[Skip to Content]

Register to receive the OMFIF Daily Update and trial the OMFIF membership dashboard for a month.

* Required Fields

Member Area Login

Forgotten Password?

Forgotten password

Analysis
Why the ECB rules on ‘risk limitation’ do not obviate shared liability

Why the ECB rules on ‘risk limitation’ do not obviate shared liability

National central banks cannot avoid exposure to bond purchases

by Frank Westermann, Osnabrück University

Thu 5 Feb 2015

An important element in the European Central Bank quantitative easing programme starting next month is the limit on risk-sharing, with risks from asset purchases residing largely with national central banks carrying out the operations.

Mario Draghi, the ECB president, has played down the issue, terming discussion on the subject ‘quite futile’. In fact, delving further into the risk-sharing agreement is not futile at all. The accord was a concession to German and Dutch concerns about risk exposure, but a study of the underlying economics shows that individual euro members, or the euro system as a whole, really cannot avoid being exposed to risks from individual national central bank's (NCB) bond purchases.

The risk-limitation agreement conceals, but does not obviate, the reality of joint central bank liability. On the other hand, the technical practices and accounting rules of the euro system have been framed in such a way that the issue of possible central bank losses arising from purchase of government securities can be successfully hidden for many years. The timing of the crystallisation of losses, and the definition of who will bear them, will remain extremely opaque.

Through NCB asset purchases, new money can flow freely across borders, reshaping frameworks of assets and liabilities in a way that renders ineffective the ECB’s risk-limitation arrangements.

A number of scenarios make this clear. We can consider first the position of a euro area government that anticipates imminent difficulty repaying bond issues purchased by a NCB under the QE programme. As long as these bonds are eligible collateral, and the ECB maintains its full allotment policy, the government concerned should be able to find a private bank that will buy newly issued bonds. The private bank can use these bonds as collateral to obtain refinancing loans from the central bank. And the government can use the revenues to replay the maturing bonds held by the NCB.

From the central banks’ perspective, collateralised loans would replace the purchases of the now-maturing bonds – but unlike bonds, these loans are subject to loss sharing. Also, the government does not need to default, as – indirectly, via the private banks – it will continue to have access to central bank financing. Furthermore, a fall in prices will not immediately lead to losses; there is no ‘fair value’ accounting for bonds purchased under QE, as they are classified as ‘held to maturity’.

The above scenario holds good as long as a member country’s securities are deemed eligible as collateral. If however the ECB decided that the bonds of a particular country were no longer eligible, a solvency problem could arise. In that case, it is unlikely that the other central banks could avoid joint liability.

Suppose a debt rollover is due that the government is unable to finance. Private investors would anticipate this and move funds across the border to other euro area countries. The liquidity created by QE bond purchases could move to another country in a single day. In this case, the recipient national central bank would accumulate Target-2 assets on their balance sheets, measuring a build-up of claims on the euro system; the central bank under pressure would itself record a rise in its Target-2 liabilities vis-à-vis the system.

Since Target-2 assets are also subject to loss-sharing, the anticipated reaction of financial markets to the perceived insolvency risk would create a joint liability. This would arise well before any test of an individual NCB liability over its asset purchases. Although Target-2 balances have beenconverging since summer 2012, Italy’s Target-2 liabilities have increased in the past few months by nearly €80bn. Against this background, the rest of the euro area has a considerable incentive to help prevent the country concerned from becoming insolvent.

We need to consider, too, the possibility of a surprise move, under which a NCB writes off anticipated losses from QE bond purchases before the private sector had a chance to move its assets abroad. However, this scenario is unlikely. When it was set up, the ECB created its own set of accounting rules, largely following international financial reporting standards. This set of rules lays down that, when assets are held to maturity, the euro system applies an ‘incurred loss model’. A write-off can occur only when the rollover of a maturing debt actually proves to be unsuccessful.

If markets anticipated today that insolvency would occur in the next months, they would have plenty of time to bring their money to safety before a central bank would be allowed – and forced – to write off the losses. While this accounting rule, up to now, is in line with IFRS (see IAS39), it stands in sharp contrast to the precautionary principle, which is also part of the ECB rulebook.

The International Accounting Standards Board has recently introduced the IFRS9, which ensures a more conservative approach by applying the ‘expected loss model’. As long as the euro area does not implement this, however, and countries do not explicitly default, sudden write-offs of losses – individual or shared – are unlikely to happen in the near future. The issue of central bank losses from QE will preoccupy us for many years. But, if and when they are crystallised, the losses are likely to be borne by the euro system as a whole.

Frank Westermann is professor of Economics at Osnabrück University, and a member of the OMFIF Advisory Board.

Tell a friend View this page in PDF format