The third leg of European banking union, the common deposit insurance scheme, has been postponed until an unspecified reduction in bank risk has been achieved. A reduction in banks’ domestic sovereign exposure has (mostly rightly) been identified as an objective, potentially via risk-weighting sovereigns. However, risk-weighting sovereigns is the wrong way to go because it both undermines the most fundamental concepts of government and governance (trust), and would build a strongly pro-cyclical element into policy making. There are simpler and better ways to achieve standardised limits on banks’ exposure to sovereigns.

1. The context: The doom-loop and the importance of banking union

Few people would question the desirability of reducing the interdependencies between sovereigns and banks. The eurozone debt crisis is an acute reminder of this. During the crisis, a number of systemically important banks were bailed out by their government, including in Greece and Ireland (each to the tune of about 35% of GDP), the Netherlands (about 17% of GDP), Germany and the UK (each about 12% of GDP) and Spain (some 8% of GDP), causing public debt to increase significantly, hence burdening future taxpayers and negatively impacting the creditworthiness of the sovereign itself. In the eurozone periphery, this led to a severe negative feedback loop.

So far the solution to break this doom-loop in the eurozone has been European banking union. The banking union is supposed to stand on three legs: common supervision, common resolution and common deposit insurance – these are the three responsibilities, that will be transferred from the national level to the common European level, thereby reducing the interdependencies between banks and their home-country sovereigns.

Specifically, the banking union will secure uniform supervision, thereby helping to level the playing field. With additional capital requirements, regulatory tightening and beefed-up supervision, it should reduce the risk of bank failures, and – when failures do happen (as they must occasionally in a well-functioning market economy) – private creditors would take the hit. Furthermore, in the extreme case where the cost exceeds bail-in-able liabilities, the cost to the home-country sovereign would be reduced by spreading the cost across all eurozone governments.

However, when it came to that (for society) most important group of creditors, the depositors, progress towards full banking union took a significant step back in mid-June, when the Eurogroup of finance ministers kicked the third leg of the banking union, the European Deposit Insurance Scheme (EDIS), firmly into the long grass.

Speaking ahead of the Eurogroup meeting, German Finance Minister, Wolfgang Schäuble, said that talks “will not start with a deposit insurance to strengthen banking union, but with reducing the risks in the banking sector step by step”, a sentiment that was reflected in the post-meeting communiqué, which said that negotiations on the EDIS will start “as soon as sufficient progress has been made on the measures on risk reduction”.

It is not entirely clear what specific risks in the banking sector the Eurogroup was referring to, or what degree of reduction in such risks they are seeking in order to allow the EDIS to go ahead, but earlier this year, the president of the Eurogroup, Dutch Finance Minister, Jeroen Dijsselbloem, included on the agenda of a Eurogroup meeting a proposal to introduce risk weights for sovereign bonds on banks’ balance sheets, aimed at forcing a reduction in banks’ holdings of their home-country domestic sovereign debt.

Why do the authorities want to reduce banks’ holdings of their domestic sovereign debt? The answer is that the doom-loop between banks and their sovereigns works both ways:

When big (in the context of the local economy) – or politically connected − banks get into trouble, history shows that sovereigns step in, and not only when a major financial crisis is on the doorstep. As recently as in May 2016, the European Commission finally approved another EUR 3bn in guarantees for HSH Nord Bank in northern Germany, provided by the states of Hamburg and Schleswig-Holstein.

And, when sovereigns get into trouble in terms of their creditworthiness, banks suffer as well. Importantly, however, while it is a historical fact that sovereign states have restructured their debt in the past, it is a rare event, particularly outside emerging markets. In Europe, sovereign defaults have happened only in the aftermath of major political upheavals, including after WWII and in Central Europe following the transition from communism to market-based democracies, as well as in Greece, following the revelation of years of untruthful statistical reporting (a situation that might have been avoided if Eurostat auditing of national accounts throughout the EU had been approved years earlier, as proposed, but rejected by the Council).

Also importantly, historically, private-sector exposure, rather than sovereign debt, has been the weak point on banks’ balance sheets. It was not excessive exposure to domestic sovereigns that caused trouble during the eurozone sovereign debt crisis (Greece is a special case because it was not being honest about its accounts). Rather, it was excessive exposure to real estate that necessitated bank bailouts in Ireland and Spain, and exposure to derivatives and US CDOs, which they basically had not understood, that caused the need to bailout banks in Germany and the Netherlands, both areas which – with the benefit of hindsight – could have been avoided relatively easily with proper supervision.

In other words, in reality the link between sovereigns and banks runs – the vast majority of the time – from banks to sovereigns, and very rarely the other way around. No measurable bailout was needed in France and Italy, but the vulnerability of Italian banks today, namely the NPLs, rests on exposure to the SME sector, not the sovereign.

Moreover, if the 2008-12 crisis holds any lesson, it is that domestic banks play an important stabilising, not a de-stabilising, role for their sovereigns. By any measure, the peripheral sovereign debt crisis was severely exacerbated by the sudden and large withdrawals of exposure by non-domestic banks based in core Europe, often encouraged, if not outright dictated, by national supervisors. BIS statistics show that core European banks withdrew around a trillion euros from the periphery during the critical 2-3 years. Had domestic banks not held on to their sovereign securities, and at times even added to their holdings, it is almost inconceivable that a full-scale crisis with either substantially greater external official assistance and/or sovereign defaults could have been avoided.

If banking union with common supervision had been in place then almost certainly the withdrawals would not have been so large and politicians would not have had to worry to the same extent about depositors. The run on Northern Rock in 2007 ended only when then chancellor Alistair Darling announced on TV that all deposits in the UK were safe. And when Hypo Real Estate collapsed in Germany a year later, Angela Merkel announced that “we tell all savings account holders that your deposits are safe. The federal government assures it.” In neither of these cases did the government have the fully funded resources to back the claims, or even an established parliamentary majority for the contingent liability, but the political commitment was sufficient to stabilise the situation.

2. Banks’ exposure to the sovereigns

While history is important, regulators must not fight the last war, but rather look to the next one, and here – given the dramatic increase in sovereign debt-to-GDP ratios in recent years – it is unquestionably important to pay attention to banks’ sovereign exposure.

To be sure, it is not the case at the moment that banks can hold unlimited amounts of sovereign debt on their balance sheets. Banks are heavily regulated and supervised businesses, with their supervisors signing off on the risk-weighting approach used by any bank – whether it’s a simple standardized approach (used by most small banks), or the internal or advanced internal approach used by bigger banks – when assigning risk-weights to the various assets held on the balance sheet.

Of course, the implicit (ad hoc) limits currently in place could beneficially be replaced by a standardised ex ante approach, but doing so by introducing risk-weighting of sovereign debt is misguided – and the fact that it now seems to be holding up the critical third leg of the banking union is deeply troublesome because we are left in a situation where the supervision and resolution authorities have been transferred away from the national level to the European level, while part of the risk (the deposit insurance), if anything goes wrong, has been left at the national level.

Nobody can be expected to transfer power away to another level of government, while keeping the risk – just as nobody can be expected to accept the assignment of new risk without also receiving the authority to manage it.

Constructing a three-legged stool, where the fitting of the third leg has become conditional on other, albeit related, issues, is a troublesome enterprise, which implies a significant risk of it being unable to function as intended. Indeed, the chances are that as long as this deficiency remains in place, the national supervisors will, with some justification, demand deep involvement in the ECB’s supervisory work, and if or when bank resolution has to take place, the national authorities will lead, if not monopolise, the process because they know that if a run were to emerge, they, and they alone, would have to carry the cost. The Italian governments’ handling of the trouble surrounding Banca Populare di Vicenza and Veneto Banca, and the establishment of Atlante, can be seen in this context.

In what follows, I’ll first explain why risk-weighting of sovereign exposure is philosophically troublesome, then I’ll discuss the huge practical obstacles if one were to get comfortable with the philosophical hurdles, and finally I’ll outline the (better) alternatives.

3. The philosophical problems with risk-weighting sovereign debt

While sovereign defaults are historical fact, to institutionalise sovereign risks ex ante is troublesome on several fronts. First, it carries significant risk in terms of how the population and businesses see their elected governments.

The role of government is to provide a set of services to the population, including internal and external security, a judiciary function, education, infrastructure, etc, financed by taxes and (usually) periods of borrowing. In other words, virtually every government faces a set of obligations, ranging from legally contracted financial obligations (e.g. debt and procurements once contracts have been signed), to politically contracted financial obligations (e.g. pensions in PAYG systems) and politically promised obligations (e.g. education.)

It is therefore not surprising that no government would support an institutionalised risk on its own legal claims (apart, apparently, from a very few eurozone governments seemingly trusting that whatever system they concoct would deem themselves virtually risk free – at least for the time being!) in return for what they (incorrectly) see as effective risk reduction among banks in other eurozone countries.

Indeed, it seems odd that a government – preaching the rule of law – should consider defaulting on its legally contracted obligations before having exhausted all other (non-legally binding) options, including raising taxes and cutting expenditure. But the political reality is more complex. While tax increases and spending cuts have been employed in all crises, as they evolve (sometimes with the help of IMF programs), there are, of course, limits to how far down this road any government would, or could, go.

More importantly, a sovereign debt restructuring is never an orderly and pre-designed process. Indeed, there is no example of a government restructuring its debt without having first run up arrears on their procurements, including medical and educational supplies, and fallen behind in public sector wage and pension payments. In other words, default on debt is always closely associated with – and trails – a broader decaying of government functions, if not an outright breakdown of governance. This is not a scenario any responsible government would want to institutionalise ex ante.

Specifically in the context of the EDIS, if indeed at some point in the future sovereigns were to become the weak point on banks’ balance sheets, institutionalising a quantitative risk assessment of sovereigns – particularly at a time when they still underwrite each country’s deposit insurance scheme – could be dangerous. You would basically be telling the population in several countries that 1. the European authorities assign a risk of x% to a default of your country, and 2. if you, and your fellow citizens, were to withdraw your money from your banks, then the only underwriter of those deposits would be…your government. Does that not imply a risk of a self-fulfilling disaster?

Finally, in addition to the profoundly political issue of trusted governance, there is a more technical, yet, very important, issue: If sovereign debt in the eurozone were to be risk-weighted, the euro would be a currency explicitly without a “risk-free rate” (or an anchor for pricing securities), unless a new common “risk-free” (anchor) asset is created. This is a status “enjoyed” only by the poorest emerging markets, and indeed a factor holding them back from the market-based pricing of assets. To imagine a well-functioning market economy without a “risk-free rate” in its own currency would defy the finance theories underpinning any known market economy.

4. The practical issues of implementing risk weights

Suppose one could get comfortable with the issues discussed above, one would then have to decide who the judge would be for these relative probabilities of default among countries and the changes in them?

Three possibilities are being mentioned, none of them comfortable:

• The credit rating agencies: If you think that’s a good idea, would ask you to read our paper from March 2014 (The Damaging Bias of Sovereign Ratings; since then also published – after proper academic review – in the academic journal Economic Notes^[1]D. Vernazza and E. F. Nielsen (2015), “The Damaging Bias of Sovereign Ratings”, Economic Notes, 44(2), pp.361-407.). In it, we document the disastrous overreaction by the agencies in terms of downgrades, well beyond the fundamental deterioration in the eurozone periphery during the crisis – a time when the agencies fell in love with the “Fragile Five” and upgraded them well beyond changes in fundamentals.
• A mark-to-market approach: This would be dangerously pro-cyclical. If markets worry about the creditworthiness of a country, it will sell that country’s debt, thereby imposing a tightening of financial conditions in that country – on top of that you demand fiscal tightening, and hence still more drag on the economy. This is precisely what happened in Italy when Mario Monti took over as PM: markets sold BTPs and imposed severe monetary tightening on the Italian private sector, and – “dictated” by markets and credit rating agencies – Mr. Monti imposed massive fiscal tightening, thereby further punishing the real economy. But, rather than seeing through the hardship to the better future, markets began to question whether Italy would ever grow again and the negative spiral – the true doom-loop – began (only to be broken when Mario Draghi promised to do whatever it takes, implemented as the OMT). This is particularly true in a monetary union where the FX cannot do the adjustment.
• You could impose politically-agreed sovereign-specific limits on banks’ holdings, but it would be uncomfortably close to credit allocation by degree. Hardly a recipe for growth and healthy financial institutions.

Whichever way you do it, you’ll get something pro-cyclical and ultimately dangerous. Imagine Europe being hit by another downturn (as happens every so often), and everyone, including Germany, sees their tax revenues drop and unemployment payments go up. Fiscal deficits rise, and rating agencies and markets get nervous, as they tend to do in these circumstances. As a result, in this new regime, banks will need to reduce their holdings of sovereign debt, including that of Germany, so the bund curve (and other sovereign curves) move higher, which means that financial conditions tighten. Is that really the policy response to a slowdown you want to build in?

5. Alternative ways of limiting the sovereign-bank nexus

In the long term, the key will be to lower the share of banks in credit intermediation. As is well-known, the vast bulk of lending in the eurozone is done by banks, far higher than in the US. Here the Capital Markets Union, which is complementary to banking union and already in train, is the right approach. Taxation could also be changed to encourage equity financing over debt financing. Importantly, however, given the structure of European businesses, with the important part played by SMEs, these are steps that would need to be calibrated carefully before being introduced.

Meanwhile, the following three measures should be considered:

1. Encourage geographical diversification of banks’ private-sector lending and other financial services.

First and most importantly, sovereign exposure cannot be equated with private sector exposure because of the interconnectivities between the sovereign and the private sector. After all, to a very large extent, the sovereign is a mirror imagine of the domestic private sector. Therefore, the focus should be on banks’ exposure to the home-country sovereign as a share of total GDP. This would reflect both the totality of government and the connectivity between the sovereign and the private sector that generates GDP. (On this more appropriate measure, European banks look quite alike: German banks hold 8.7% of GDP in claims on the German general government; French banks hold 8.2% of claims on the French government; Italian banks hold 11.5% and Dutch banks 10.6%. Spain is a bit of an outlier with 18.1%.)

To illustrate the interconnectivities, imagine an extreme scenario in which Italian banks no longer hold any claim on the Italian government, and imagine a restructuring of the Italian government’s debt – of which more than 50% is held by Italian entities (and let’s even assume that the part of this holding previously held by Italian banks is now distributed to creditors outside Italy.) Now work through an Italian sovereign debt restructuring scenario and consider how Italian banks (with no direct exposure to the sovereign) would fare. How would Italian households – presently holding 8% of the debt – react economically (let alone politically)? And how would non-bank Italian institutions, holding 40-50% of the debt, react? How far would GDP drop?

And with the remaining 40-50% of the debt held by foreign financial institutions, almost certainly in countries whose business sectors are deeply integrated with Italy, the impact on the rest of Europe would be measurable as well, with a feedback loop back to Italian GDP. Now consider what would happen to all other assets on Italian banks’ balance sheets in this scenario. The point is that you may be able to take the home-country sovereign debt out of the banks, but you cannot eliminate the risk to banks of a domestic sovereign default.

The solution is to encourage, via regulation, and maybe even taxation, a greater degree of geographical diversification in banks’ core business of servicing the non-bank private sector.

2. Consider incentives for banks to change their registration to national legislation to a European one.

To supplement the point above, Europe could consider the establishment of a framework for an “elite” group of banks e.g. by offering an advantageous European license to banks that fulfil a certain number of criteria with respect to geographical distribution of their business and particularly solid capital ratios.

3. Cap sovereign exposure as a share of equity and/or GDP.

A straightforward way to reduce the link between the home-country government and banks is a simple cap on the sovereign bond holdings of banks as a share of capital, e.g. at 100%, and/or of GDP, e.g. at 15%. Crucially, the cap would be the same for all banks, irrespective of where they are registered. This could come with a bonus for diversification (assuming you properly account for the covariance of the sovereign bond yields). Such an approach would circumvent many of the problems I discussed above, but not all, as many banks will still be highly dependent on the economy of their home country.