The Eurosystem, headed by the European Central Bank (ECB), provides liquidity to banks, and through them to the entire economy, through five channels. One, the Marginal Lending Facility, has always been small and has practically disappeared since 2008, when the ECB greatly increased the size of its operations. Two of the other four channels, refinancing operations (Main Refinancing Operations, Longer Term Refinancing Operations, Targeted Longer Term Refinancing Operations) and various forms of quantitative easing (including the Expanded Asset Purchase Programme and the Pandemic Emergency Purchase Programme), are decided by the ECB and determine the stance of monetary policy. The ECB tightly controls the remaining two, though they are implemented by national central banks (the investments of national central banks and Emergency Liquidity Assistance) and have no significant macroeconomic impact at euro-area level.
Emergency Liquidity Assistance (ELA) has attracted less attention than the other channels. This is consistent with its insignificant macroeconomic impact on the euro area, but contrasts with its importance for banking developments and for the fate of some euro-area countries that have seen large ELA flows relative to the size of their economies. In this post we want to shed some light on this neglected child of the Eurosystem family.
Let’s start with the name, in which the word ‘emergency’ is prominent. This immediately suggests a hospital emergency room. When an injured person is brought to hospital, the hope is that the injury can be treated and the patient’s normal life can resume. However, a fatal outcome cannot be excluded. The likelihood of a fatal outcome will depend on the gravity of the injuries as well as the treatment received in the emergency room.
There is an analogous situation when banks receive Emergency Liquidity Assistance from a Eurosystem national central bank. We are interested in the percentage of banks that recover after treatment in the Eurosystem’s emergency room, and in the role played in determining their fate by their condition before ELA is applied.
Answering these questions is not easy. The ECB does not publish information on ELA operations and national central banks (NCBs) exhibit variable degrees of transparency (for instance the Central Bank of Cyprus has been very open, while the Banca d’Italia and the Banco de Portugal have been quite opaque). Furthermore, NCBs may change over time the information they publish. While there are good reasons for non-communication about ELA operations when they are being carried out, no justification seems to exist for not communicating historical information about ELA, after an adequate lapse of time. We carried out a painstaking search for evidence about ELA, but think it unlikely we identified all cases because historical information is so haphazard (see Annex 2). The Eurosystem, which often boasts of its transparency, should do better in this area.
What we do know about ELA
A brief examination of European Central Bank documentation shows that ELA rules have become more detailed over time: the 2013 rule consists of two pages while the 2017 and 2020 versions both cover seven pages. The basics, and arguably the procedure prevailing before the codification of ELA in 2013, are, however, the same:
- All responsibilities, costs and risks are borne by the relevant NCB;
- The ECB requires increasingly detailed information from the NCB that grants ELA and from the bank that receives it, also to assess the requirement that the bank should be only illiquid and not insolvent;
- Two thresholds are established for ELA, the first at €500 million and the second at €2 billion. When the first threshold is reached, the ECB Executive Board must be informed before granting ELA. When the second threshold is reached, the ECB Governing Council must be informed;
- When the €2 billion threshold is reached, the Governing Council can decide to object, and in fact prohibit, the provision of ELA if it judges that it would interfere with the single monetary policy of the Eurosystem. This is also the case if the provision of ELA is not compliant with the prohibition of monetary financing, which would be the situation if the bank were insolvent;
- ELA should be granted at a punitive cost, at least 1% above the rate on the Marginal Lending Facility, ie the ceiling of the interest rate corridor;
- The NCB granting ELA can communicate about it, only if ELA is not provided to a single bank, subject to ECB non objection.
While legally, NCBs are responsible for granting ELA, economically, when the amount is large, the fact that the entire Eurosystem stands behind the granting NCB becomes critical. For example, ELA granted to Greek banks during the crisis exceeded €100 billion, which was only possible because the Bank of Greece was part of the Eurosystem. A similar consideration applied to Ireland and Cyprus in some years.
The disconnect between the practical and the legal situation is arguably the reason why former ECB President, Mario Draghi, said at a March 2018 press conference that ELA should be centralised at the ECB and be decided on by the Governing Council. This, however, would require legal changes and thus could not be implemented immediately. Three years after that statement, nothing seems to have moved and the curious situation remains that the ECB is responsible for both monetary policy and banking supervision, but ELA legally remains a prerogative of NCBs.
Table 1 shows our estimate of the total amount of ELA granted by Eurosystem central banks (see also Annex 1).
Our estimate is clearly downward biased. For instance, we found that some ELA was granted to Italian banks, but we could not find evidence of the amounts in the publications of the Banca d’Italia. Even if the figures are incomplete, it is a fair conclusion that the amounts were never really macroeconomically significant at euro-area level, and were small as a share of ECB monetary policy operations, including temporary operations and various forms of outright purchases (QE) (Table 1, last row). From this point of view, ECB Governing Council non objections to the provision of ELA by NCBs were fully justified. ELA amounts, however, were very significant for both the economy and the banking system, in some years, for the small economies of Ireland, Greece and Cyprus.
Dead or alive?
Let’s now come to the original question: what percentage of banks exited ELA alive and what percentage died? Note, we use a broad definition of post-ELA ‘death’ of a bank, identifying it as the end of the institution as an ongoing, independent concern. We thus include cases in which a bank was resolved, was liquidated through normal insolvency proceedings or was bought by another bank.
Answering this question requires forensic work, including close examination of NCB publications, documents issued by banks, press reports, parliamentary enquiries and so on (see Annexes 1 and 2). For the 2008-2019 period, we identified 36 banks in 11 countries as receiving ELA, of which 26 did not survive (Table 2; as noted above, in all likelihood the number of banks that in reality received ELA is higher). Annex 1 details uncertain cases and whether or not we have included them in our sample. While we have identified all ELA cases in Belgium, Cyprus, Germany, Ireland, Slovenia and Spain, we are uncertain about some others. In addition, it could be that we identified more cases of ELA that ended with the disappearance of the bank than cases in which ELA ended with the rehabilitation of the bank. However, for Belgium, Cyprus, Germany, Ireland, Slovenia and Spain we are confident we have found all relevant ELA cases and 10 banks out of the 12 which received ELA in these countries did not continue as independent ongoing concerns. Overall, our reading of the evidence is that our fundamental conclusion should not change significantly.
The fact that a bank in receipt of ELA is twice as likely to die than to be rehabilitated (Table 2) does not mean that ELA is useless or even damaging for banks that receive it. It could just mean that when banks must have recourse to ELA they are, overall, in very poor conditions, which cannot be cured by ELA injections. But then, this raises doubts over the requirement that, according to the rules, only solvent banks should receive ELA: if they were all solvent, how come ELA only rehabilitated a third of them? And if a majority were not solvent, there must have been multiple cases of non-compliance with the prohibition of monetary financing.
An additional item of evidence is that ELA was extended to individual banks for periods extending from one week to eight years, with an average of two and a half years (Annex 1), showing that the term ‘emergency’ has been stretched to cover all cases in which ELA was used. It seems hard to characterise an operation lasting two and half years on average and up to eight years as an ‘emergency’.
Confirming the hypothesis that non-solvent banks have also received ELA requires disentangling their initial conditions from the effectiveness of the treatment they received, by assessing their financial health conditions when they entered ELA. Table 3 shows relevant rating information.
Unsurprisingly, all banks that received ELA were not in good shape as far as their rating was concerned: hardly any was investment grade in the year before receiving ELA. What is more interesting is that, on average, those that were eventually resolved, rated B on average, were in a worse condition than those that survived, rated BB. Thus, there is some evidence that the condition of banks entering the ELA emergency room influenced whether or not they would survive.
Drawing conclusions is not straightforward because, notwithstanding our efforts, the information on ELA is probably not exhaustive and, more importantly, because in the 12 years covered by our observations, the institutional set-up has improved significantly, especially with the creation of the Single Supervisory Mechanism and the Single Resolution Board, and the adoption of the Bank Recovery and Resolution Directive. One can argue that, in many cases before these improvements, ELA was used for lack of a better option to preserve financial stability, even at the cost of bending the rules. Still, we put forward four conclusions:
- While further work is needed to see more precisely what condition banks that accessed ELA were in, in order to disentangle the impact in determining their fate, relative to the quality of the received treatment, the available evidence that death is twice as likely than recovery is strong enough to wonder whether recourse to ELA is really an effective cure.
- The term ‘emergency’ does not cover all cases of ELA; an emergency cannot last on average two and a half year and in some cases eight years.
- It is unlikely that all banks that received ELA were really solvent, as required by the rules. The fact that they were all still licensed and received ELA is, to say the least, surprising.
- ELA amounts were not significant for aggregate monetary conditions in the euro area, but were of critical importance for some countries in some years.
Annex 1 - List of ELA banks
Annex 2 - ELA related disclosures and publications (2008-2019)
Cadamuro, L. and F. Papadia (2021) ‘Emergency Liquidity Assistance: A new lease of life or kiss of death?’, Bruegel Blog, 28 May