The Eurosystem: An accident waiting to happen

Willem Buiter 01 October 2020



Unlike the Fed, the Eurosystem is not a decentralised unitary central bank.  Instead it is a currency board system with 19 nations, each with a national central bank (NCB), tied to a single currency – the euro – which no NCB can issue at its discretion.  The monetary issuance of each NCB is a collective decision: it must be approved by a majority of the Governing Council of the ECB – its highest decision-making body.1   

For a normal unitary central bank that holds few or no foreign currency-denominated liabilities, default and insolvency are always a choice rather than a necessity because of the option of unbounded monetary issuance. This would be a fair characterisation of the ECB and the consolidated Eurosystem if the NCBs were not independent national profit and loss centres, but mere branches of the ECB. It would also hold for the individual NCBs if their net income or profit were the capital key share of the net income or profit of the consolidated Eurosystem – complete shared-risk assets and activities.  With complete profit and loss sharing, an individual NCB can be insolvent only if the entire Eurosystem chooses to become insolvent – a choice I assume the Governing Council would not make.

The Eurosystem has 20 independent profit-and-loss centres: the ECB and the 19 NCBs.  This unprecedented arrangement is sustainable only if either there is no risk associated with the assets and activities of the NCBs or the risk is fully shared.  This risk sharing can be ex ante, through the imposition of a shared-risk rule for NCB assets and activities, or ex post, through the bailout of an insolvent NCB by the ECB and the remaining 18 NCBs.  To have ‘own-risk’ assets held by and ‘own-risk’ activities undertaken by an individual NCB means that profits and losses associated with these assets and activities are for the account of that NCB alone. Ex ante ‘shared-risk’ means that profits and losses are shared according to the capital key shares in the paid-up capital of the ECB.

NCBs with substantial holdings of risky sovereign debt (mostly debt of their own sovereign) on an own-risk basis (Italy, Spain, and Greece are examples) are at material risk of default if their sovereign defaults.  The odds on recapitalisation of an insolvent NCB by its defaulted sovereign are obviously low to negligible.  Therefore, risk exposures of an NCB that are out of line with the risk exposures of the consolidated Eurosystem are an existential threat to the monetary union. Should a highly exposed NCB suffer losses large enough for it to become insolvent, it would become an ineligible counterparty in TARGET2. It would, in effect, cease to be part of the Eurosystem and its nation would likely crash out of the monetary union.  If the rest of the Eurosystem does not allow the financially afflicted NCB to borrow against its capital key share of the future seigniorage of the Eurosystem as a whole, insolvency of the NCB could occur with quite modest losses.

It is possible that, should an NCB face insolvency because of losses on own-risk assets (most likely own-risk holdings of its own sovereign’s debt), the ECB and the 18 other solvent NCBs would choose to bail out the at-risk NCB, ignoring the own-risk rules of the Eurosystem.  Ex-ante own-risk would morph into ex-post shared-risk.  That removes the risk of a eurozone exit by the nation with the insolvent NCB.  It would, however, likely result in a political backlash in at least some of the 18 other euro area member states (the ones with solvent NCBs) that could result in a voluntary exit from the euro area by one or more member states.  

These are not purely hypothetical issues.  The Eurosystem has risky private and public assets on its balance sheet.  That leaves risk sharing, ex ante or ex post, as a necessary condition for viability and solvency of the individual NCBs.  One bit of good news is that central bank insolvency is not (or ought not to be) associated with negative net worth on the conventional central bank balance sheet.  The conventional balance sheet omits a key asset – the present discounted value (PDV) of future seigniorage – future issuance of central bank money net of any interest paid on central bank money.  A central bank is insolvent only if its comprehensive net worth (allowing for the PDV of its future seigniorage) is negative, i.e. if it violates its intertemporal budget constraint and the associated no-Ponzi finance condition.  For an NCB in the Eurosystem, seigniorage wealth is its capital key share of the PDV of the future seigniorage of the entire Eurosystem.  It is an interesting question whether the Governing Council would permit a financially challenged NCB to borrow against its capital key share of the PDV of the future seigniorage of the Eurosystem as a whole, or, if the markets are unwilling to provide such funding, whether the rest of the Eurosystem would fill the funding gap.

Why does the Eurosystem have own-risk assets and activities?

Protocol No 4 of the Treaty on the Functioning of the European Union (TFEU) asserts that the sum of the “monetary incomes of the NCBs (where monetary income is defined as ‘the income accruing to the national central banks in the performance of the ESCB’s monetary policy function’) shall be allocated to the NCBs in proportion to their capital key shares in the subscribed capital of the ECB.  Likewise, the net profit of the ECB, after an appropriate transfer to the general reserve fund, is to be distributed to the NCBs in proportion to their capital key shares.”

Unfortunately, there are major departures from full risk sharing in the modus operandi of the Eurosystem.  The most important ‘own-risk’ activities are the following.  First, there are the ANFA assets – the gross assets held under the Agreement on Net Financial Assets. These started off as legacy assets of NCBs relating to national, non-monetary policy-related policy tasks.  They should have been wound down, but they were not.  Now included in ANFA activities are Eurosystem lending operations under Emergency Liquidity Assistance (ELA).  I consider the lender-of-last-resort and market-maker-of-last-resort responsibilities of the Eurosystem conducted under the ELA arrangements to be an integral part of its ‘monetary policy function’.  Treating them on an ‘own-risk’ basis is therefore a violation of the TFEU.

Another set of illegitimate own-risk activities are the lending operations under the de facto resurrected tier-2 collateral system of the early monetary union.  Here, an NCB is responsible for any losses on collateral accepted (with the approval of the ECB) by that NCB but not generally accepted in the Eurosystem. The original two-tier collateral framework, in operation since the start of EMU to handle idiosyncratic legacy national collateral, was replaced on 1 January 2007 by a single collateral system. On 15 October 2008, the Governing Council decided to expand – initially until the end of 2009 – the list of eligible Eurosystem collateral on a temporary basis. This was the first of a sequence of collateral standard reductions.  Lending against this new, lower-grade collateral was at the own risk of the NCB providing the credit. There appear to be no publicly available data on how much of this “temporary” tier-2 collateral has been accepted since 2008 by NCBs at their own risk and how much, if any, remains outstanding. My best guess is that these loans have all been repaid by now.

By far the largest own-risk assets are the own-risk purchases under the Public Sector Purchase Programme (PSPP). Under the expanded asset purchase programme (APP) announced by the ECB on 22 January 2015, 80% of the additional asset purchases under the PSPP (consisting of bonds issued by EA central governments, agencies, and European institutions) were own-risk.  In August 2020, Eurosystem holdings of assets under the PSPP were €2,273 billion.  The Pandemic Emergency Purchase Programme (PEPP) has an envelope of €1,350 billion, of which €512 billion had been used by 4 September 2020.  The vast majority of asset purchases under the PEPP consisted of public sector securities: end-July 2020 total holdings were €440 billion, of which €384 billion were public sector securities.  The same rules for risk sharing apply to the PEPP as to the original extended APP: only 20% of the public sector assets purchased under the PEPP are shared-risk. This is clearly a violation of both the spirit and the letter of the TFEU.

Three approaches to reducing or eliminating NCB insolvency risk

There are three strategies to reduce the default risk exposure of an individual NCB. The first, already discussed, is to permit an NCB, following a serious capital loss, to borrow against the security of its capital key share of the future seigniorage revenues of the consolidated Eurosystem – in principle, up to the amount that would reduce its comprehensive net worth to zero. If the markets do not supply that amount of financing, the other members of the Eurosystem should. This does not repair the fundamental flaw in the design of the Eurosystem – the existence of own-risk activities, assets and liabilities in a central banking system with 20 independent profit-and-loss centres – but it reduces the likelihood of this flaw turning out to be fatal. The second is to reduce the weight of risky assets in its balance sheet – risk reduction. The third is to increase the scale and scope of risk-sharing activities.

As regards the second option, risk reduction for the Eurosystem, there is a long list of possible approaches.  All involve increasing the supply of low-risk or risk-free assets to the Eurosystem.  

Financial engineering approaches can create additional safe(r) assets without changing the overall riskiness of the total stock of assets outstanding.  Public or private financial entities, including banks, could issue asset-backed securities (ABS) or collateralised debt obligations (CDOs) backed by a pool of sovereign debt instruments and tranched according to seniority. European safe bonds (ESBies), as proposed by Brunnermeier et al. (2011, 2017a, b), or sovereign bond-backed securities (SBBS) as proposed by the European Commission (2018), are examples of such synthetic safe instruments.  Many other ways of increasing the supply of safe assets through financial engineering involving euro area public debt (or indeed private debt instruments) have been proposed (see Buiter 2020 for a discussion of a representative sample).  None of them reduces the total amount of sovereign risk in the economic system as a whole.  More effective fiscal risk-sharing arrangements between euro area governments could reduce the overall riskiness of the outstanding aggregate stock of euro area sovereign debt.  That would definitely be a desirable development, regardless of whether it is associated with a reduction in sovereign risk on the Eurosystem’s balance sheet.

Indeed, even if it were possible to restrict the balance sheet of the Eurosystem to safe assets only, this would be very counterproductive.  A proper central bank, with limited foreign currency-denominated liabilities, is a natural key player in the joint monetary and fiscal response to aggregate demand and supply shocks and financial instability.  Recognising this would mean accepting the reality that central bank independence is an illusion.  This is quite consistent with granting the central bank a measure of operational independence in the secondary technical activity of setting the policy rates when these are not constrained by the effective lower bound. The fiscal-financial-monetary response to the Global Crisis and the COVID-19 pandemic have made this absence of central bank independence abundantly clear.

In a standard, single-country monetary system, the relationship between the central bank and the fiscal authority (the treasury) is one of fiscal dominance.  The political legitimacy of the treasury (the beneficial owner of the central bank) trumps whatever legitimacy can be achieved (typically output legitimacy) by a central bank that is run by appointed officials.  There is no good reason why such central banks should not put risky assets on their balance sheets when (a) financial instability undermines the viability of the financial sector and threatens the proper functioning of the real economy, and (b) the decisions on the size and composition of the risky central bank balance sheet are agreed with the treasury.  The treasury should, of course, provide a guarantee (or equity) for the risk exposures of the central bank, which through its risky loans and risky debt purchases is engaged in manifestly fiscal activities.  In practice this should amount to a treasury guarantee for all assets owned by the central bank – on- or off-balance sheet.

The one major exception to fiscal dominance is found in the relationship between the monetary and fiscal authorities in the euro area.  Here we have a single monetary authority (the Governing Council of the ECB) facing 19 national fiscal authorities with different objectives and, despite the admirable efforts of the Eurogroup, a proven record of being unable to coordinate fiscal and financial policies and actions.  Even the German finance minister is a midget when confronting the president of the ECB.  The TFEU prevents monetary dominance by banning (through Article 123 TFEU) central bank financing of government budgets through credit expansion or through sovereign debt purchases in the primary markets.  There also are no sovereign guarantees for risky assets the ECB/Eurosystem puts on its balance sheet, and the Treaty does not oblige national governments to ensure that their NCB’s capital and reserves plus revaluation reserves remain adequate.  The result is neither fiscal nor monetary dominance but instead a kind of stalemate – a damaging paralysis that prevents the design and implementation of effective joint monetary, fiscal, and financial policies. 

The first-best solution would be far-reaching fiscal integration in the euro area, combined with a rewriting of the TFEU (starting with deleting Article 123 and followed closely by the deletion of Article 125), which would enable a single euro area fiscal authority to establish fiscal dominance over the ECB/Eurosystem.  The likelihood of anything like this happening in the foreseeable future is zero.

As regards the third option, phasing out all ‘own-risk’ activities should be the ultimate aim. The ideal outcome would be for the euro area NCBs to cease to be separate legal personalities with their own profit-and-loss accounts and balance sheets and to become instead mere branches of the ECB.  It makes obvious sense, but I am not holding my breath.  A gradual approach to eventual full consolidation has a slightly better chance of being politically feasible.  New ELA facilities would be shared-risk facilities.  Legacy ELA facilities would be wound up.  Legacy ANFA activities would also be wound up and no new ANFA activities would be permitted.  Idiosyncratic ‘own-risk’ collateralised lending would be phased out if it still exists and would never be reactivated.  All new PSPP purchases would be shared-risk.  If turning own-risk assets and activities into shared-risk assets and activities applies only to the flows and leaves the outstanding stocks untouched, there could still be an extended period of material NCB insolvency risk. 


It is hard to think of a more ill-designed multilateral monetary system than the Eurosystem. The design failure to ensure that all NCB assets and activities would be shared-risk has been compounded by blatant violations of the TFEU Protocol No 4, which states that the income accruing to the NCBs in the performance of the monetary policy function is to be allocated to the NCBs in proportion to their capital key shares in the subscribed capital of the ECB.  The public debt explosion associated with the COVID-19 pandemic includes high-risk sovereign debt of countries like Italy, Greece, and Spain.  We may well see sovereign defaults by the governments of one or more euro area member states in the next few years.  An NCB whose sovereign has defaulted could have an exposure to the defaulted sovereign large enough to threaten that NCB’s solvency.  The viability of the Eurosystem will be at risk.  

It is high time to turn the Eurosystem into a proper central bank.  The necessary substantive step towards this is to eliminate all own-risk assets and activities.  A nice symbolic finishing touch would be to turn all NCBs into ECB branches.  


Brunnermeier M, L Garicano, P Lane, M Pagano, R Reis, T Santos, D Thesmar, S van Nieuwerburgh and D Vayanos (2011), “European Safe Bonds (ESBies)”, The Euronomics Group. 

Brunnermeier, M, M Pagano, R Reis, S van Nieuwerburgh and D Vayanos (2017a), “ESBies: safety in the tranches”, Economic Policy 32(90): 175-219. 

Brunnermeier, M, M Pagano, R Reis, S van Nieuwerburgh and D Vayanos (2017b), “A Safe Asset for Europe”, Financial Times, Alphaville Guest Post, 28 April.

Buiter, W H (2020), Central Banks as Fiscal Actors; the Drivers of Monetary and Fiscal Policy Space, Cambridge University Press, forthcoming.

European Commission (2018), “Frequently asked questions: Enabling framework for sovereign bond-backed securities”, Press Release Database, Brussels, 24 May.


1 The Governing Council of the ECB consists of the six members of the Executive Board (each appointed for a single eight-year term) plus the 19 NCB Governors, whose terms of office are determined according to national procedures.  The 19 NCBs share 15 voting rights by rotation, with each NCB’s voting right determined by the size of its economy and its financial sector.



Topics:  EU institutions Financial regulation and banking

Tags:  Eurosystem, sovereign risk, insolvency risk, ECB

Visiting Professor of International and Public Affairs, Columbia University and CEPR Research Fellow,


CEPR Policy Research