improving the availability of comparable data in the EU, some gaps in reporting remain. Moreover, the length of time series is still very limited in most cases. Both the quality of data and the length of time series are expected to improve with time, reflecting the effort of the EU statistical community. The international organisations and Member States have also started working on the ways to mitigate fiscal risks associated with contingent liabilities. The most promising avenues involve making such liabilities self-financing or budgetary neutral in the longer run, by charging fees and making provisions that reflect expected future losses. The strengthening of institutional and statistical frameworks and improved transparency are among other ways of mitigating such risks. References European Commission (2014), The Economic Adjustment Programme for Portugal , European Economy, Occasional Papers, 202 Eurostat (2014), Manual on the changes between ESA 95 and ESA 2010 ( ) Fiscal Discipline Council of Latvia (2015), Fiscal discipline monitoring report 2015, nexes.pdf Hofmans, Heleen M.J. and van de Coevering, Clement R., How to deal with contingent liabilities Lessons from the Dutch experience, OECD Journal on Budgeting, Vol. 2014/1 IMF Fiscal Affairs Department (2014), The Fiscal Transparency Code ( IMF Fiscal Affairs Department (2007), Manual on Fiscal Transparency (

15 International Public Sector Accounting Board, IPSASB (2014), Handbook of International Public Sector Accounting OECD (2013), Budgeting for Contingent Liabilities, Discussion Paper ( 013)7&doclanguage=en) OECD (2002), Best Practices for Budget Transparency ( Polackova Brixi, Hana and Mody, Ashoka (2002), Dealing with Government Fiscal Risk: An Overview, World Bank and Oxford University Press The Swedish National Debt Office, 2015,