The International Monetary Fund (IMF) played a ground-breaking role in understanding the financial-sector dynamics of the euro-area crisis. It was the first public authority, and one of the first more generally, to acknowledge the role of the bank-sovereign vicious circle as the central driver of contagion in the euro area. It was the first public authority to articulate a clear vision of banking union as an essential policy response, building on its longstanding and pioneering support of banking policy integration in the European Union.
At national level, the IMF’s approach to the financial sector was appropriate and successful in Ireland and Spain, more limited in the Greek Stand-By Arrangement, and less compelling in Portugal where vulnerabilities remained when the country exited the programme.
The IMF should further integrate financial-sector policy together with fiscal and macroeconomic issues at the core of its operations, and should devote particular effort to adapting its processes and methodologies to the new context of European banking union.
Financial sector aspects have pervaded the euro-area crisis, which can be seen as much as a financial sector crisis as a sovereign debt crisis, even though the latter narrative has dominated media coverage and political perceptions. The fragility of sovereigns in the euro area was to a great extent (though not in Greece) the result of large implicit and explicit state guarantees to national banking sectors.
The International Monetary Fund made a significant contribution to addressing the euro area’s financial sector challenges. The euro-area crisis exposed the unsustainability of the then-existing European Union banking policy framework. National authorities were ineffective in supervising banks adequately in the run-up to the crisis, and did not manage and resolve financial sector aspects of that crisis in an effective and timely manner. EU institutions, including the European Central Bank (ECB), did not generally have the skills, experience or mandate that would have enabled them to offset the national authorities’ shortcomings. The IMF was thus in a position to make a major positive difference.
At the euro-area level, the IMF played a ground-breaking role in understanding the dynamics of the crisis and promoting banking union as an essential policy response. The IMF was the first public authority, and one of the first movers more generally, to acknowledge the role of the bank-sovereign vicious circle as the central driver of contagion in the euro area. It was also the first public authority to articulate a clear vision of banking union as a policy response, building on its longstanding and pioneering support for banking policy integration in the EU. Even though its advocacy on these issues was not entirely consistent and continuous, the IMF can claim some credit for the euro area’s breakthrough decision in mid-2012 to initiate the banking union; this decision was the most important turning point in the entire sequence of crises.
In individual countries, the IMF’s approach to the financial sector was appropriate and successful in several, but not all, cases. The Stand-By Arrangement (SBA)-supported programme in Greece preserved short-term financial stability, but many of its financial sector aspects are difficult to assess on a stand-alone basis since it was followed by further IMF assistance (which falls outside the scope of this evaluation). With the Extended Fund Facility (EFF)-supported programme in Ireland, the IMF contributed significantly to the effective resolution of a major banking crisis in that country, and so did the Financial Sector Assessment Programme (FSAP) and subsequent IMF technical assistance in Spain. But the opportunity to clean up the financial sector was missed in the EFF-supported programme for Portugal.
The IMF should preserve, update and develop the institutional knowledge it has acquired about the EU financial sector framework. The Fund’s positive contribution to addressing financial sector aspects of the euro area crisis is widely acknowledged by European policymakers, providing a promising basis for future engagement. The IMF should devote particular effort to adapting its processes and methodologies to the new context of banking union, which, for the euro-area countries, makes it increasingly difficult to consider financial sector issues on a national basis.
Financial sector issues, especially those relating to banks, played a central and generally under-recognised role in the euro-area crisis. Poorly controlled risk-taking by European banks throughout the 2000s left the EU banking sector highly vulnerable at the onset of the financial crisis in mid-2007, and the subsequent shocks of 2007-08 left it in a situation of systemic fragility. Unlike in the United States, this fragility was not addressed head-on and was allowed to linger. Europe’s banking problem thus long predated the emergence of the sovereign debt sustainability challenges starting at the end of 2009 (see, for example, Posen and Véron, 2009; Rehn, 2016).
The key mechanism of the euro-area crisis was what has become widely referred to as the bank-sovereign vicious circle. The modalities of this mechanism varied in different countries but its ‘doom loop’ pattern became increasingly visible as the crisis worsened, even though it was initially obscured by the unique features of the Greek situation. The bank-sovereign link is deeply embedded in the political economy of each member state. It covers the use of banks by governments as instruments of national policy, including the preferred financing of favoured sectors and of the state itself (‘financial repression’) and, conversely, the protection and promotion of domestic banks by national governments (‘banking nationalism’). Europe’s bank-sovereign links were made explicit by a joint commitment given by EU leaders in mid-October 2008 to provide national funding, capital and guarantees to their respective banking systems so that no bank would be allowed to fail (Council of the European Union, 2008).
The bank-sovereign link exists in all jurisdictions, but it became uniquely destabilising, and thus a vicious circle, in the context of the EU’s single market and single currency. From the 1990s onwards, cross-border market integration and a competitive level playing field were enforced through increasingly powerful European policy frameworks, including regulatory harmonisation and EU competition policy. Banking policy frameworks covering supervision and crisis management, however, remained almost entirely national until well into the crisis, despite reforms such as the so-called Lamfalussy Process of EU-level regulatory and supervisory cooperation, introduced in the early 2000s, and the creation of three European supervisory authorities in January 2011 following the Larosière Report of February 2009. This mismatch created perverse incentives for national authorities to neglect prudential aims for the sake of banking nationalism, which prevented banking supervision from being sufficiently effective in almost all advanced EU member states (Véron, 2013). In the euro area, the problem was compounded by the impossibility of devaluing in the event of a sudden stop.
The set of reforms known as banking union provided a fundamental response to this policy challenge, even though it came late and remains incomplete. Initiated at a euro-area summit on 28-29 June 2012, the banking union policy package aims explicitly to break the bank-sovereign vicious circle through a transfer of most instruments of banking sector policy in the euro area from the national to the European level. Its inception was instrumental in enabling the ECB to announce its Outright Monetary Transactions (OMT) programme, which in turn marked the turning point of the crisis and the start of a broad normalisation of sovereign credit conditions. As described below, however, banking union remains incomplete and will require new policy initiatives if it is to achieve its stated objective of breaking the bank-sovereign vicious circle in the euro area.
Banking union consists of three pillars, under a standard though somewhat simplified classification. These are: (1) a Single Supervisory Mechanism (SSM) that establishes the ECB as the central supervisor of euro-area banks; (2) a Single Resolution Mechanism (SRM) that establishes a new framework for bank crisis management and resolution (though less centralised than the SSM), with a new agency, the Single Resolution Board (SRB), as the hub for corresponding decision making; and (3) a European Deposit Insurance Scheme (EDIS), designed to eventually mutualise the resources and mechanisms through which euro-area countries protect guaranteed deposits.
Even though the three pillars are mutually dependent, banking union is being implemented in a staggered and protracted sequence. The SSM was announced in late June 2012 and has been in force since 4 November 2014. The SRM was announced in December 2012 and has been in force since 1 January 2016, but its financial arm, the Single Resolution Fund (SRF, managed by the SRB) will only reach its steady state in 2024, and even then might lack an effective fiscal backstop. A proposal for the European Deposit Insurance Scheme, published by the European Commission in November 2015, would see the EDIS reach a steady state in 2024 at the same time as the SRF but, at the time of writing, no decision has been made to implement this scheme.
Given the essential importance of the bank-sovereign vicious circle and of banking union in the euro-area crisis, section 2 of this paper is devoted to the IMF’s role in identifying and addressing these topics. It first reviews the IMF’s surveillance of the European financial system since the start of the crisis, with a focus on the role that the Fund played in the gradual identification of the bank-sovereign vicious circle and its acknowledgement by European policymakers in the period 2010-12. It then analyses the sequence of IMF contributions to the elaboration of the policies now known as banking union. The rest of the paper focuses on individual countries. Section 3 discusses the financial-sector aspects of the assistance programmes for Greece (Stand-By Arrangement 2010-12), Ireland (Extended Arrangement 2010-13) and Portugal (Extended Arrangement 2011-14), as well as the IMF’s involvement in Spain (2012-14). Section 4 analyses salient selected themes from the reviewed cases. Section 5 concludes.
While each IMF programme was country-specific, their contents were partly determined by the shared EU and euro-area policy framework. The IMF acted in coordination with the European Commission and the ECB, forming a ‘troika’ with these two institutions (Kincaid, 2016). The European Commission was mainly represented in troika discussions by its Directorate-General for Economic and Financial Affairs (DG ECFIN). On financial sector aspects, however, the Commission’s Directorate-General for Competition (DG COMP) played an influential and autonomous role as the enforcer of the European Union’s unique policy framework for state aid control, with oversight of any national publicly-funded interventions in the banking sector. Another significant EU-level policy framework governs central bank lending operations to banks. The ECB transacts with financial institutions under its monetary policy mandate. In addition, national central banks of the Eurosystem play a lender-of-last-resort role when providing emergency liquidity assistance (ELA), which is supplied under the control of the ECB to ensure compatibility with monetary policy. These central bank interventions powerfully interacted with sovereign financing, through the banks’ purchases of sovereign securities and other mechanisms.
2 Euro-Area-Level Aspects
A. Financial system analysis
Pre-crisis surveillance by the IMF largely missed the build-up of risk in the euro-area banking system. Successive Article IV reports on euro-area policies aptly characterised the shortcomings of cross-border financial integration and corresponding policy challenges (as discussed below), but otherwise devoted only partial attention to financial system developments. They focused on a limited set of indicators, such as banks’ profitability, share prices, market indicators of distance-to-default and reported capital ratios, which did not adequately capture the accumulation of risks in banks’ balance sheets. This observation echoes the identification of shortcomings in general evaluations of the IMF’s surveillance in the run-up to the financial and economic crisis (IEO, 2011; Pisani-Ferry, Sapir, and Wolff, 2011).
The IMF’s underwhelming performance in this respect was comparable to that of most other observers, including most market participants. They similarly failed to identify the overextension of banks’ balance sheets that ultimately provided the basis for Europe’s banking crisis starting in late July 2007. The IMF correctly analysed just before that date that “financial indicators may have peaked [and] there are signs that the credit cycle is gradually turning,” but it also stated that “the [euro area] financial system is viewed as relatively healthy” and that analyses by the European Commission and ECB “had confirmed the robustness of the financial system” – statements that would be shown to be questionable shortly afterwards (IMF, 2007: 17-18).
After the crisis began in July/August 2007, the IMF took a long time to adjust its assessment of the soundness of the euro-area banking system. In 2008, the Euro Area Policies Article IV Staff Report observed that “The euro area’s financial system entered the turmoil from a position of strength” and that “the area’s financial system remains sound” (IMF, 2008a: 3, 7), and correspondingly missed the imminent financial panic and subsequent recession. This was in spite of widespread concerns expressed by market participants and other observers at the time (for example Borio, 2008; Véron, 2008), and it appears that the IMF took national and European authorities’ reassurances too much at face value. At an IMF Executive Board meeting in July 2008, staff made sanguine statements in response to questions and comments from executive directors:
“With respect to [bank] balance sheets, the situation is generally stronger in the euro area than elsewhere, with some exceptions within the area. (…) It is true that European banks are more highly leveraged than U.S. banks, but the former hold relatively less risky assets than the latter. (…) Overall, the staff is fairly confident that the regulatory and legal and accounting frameworks currently in place – Basel II and the International Financial Reporting Standards – can ensure there will be adequate recognition of bank losses in the euro area, even if this may be somewhat less timely than in the United States”.
By contrast, in the months following the panic of late September and early October 2008, the IMF was ground-breaking in highlighting European banks’ unaddressed vulnerabilities. The April 2009 Global Financial Stability Report (GFSR) was a landmark contribution that shed an unflattering light on European banks’ unacknowledged losses, and contrasted them with the more timely disclosures in the US and the far lower exposures in Japan (IMF, 2009a: Table 1.3). This analysis was updated in the three subsequent GFSRs of October 2009, April 2010 and October 2010. Though it attracted a lot of attention from outside analysts, and also considerable pushback and criticism from European country authorities and the ECB, as well as internal debate within the IMF, it was comprehensively vindicated by later developments. The banks’ vulnerabilities in terms of undercapitalisation, funding, asset quality and sovereign risk exposures were appropriately identified and characterised. Correspondingly, the 2009 Article IV Staff Report for the euro area frontloaded its analysis of the financial sector, in contrast to the practice in previous years, noting that “the financial sector remains key to the shape and the robustness of the economic recovery.” That report aptly emphasised the “need to take further decisive action, especially in the financial sector. (…) A resolute and coordinated cleanup of the banking system is essential to restore trust” (IMF, 2009d: 4, 9).
As a consequence, the IMF appropriately pressed for aggressive bank stress testing and recapitalisation. The IMF publicly criticised the lack of disclosure of results of the first round of EU stress tests in the late summer of 2009. Subsequently, the IMF correctly emphasised that the stress-testing rounds of mid-2010 and mid-2011 would not be sufficient to restore trust in the European banking system unless they were followed up with appropriate action (IMF, 2010c: 14; IMF, 2011b: 13-15).
In 2009, staff in the IMF European Department were first to identify the bank–sovereign vicious circle in the euro area. Channels of contagion, from sovereigns to banks and to a lesser extent from banks to sovereigns, had been described in pre-crisis literature, but less so the mutual reinforcement between them in the context of a supranational financial system that was integrated by binding policy instruments. Following the Icelandic crisis of late 2008, research on bank-sovereign linkages was jointly undertaken by staff at the Fund’s European and Research departments. The European Regional Economic Outlook report of May 2009 included a chapter on the fiscal risks resulting from “the use of public balance sheets to shore up the financial system” (IMF, 2009c: Chapter 2). At the same time, Mody (2009) exposed “the possibility that sovereign spreads, the health of the financial sector, and growth prospects support a mutually reinforcing [bad] equilibrium”. In the literature analysis conducted for this evaluation, the author found no other authors or organisations that identified the bank-sovereign vicious circle with similar clarity in 2009, let alone earlier.
In 2010, the bank–sovereign vicious circle was further characterised by a few analysts, but it did not feature widely in either IMF analysis or the public debate. For example, Candelon and Palm (2010) noted that while “the consequences of fiscal imbalances for currency/banking crises has [sic] been largely investigated (…) [o]n the contrary, only few papers have scrutinised the potential mutation of banking crises into sovereign debt ones. (…) Reinhart and Rogoff even portrayed this lack of empirical studies regarding banking and debt crises as ‘a forgotten story’”. Illustrating this point, the April 2010 GFSR included a chart showing contagion from sovereigns to financial systems, as had just happened in Greece, but not the other way around. In an update of the same analysis in the October 2010 GFSR, dotted lines were added from banks to sovereigns, and “linkages to the banking system” were mentioned as among several factors contributing to elevated sovereign risks, but the emphasis remained on the sovereign-to-bank channel – underplaying an essential part of the dynamic nature of the contagion (IMF, 2010a: Figure 1.5; IMF, 2010d: Figure 1.5 and text p. 4). The fact that the IMF was slow to build on the early insights from its staff in 2009 about the bank-sovereign vicious circle may partly be attributed to its increased focus on adjustment programmes, starting in 2010 – including work on the Greek SBA, which diverted resources from euro-area-wide analytical work.
In 2011, the bank-sovereign vicious circle narrative became widely recognised. Particularly following the developments in Ireland in November 2010, this narrative became increasingly prevalent in academic and other independent studies of the euro-area crisis throughout 2011. But it took some additional time for the IMF to fully realise the implications. The widely noted speech by the managing director in Jackson Hole in August 2011 (Lagarde, 2011), which emphasised the “urgent” need for recapitalisation of European banks, illustrated the lingering ambiguities of the IMF’s stance at that date. On the one hand, the MD made the pioneering proposal “to mobilise EFSF [European Financial Stabilisation Facility] or other European-wide funding to recapitalise banks directly,” which was fully aligned with the vicious circle analysis (see Section IIB). On the other hand, she presented this proposal as only “one option” and thus implied that massive recapitalisations might be funded by national budgets instead – which would inevitably have exacerbated the vicious circle.
The bank-sovereign vicious circle became a major feature of the IMF’s interpretation of the euro-area crisis in the autumn of 2011, well ahead of European authorities’ interpretations. In July, the Article IV Staff Report on euro-area policies stated that “the approach to banking problems remains national, thus perpetuating the intertwining of banks and sovereigns” (IMF, 2011b: ‘Key Issues’). A simple and coherent IMF description of the vicious circle was developed by the IMF’s Research Department in the late summer of 2011, and from then on the bank-sovereign vicious circle was consistently and publicly exposed by IMF senior staff and management (see, for example, Chopra, 2011; Lagarde, 2012). The EU did not adopt this analytical paradigm until the spring or early summer of 2012.
In sum, the IMF was wrong footed by the financial crisis starting in mid-2007, but from early 2009 it was mostly ahead of other public institutions in the identification of the unique dynamics of the financial system crisis in the euro area. The Fund’s identification of these dynamics played an important role in the genesis of Europe’s banking union.
B. Institutional architecture and banking union
The IMF’s European Department identified and highlighted inadequacies of the European Union’s bank policy framework at an early stage, well before the start of the financial crisis. For example, Article IV consultations for the euro area in 2005 led to a characterisation of barriers to cross-border financial integration in the EU and of the need for more centralised banking supervision (IMF, 2005: Chapters IV to VI). Resulting financial stability concerns were expressed in Article IV staff reports for 2005, 2006 and 2007. Several publications that were initiated by the European Department in 2007, combined with outreach initiatives, provided important analytical contributions to the debate on EU banking policy architecture and went into some detail in outlining possible policy responses.
Differences of view between IMF departments, however, prevented this pioneering analysis from influencing policy to the extent it could have. The suggestion to pool supervisory responsibility at the supranational level was fiercely opposed by EU member state authorities. This made it impossible for the European Commission and ECB to publicly articulate proposals such as the IMF’s at that time. The European national authorities’ concerns were echoed in IMF internal debates, principally by the Monetary and Capital Markets (MCM) Department. As a result, public expression of the European Department staff recommendations only used coded, euphemistic terms. The notion of a pan-European banking supervisory authority was referred to as “joint responsibility and accountability,” and the option of a European resolution and/or deposit guarantee fund was referred to as “a burden-sharing agreement” or “ex ante mechanisms to share costs of [bank] failures” (IMF 2007: 20 and 27). These watered-down formulations reduced the impact of the IMF’s otherwise novel analysis. Even so, the European Department’s work on the EU banking policy framework proved controversial when presented to the Executive Board in July 2007. According to the minutes of the meeting, several European Directors expressed the views that the tone of staff recommendations was “too alarmist;” that there was too much emphasis on “the misalignment of incentives” of national prudential authorities; and that financial supervision, being an EU matter, “should not be a focus of discussion on euro area policies”. In 2008, staff presented a toned-down version of a ‘European mandate’ for financial sector authorities in the EU, implying no change to the institutional architecture (IMF, 2008b: Chapter IV).
Following the post-Lehman panic, IMF staff made ground-breaking policy proposals to address the challenge of bank restructuring and resolution, which were endorsed by IMF management in early 2010. The shift of focus from supervision to bank restructuring was justified by the fact that in the meantime, EU policymakers were making progress with the establishment of the European Banking Authority (EBA), which was agreed in 2009 and became effective in January 2011. In 2009, the staff complemented the IMF’s advocacy in favour of “a resolute and coordinated cleanup of the [euro area] banking system” with well-argued proposals to create special resolution regimes for banks in all EU member states, building on established US practice and the more recent UK Banking Act 2009, and prefiguring the EU Bank Recovery and Resolution Directive (BRRD) (IMF, 2009e: Chapter II; Cihak and Nier, 2009). In early 2010, staff from the European (corresponding author), MCM and Legal Departments went further by suggesting a blueprint for the supranational pooling of corresponding powers in a European Resolution Authority (ERA), with accompanying legal and financial arrangements to ensure effectiveness (Fonteyne and others, 2010). The ERA proposal was forcefully endorsed by the managing director in a public address in Brussels (Strauss-Kahn, 2010) and referred to in the euro-area Article IV Consultations in June. It can be seen as a prefiguration of the SRM.
Until late 2011, these proposals referred to EU-level arrangements, implying that the UK and other non-euro-area member states would be included. This stance was consistent with the EU legal framework, which did not contain a separate financial regulatory regime for the euro area and considered financial regulation as a component of EU Internal Market legislation, which applies to all member states. But it also heightened the political obstacles to implementing the proposals, since the UK (together with Sweden and several non-euro central and eastern European countries) was even more opposed to a pooling of banking policy sovereignty than were most euro-area member states, and was also more advanced in tackling systemic fragility in its own domestic banking sector. Developments to overcome this logjam were protracted. An important change was the acknowledgement by the UK government in mid-2011 that the “remorseless logic” of monetary union would justify further institutional build-up in the euro area. Nevertheless, the Fund’s Article IV Staff Report for the euro area in 2011 still advocated the ERA proposal in the context of “the EU financial stability framework,” with no mention of specific arrangements for the euro area (IMF, 2011b: 16-17).
The IMF eventually acknowledged the political obstacles to a pan-EU banking union, and focused its advocacy of an integrated banking policy framework on the euro area. This shift was consistent with the contemporary recognition of the euro area’s bank-sovereign vicious circle (analysed above). An early step was the managing director’s above-mentioned suggestion of direct bank recapitalisation by the EFSF in her Jackson Hole speech of late August 2011. The next development was the suggestion, in a speech by the managing director in Berlin in January 2012, that direct bank recapitalisation, supervisory integration and what had earlier been the ERA proposal should be combined into a single policy package for the euro area (Lagarde, 2012). This text appears to be the first time that a public authority set out the vision of banking union as it was eventually endorsed by European leaders later in 2012:
“To break the feedback loop between sovereigns and banks, we need more risk sharing across borders in the banking system. In the near term, a pan-euro area facility that has the capacity to take direct stakes in banks will help break this link. Looking further ahead, monetary union needs to be supported by financial integration in the form of unified supervision, a single bank resolution authority with a common backstop, and a single deposit insurance fund.”
This vision was further developed in the Article IV Consultation later in 2012 (IMF, 2012g: Chapter I). The shift of geographical scope from EU to euro area was frustrating from a principled and legal perspective, since banking union is essentially a single market initiative and should thus encompass the entire EU, but it was pragmatic, given the urgency of the euro area situation in the first half of 2012 as well as political realities in the UK and other non-euro-area member states. The IMF subsequently rationalised this shift by stating that “while a banking union is desirable at the EU27 level, it is critical for the euro 17”.
The decision to initiate banking union was eventually made by euro-area leaders. The catalyst was the increasingly widespread acknowledgment of the bank-sovereign vicious circle as the main engine of financial deterioration in the euro area, and the parallel recognition – which became increasingly clear in the second half of 2011 – that a fiscal union (enabling unlimited joint issuance of ‘eurobonds’) would not be politically feasible. As a consequence, the direct recapitalisation of banks by the EFSF or its successor the European Stability Mechanism (ESM) was increasingly seen as the most promising policy option to stem the contagion, following its early advocacy by the IMF but also by the EBA and later by several member states. The corresponding debate accelerated in the second quarter of 2012, spurred in particular by developments in the Spanish banking system (see below) and the belief that Spain’s continued access to sovereign debt markets was critical for the entire euro area. The establishment of the SSM was proposed in late June by the German government, initially as a check on the risk-sharing inherent in ESM direct bank recapitalisation. The opportunity was seized and swiftly implemented thanks to the joint leadership of the ECB, the President of the European Council and the European Commission, and the SSM subsequently became the central pillar of the fledgling banking union.
The IMF can claim some of the credit for the initiation of banking union, even though the opinions of European policymakers vary as to the Fund’s precise role. Several current or former senior officials have argued in interviews that the IMF was not among the main participants in the banking union decision-making sequence. Others, however, emphasised that the IMF’s longstanding advocacy, reiterated in various presentations to European ministers and bilateral meetings in the second quarter of 2012, helped prepare the ground for the decision. It also appears that the IMF’s credibility with the German government enabled it to help overcome initial resistance there. Unlike the ECB and other EU institutions, the IMF did not have a direct turf interest in banking union – which made its advocacy more compelling. In an interview for this evaluation, a senior EU official noted that the US Treasury was also influential in advocating banking union from outside the European Union, especially during the Los Cabos summit of the G20 on 18-19 June 2012.
After the breakthrough decision of euro-area leaders on 29 June 2012 to initiate banking union, the IMF continued to provide significant advocacy and advice. It was used as a resource by the ECB in the design and initial establishment of the SSM between 2012 and 2014. The IMF helped cement a consistent framework, including by promoting the use of the term ‘banking union’ to bring together the different policy areas of supervision, resolution and deposit insurance and, more substantially, by publishing a string of documents that holistically described the banking union policy vision (see for example IMF, 2012f; Goyal et al, 2013; the first EU FSAP, published in March 2013; and Enoch et al, 2013). Through these contributions, the IMF appropriately kept reminding euro-area policymakers that the bank-sovereign vicious circle would not be fully broken, and that banking union would remain fundamentally incomplete, as long as deposit insurance remained at the national level and in the absence of adequate backstops. This helped pave the way for the European Commission’s EDIS proposal in November 2015.
In sum, banking union is a case of influential and successful IMF policy advocacy, even though the IMF was far from alone in proposing banking union, many aspects of which were first formulated by academic and other independent experts. The IMF itself was also less than fully consistent over time and across departments. It can certainly not claim exclusive ownership of originating (let alone deciding) the set of reforms referred to as banking union, which itself still remains incomplete and a work in progress. Nevertheless, its role in the planning and implementation of Europe’s banking union should be recognised as constructive and significant.
3 Individual Countries
A. Greece (SBA 2010-12)
The Greek crisis was not triggered by developments in the Greek financial sector. In the late 2000s, the Greek banks were considered well capitalised and did not raise major concerns. The Greek banking sector was widely seen as conservative and resilient, with the last systemic banking crisis having taken place in 1932. The ratio of private debt to GDP was low (97 percent as of end-2008), similar to that in Italy (105 percent) and notably lower than in Portugal, Spain and Ireland (all above 170 percent). Several Greek banks had expanded internationally, mostly to south-eastern Europe and in one case (National Bank of Greece) to Turkey, but oversight of them was considered tight and broadly adequate. Greece’s Article IV Staff Report for 2009 mentions that “stress tests (conducted jointly by the Bank of Greece and staff) suggest that the banking system has enough buffers to weather the expected downturn” (IMF, 2009f: 16).
As the Greek government lost market access for sovereign debt in late 2009, Greek banks lost market access for wholesale funding. Liquidity provision was provided by the ECB, from a level of around €10 billion by end-2008 to €50 billion by end-2009 and €100 billion by end-2010 (IMF, 2013c: 19). Deposits were much more stable than wholesale funding, but nevertheless declined regularly and significantly from their peak level in the second half of 2009.
The IMF was involved early on in providing technical assistance to the Greek authorities on financial sector issues. An IMF technical assistance mission in February/March 2010, at the request of the Bank of Greece, helped on the management of ELA. A few weeks later, IMF technical assistance was instrumental in the design of a mechanism by which banks would issue bonds guaranteed by the government and thus eligible as ECB collateral, and use them for access to Eurosystem liquidity. These “government-guaranteed bank bonds issued for own use,” or “own-use GGBBs,” would play a major role in subsequent Eurosystem lending operations, in Greece and other programme and non-programme countries.
The main financial sector-related condition of the Greek SBA-supported programme, approved on 9 May 2010, was the creation of the Hellenic Financial Stability Fund (HFSF). Intended as an instrument to provide solvency support to any bank that would be hit by unexpected loan losses, the HFSF had €10 billion at its disposal for capital interventions. This was innovative and significant, since the Greek situation was not perceived as a financial sector crisis. The legislation establishing the HFSF was enacted on 13 July 2010, and the HFSF became operational in the autumn of that year. The only other financial sector-related conditions of the programme were the intensification of supervisory practices by the Bank of Greece and a commitment to review insolvency legislation.
The initial implementation of the financial sector aspects of the programme was comparatively uneventful. In the EU stress-testing round for which the results were announced in July 2010, all Greek banks were found adequately capitalised except the Agricultural Bank of Greece (also known as ATEbank). Majority owned by the government, the Agricultural Bank was recapitalised with government funds in the course of 2011. Frequent reviews of banks’ funding outlook and continued liquidity provision by the Eurosystem allowed the authorities to avoid disorderly developments, despite declining deposits and regularly mounting non-performing loans. Precautionary capital buffers were imposed on banks, which complied through various channels, including by disposing of assets and raising equity. Total credit started declining in mid-2010 (IMF, 2012a: 7), contributing to the economic downturn that was being driven by many other factors at the same time.
The financial sector situation deteriorated in the course of 2011. In the first quarters of the programme period it appeared natural not to take into account the potential losses of Greek banks on their portfolios of Greek government bonds (GGBs), but this stance became increasingly hard to justify as discussions about government debt restructuring, referred to as private-sector involvement, developed in 2011. In these circumstances, ECB bank lending operations, which are reserved for ‘sound’ banks, increasingly came into question, with the corresponding uncertainty in turn contributing to tighter credit conditions. While ECB lending was maintained, the Bank of Greece supplemented it with greater use of emergency liquidity assistance under ECB review, at a higher cost for the banks. Attempts were made to foster the purchase of Greek banks by foreign acquirers, but were unsuccessful given the general uncertainty. Two small banks had to be resolved under a newly adopted resolution law: these were Proton Bank (amid allegations of fraud and money-laundering) in October 2011 and T Bank (formerly known as Aspis Bank) in December 2011.
The materialisation of private-sector involvement triggered a major restructuring of the Greek banking system. At no point in the SBA-supported programme did the IMF impose limitations on the Greek banks’ exposure to the Greek sovereign, nor did the other troika institutions or the Bank of Greece; an IMF official interviewed for this evaluation noted that such limits were “not in the IMF’s software at the time” and were not even considered. In late 2011, the nominal (undiscounted) value of Greek banks’ GGB portfolios was about €45 billion, compared to an aggregated core capital of €22 billion (IMF, 2011g: 38, Box 3). In addition, an asset-quality review of Greek banks that was performed by BlackRock Solutions for the Bank of Greece in the second half of 2011 identified further non-performing exposures. Eventually, €50 billion was reserved under the EFF-supported programme approved on 15 March 2012 to cover bank recapitalisation needs and resolution costs.
The governance of Greek banks posed challenges that the IMF did not address decisively. The IMF did not initially (during the SBA-supported programme) focus on issues of connected lending that may have contributed to the banks’ risk profile, and thus neglected lessons from its earlier crisis interventions, for example in Asia in the late 1990s. The HFSF’s governance itself was revised on many occasions, but the IMF was not forcefully involved in overseeing it. There were no easy solutions to this challenge, because the uncertainties affecting the Greek banking sector during the SBA-supported programme made it difficult to mobilise non-conflicted sources of private-sector capital for purposes of bank recapitalisation. The IMF’s priority appears to have been to ring-fence the banks as much as possible from government interference. In a late-2011 programme review, the IMF argued that “the government of Greece has a poor track record of properly managing state-owned banks and managing its own finances. This suggests that an effort needs to be made to keep a part of the core banking system in private hands, run by competent managers” (IMF, 2011g: 38, Box 3). This stance was understandable but it carried the risk of insufficient scrutiny of private-sector management, as well as of distortions between ownership and control that could pose governance challenges, especially as banks became increasingly dependent on public sources not only for liquidity support but also for capital.
A separate issue was the impact of the Greek programme on the rest of the EU banking system, and the possible implications of this impact for decisions on the design and timing of private-sector involvement (PSI). The design of the SBA-supported programme in 2010 had the consequence that banks outside Greece that held significant portfolios of GGBs, prominent among which were several French and German banks, would receive scheduled repayments in full as long as PSI was delayed (Wyplosz and Sgherri, 2016). Several banks announced their intention to maintain exposures, but there was no binding mechanism to enforce a standstill and the incentives of banks and governments were largely misaligned in this respect. The protracted discussion on PSI in 2011-12 implied that eventual losses on such portfolios were significantly lower for these banks than they would have been had a comparable restructuring taken place at an earlier date. It also implied that to achieve a given amount of sovereign debt reduction, a larger haircut would be needed now, compared with the earlier period when the total pool of Greek sovereign debt held in the private sector was larger. The IMF appears not to have had a clear picture of exposures to Greece in the early phase of reflection about a future PSI in the late spring and summer of 2010, and indeed to have struggled to collect reliable data on these (IMF, 2010b: 12, 16).
The extent to which concerns about banks outside Greece weighed on decisions on the timing and design of private-sector involvement is debatable. Some observers have argued that the choice not to have a PSI in May 2010 was largely or even primarily motivated by a desire to protect the French and German banks that held Greek GGBs, which would have been affected by ‘direct’ contagion (ie losses on the Greek debt that would be restructured) (see, for example, Véron, 2010). But there were simultaneous concerns about ‘indirect’ contagion to other euro-area sovereign issuers that were perceived as fragile (such as Ireland, Italy, Portugal and Spain), through a general decrease in confidence in the safety of euro-area government debt if this were shown to carry credit risk. Concerns about indirect contagion, which were partly informed by the experience of the post-Lehman panic, appear to have been the decisive driver of the choices made, at least by the IMF and ECB, and possibly also by the French and German governments. The delay in agreeing on PSI in 2011 can be attributed primarily to the flaws in European decision-making processes and to insufficient focus of attention and decisiveness (compounded by an unexpected management transition) at the IMF, even though it also suited the financial interest of the banks that were receiving reimbursements from Greece in the meantime.
The fact that only the SBA-supported programme is evaluated here makes it difficult to carry out a holistic assessment of the financial-sector aspects of the Fund’s work in Greece. On the positive side, programme conditions related to the financial sector were broadly met, and no major disorderly developments occurred in the Greek financial sector in spite of the highly challenging environment. On the negative side, the lack of sufficient attention paid to issues related to Greek banks’ governance and ownership might have contributed to difficulties at a later date; no effort was made to limit Greek banks’ exposure to the Greek government; and the lack of progress on insolvency reform contributed to the accumulation of unaddressed NPLs. It should be noted, however, that any impact of these shortcomings was dwarfed by other drivers of the Greek sequence, including the lack of national ownership, structural features of the Greek economy and political system, and the reluctance of other EU member states to provide unambiguous support to Greece as a member of the euro area.
B. Ireland (Extended Arrangement 2010-13)
Ireland was a textbook case of the banks-to-sovereign part of the bank-sovereign vicious circle – namely a major banking crisis leading to the sovereign’s loss of market access. It differed from the Greek case, where the contagion was from sovereign to banks. The Irish banks aggressively expanded their balance sheets and risk-taking in the 2000s in an environment of connected lending and inadequate national supervision (Regling and Watson, 2010; CBI, 2010; CIBSI, 2011; IMF, 2015). In response to the post-Lehman turmoil in late September 2008, the government extended a blanket two-year guarantee on all bank liabilities, which was later partly extended into 2011. The government then nationalised Anglo Irish Bank in January 2009; recapitalised the two largest domestic banks, Allied Irish Banks (AIB) and Bank of Ireland (BoI), in February 2009; established the National Asset Management Agency (NAMA) with the aim to purchase impaired loans from the banks, later in 2009; and nationalised the Irish Nationwide Building Society (INBS), in December 2009. The resulting contingent liabilities, combined with the sharp economic downturn, led to market speculation as early as January 2009 that Ireland might need IMF support, and indeed this scenario was discussed by Fund staff with the authorities in early 2009 (Donovan, 2016).
The IMF’s surveillance did not anticipate the Irish crisis. In 2007, the Executive Board “welcomed the indicators confirming the soundness of the Irish banking system, including the stress tests suggesting that cushions are adequate to cover a range of shocks even in the face of large exposures to the property market”. Donovan (2016) mentions, among various shortcomings, the inadequate attention paid to risks linked to the commercial property market. He notes that in 2008 “the Fund essentially absented itself from the Irish stage,” in part because the mission chief was reassigned to work full time on the UK; and the 2008 Article IV consultation never took place. The authorities appear not to have interacted at all with the IMF when they guaranteed bank liabilities in October 2008. The IMF’s engagement with Ireland caught up rapidly in 2009, however.
Irish banks lost wholesale market access in 2010. Considerable losses materialised for the banks as the NAMA purchased property loans at a steep discount to book value, even though their market value was even lower. Corporate clients also increasingly withdrew deposits. ECB liquidity to Irish banks rose rapidly during 2010 and eventually peaked at around €90 billion in early 2011, supplemented by emergency liquidity assistance from the Central Bank of Ireland (CBI) at an additional €60 billion. Moreover, in spite of very large public capital injections into banks (€46 billion in aggregate by mid-2010), the banks’ solvency came increasingly into question. Successive rounds of stress testing by the CBI in 2009 and 2010, dubbed the Prudential Capital Assessment Review (PCAR), failed to establish trust because analysts remained less than convinced by methodological choices and stress assumptions. The same applied to Ireland’s participation in the EU-wide stress-testing exercises of September 2009 and July 2010. Since Eurosystem liquidity is reserved for ‘sound’ and solvent banks, the ECB made it clear to the Irish government in October 2010 that resolute actions would be needed to enable its continued provision. Meanwhile, the aftermath of the French-German declaration in Deauville (16 October 2010), which appeared to encourage PSI as a component of future assistance programmes, led to further deterioration of financing conditions. The Irish authorities requested assistance in early November, and the staff-level agreement on the EFF-supported programme was announced by the IMF on 28 November 2010.
The treatment of senior bank bonds, widely referred to in Ireland as “burning the bondholders,” was a salient issue for the IMF. Former senior Irish officials who were involved in the programme discussions said in interviews conducted for this evaluation that the option of imposing losses (or ‘bail-in’) on senior bank bonds was introduced informally by the IMF mission staff during pre-programme discussions in November 2010, to the considerable frustration of the ECB, and was immediately supported by the Irish authorities. The authorities, however, had concerns that the proposal might not have been fully thought through from a technical and legal standpoint. From interviews conducted for this evaluation, it appears that European Department staff at the time estimated the potential impact of the bail-in at as much as €16-17 billion, as they argued it could be applied to the ‘going-concern’ Bank of Ireland and Allied Irish Banks (also known in Ireland as ‘pillar banks’), as well as to the ‘gone-concern’ Anglo Irish and Irish Nationwide Building Society. By contrast, the Irish authorities believed that bailing-in the bonds of the pillar banks might create broad uncertainty and result in high indirect costs for the Irish economy. Nevertheless, imposing losses on gone-concern banks’ bonds alone, the outstanding amount of which was €4-5 billion, would potentially have yielded €2-3 billion of relief, a not-insignificant amount in the Irish context. The Fund’s MCM Department was more worried about contagion to other euro-area banks.
The issue of “burning the bondholders” was discussed in a G7 conference call on 26 November 2010. The IMF’s managing director participated and presented the case for bail-in. As was widely covered in the Irish press, the ECB’s adamant opposition to this action was supported by several participants in the call including, crucially, the US Treasury Secretary, on the basis that the financial gain for Ireland would not justify the risk of destabilising bank bond markets well beyond Ireland, even assuming that the bail-in would only affect gone-concern banks. This temporarily settled the matter. The staff report for the programme request, issued two days later, nevertheless included a reference to criteria for future decisions on bank bonds which left open the possibility of re-examining the bail-in possibility at a later stage (Chopra, 2015: 8). Meanwhile, the IMF’s pro-bail-in stance, which quickly became publicly known, made the Fund comparatively more popular in Ireland.
The “burning the bondholders” controversy resurfaced in March 2011. The matter was intensely debated in the campaign for the national election of 25 February 2011, even though at that time it was only about the gone-concern banks, namely Anglo Irish Bank and INBS, which were to be wound up. In late March, the new Irish finance minister intended to refer to it in his maiden speech to the Irish Parliament, a draft of which he submitted to the troika institutions. While the IMF was still in favour of bail-in, the ECB again successfully opposed it. As a consequence, only junior (subordinated) debt was bailed-in, for all four banks (AIB, BoI, Anglo Irish and INBS), with corresponding losses for creditors of €5 billion in aggregate (see below). The episode left a sense of unfairness against Ireland that played a big role in later enabling a financial restructuring known as the “promissory notes transaction”. Interviewees from the Irish authorities assessed this transaction as having been eventually much more beneficial to Ireland than the bail-in of gone-concern bank bonds would have been – even though most commentators in public debate and the media, and the Irish public more generally, do not appear to acknowledge the advantageous nature of this trade-off.
The IMF’s decisions in the course of the “burning the bondholders” controversy were broadly appropriate. It was reasonable for the Fund to discuss the option of bailing-in senior bank bonds, given the potential gains for Ireland, and to escalate the discussion once it became clear that Irish authorities were supportive and the ECB was not. Given the potential implications for global financial stability, it was also appropriate that this matter be discussed among principals of the IMF’s key stakeholders. The G7 format for such discussion was certainly less than optimal from an IMF standpoint because it excluded major emerging economies, but no evidence was found that this choice of format came from the IMF. Once G7 principals had rejected senior bank bail-in, the IMF could not include that option in the programme, but the Fund was justified in keeping it open for the future and in supporting it (restricted to gone-concern banks) during the next round of discussions in March 2011. In the end, Ireland benefited from the promissory notes transaction and thus cannot be viewed as having been treated unfairly. One key IMF participant later implied in parliamentary testimony that the IMF was active behind the scenes and contributed to the formation of the promissory notes deal, even though the deal “was primarily in the bailiwick of the ECB” (Chopra, 2015: 10).
The EFF-supported programme largely focused on financial sector measures. Out of the €85 billion programme adopted in late November 2010, a notional €35 billion was reserved for banking sector support, with €10 billion initially earmarked for immediate recapitalisation, and an additional buffer of €25 billion for later interventions (IMF, 2010e: 16). The programme mandated significant recapitalisation to reach a core tier-one capital ratio of 12 percent by February 2011, and a new round of balance-sheet assessment and stress testing, known as the PCAR 2011, to restore trust in the underlying capital assessment.
The PCAR 2011 was a significant and successful milestone. The CBI chose BlackRock Solutions to assist it in this exercise, which was conducted at a time of political change (elections on 25 February; new government on 9 March) and for which results were published on 31 March 2011. The asset-quality review (or loan-loss forecasting exercise), performed by BlackRock, was the first such independent system-wide assessment in the euro area. As such, it constituted a notable IMF contribution to European crisis-management strategies, prefiguring similar exercises in other programme countries and the euro-area-wide asset-quality review under the ECB-led Comprehensive Assessment of 2014, which was itself a key milestone in the transition to banking union. The results of the PCAR 2011 formed the basis for later supervisory action in Ireland and, unlike the CBI’s PCAR rounds in 2009 and 2010, did not require significant subsequent revision. The exercise identified a capital need of €24 billion, of which about €17 billion was injected by the government, €5 billion came from the bail-in of junior debt and €1.6 billion from private investors (into BoI) (IMF, 2011d: 12).
The banking sector was further restructured. EBS Building Society (formerly the Education Building Society) was merged into AIB in July 2011; Anglo Irish and INBS were merged in July 2011 to form the Irish Bank Resolution Corporation (IBRC), which was later gradually wound up and liquidated in February 2013; the Irish Life & Permanent group was purchased by the government in 2012 and split, with its insurance operations (Irish Life) resold to a Canadian insurer in mid-2013 and its banking operations (Permanent Trustee Savings Bank or PTSB) restructured in 2012; the credit unions sector was also overhauled, with a new Credit Union Bill adopted in September 2012 and a number of local credit unions closed or recapitalised. Rather than renewing the whole PCAR approach every year as initially envisaged, the CBI performed a so-called “point-in-time balance sheet assessment” as of end-June 2013 in the last quarter of that year, in anticipation of the ECB-led Comprehensive Assessment of 2014.
Controversy arose about the fate of Irish Life & Permanent (later PTSB). The ECB advocated its merger into the Bank of Ireland, an option that would have allowed faster reimbursement of the Eurosystem liquidity on which it kept relying. But DG COMP insisted that keeping alive a third significant bank (aside from AIB and BoI) was needed to ensure a sound financial sector structure, and that saddling BoI with IL&P/PTSB would hamper its recovery. The IMF initially sided with the ECB, apparently assuming that the Irish banking sector would remain contestable and that a temporary duopoly would thus not be overly harmful. The protracted weakness of the Irish banking sector since then suggests that the Fund may have overestimated the short-to-medium-term potential for foreign-bank entry. Eventually the stance of DG COMP prevailed. Overall it appears that there was no first-best option on PTSB, and in this context the sequence of decision making appears to have been reasonable. Former participants from the IMF, ECB and DG COMP stated in interviews that the discussion remained focused on policy substance throughout.
The banks’ deleveraging weighed negatively on Irish growth, but this was mitigated by the large amounts of foreign assets held by the banks, which could be sold without contributing to domestic deleveraging. It was also helpful that a key engine of the Irish economy –operations of large foreign (including many American) multinationals – was not primarily dependent on the domestic banking system. Nevertheless, the credit reduction that occurred in Ireland was significant.
The IMF’s technical input was valued by the Irish authorities. The IMF helped the Central Bank of Ireland to enforce better implementation of International Financial Reporting Standards, especially in loan provisioning and disclosure practices, and to enhance supervisory quality through the assessment of compliance with the Basel Committee’s core principles for effective supervision. A difficult point was the treatment of NPLs and the recovery of mortgage arrears, but though the outcome was frustrating – given delays in corresponding actions by banks – the IMF’s contribution was considered helpful by several stakeholders. Interviewees also acknowledged that the IMF’s assistance on credit union reform was significant.
In sum, the EFF-supported programme for Ireland can be considered a clear success in terms of financial sector restructuring and reform, which was itself a major component of the programme. While the Irish authorities’ ownership was the major driver of the programme’s success, the quality of the IMF staff team and the value of the IMF’s contribution to programme policies were widely praised. The IMF’s interaction with its troika partners, while not always consensual, was also considered highly professional and constructive by all parties.
C. Portugal (Extended Arrangement 2011-14)
In Portugal, problems in the financial sector were less central to the need for assistance than in Ireland, but more so than in Greece. As in Ireland, private debt was very high, in excess of 260 percent of GDP (IMF, 2011a: 4). As in Greece, there were significant fiscal and structural challenges unrelated to the financial sector, including hidden public debt in state-owned enterprises (SOEs) and public-private partnerships (PPPs).