Europe Investment Report 2013 — Deciding Who Pays When a Bank Fails

Between 2008 and 2012, the answer to "who bears a failing bank's losses" had been, with few exceptions, the home state's taxpayers.

The reasoning was contagion. Imposing losses on bank creditors — particularly bondholders and depositors — risked triggering withdrawals from other banks perceived as similar. Given the fragility described in the Europe Investment Report 2011, that risk was taken seriously.

The cost of that answer was the doom loop. If a state must absorb its banks' losses, then bank weakness becomes sovereign weakness, per the Global Investment Outlook 2010. And the state's capacity is limited by the banking system's size relative to its economy, which for several members was insufficient.

Bail-in is the structural answer: require a bank's creditors to absorb losses before any public money is used, in a defined order, established in advance.

What it achieves:

• The loss falls on those who lent to the bank, which is where credit risk normally falls.
• The sovereign is insulated, breaking the loop.
• And market discipline is restored — creditors who can lose money have reason to assess the bank, which the implicit guarantee had removed, per the Asia-Pacific Investment Report 2013.

What it costs:

• Bank funding becomes more expensive, since creditors now price actual risk.
• And it creates a run incentive. A creditor who can be bailed in has reason to withdraw at the first sign of trouble — which is the funding run this archive documents repeatedly.

The 2013 episode was the first application at scale, and it went further than expected by reaching deposits above the insured limit. Two features made it a genuine turning point: it demonstrated that bail-in would actually be used, and it required capital controls inside the currency union — which showed that a euro in one member state was not, under stress, fully interchangeable with a euro in another.

The hierarchy is the mechanism

Bail-in is entirely a question of order, and the order determines whether it stabilises or destabilises.

A bank's liabilities in typical order of loss absorption:

1. Equity — first loss, by design, no expectation of protection.
2. Subordinated debt — explicitly junior, priced accordingly.
3. Senior unsecured debt — historically treated as safe in practice, though never legally guaranteed.
4. Uninsured deposits — deposits above the guarantee limit.
5. Insured deposits — protected by the guarantee scheme.
6. And secured liabilities — covered bonds and repo, backed by specific collateral and generally excluded.

Where the loss stops determines everything about the outcome:

Stopping at subordinated debt is uncontroversial. Those investors accepted junior status and were paid for it.

Reaching senior unsecured debt is a genuine change. These instruments had been treated as effectively guaranteed for decades, so the repricing is real — but the holders are institutions capable of assessing risk.

Reaching uninsured deposits is qualitatively different, and this is where the design difficulty concentrates:

• Depositors are not investors. A company holding operating cash at a bank is not making a credit decision — it is using a payment service.
• They can withdraw instantly, unlike bondholders. So a credible threat of bail-in produces a run.
• And the affected parties are frequently businesses, so the loss transmits directly into the real economy.

Bailing in a bondholder imposes a loss on someone who chose to lend. Bailing in a depositor imposes a loss on someone who was using a bank account, and it gives every other depositor a reason to leave tonight.

The framework responses — depositor preference placing all deposits above senior unsecured debt, firm insurance coverage below a threshold, and a required layer of explicitly loss-absorbing instruments sitting above deposits — exist precisely to ensure the loss stops before it reaches deposits.

The design principle: a bail-in works if there is enough loss-absorbing capacity above the run-prone liabilities. Where there is not, bail-in becomes the trigger for the run it was meant to make unnecessary.

Capital controls in a currency union

The most structurally significant feature of the 2013 episode was not the bail-in but what accompanied it.

To prevent deposits leaving during the resolution, restrictions were imposed on transfers — limits on withdrawals, on transfers abroad, and on the movement of funds out of the member state.

Why this is more than a technical measure:

A monetary union's premise is that the currency is the same everywhere. A euro in one member state is a euro in another, fully and immediately interchangeable.

Capital controls break that, temporarily and explicitly:

• A euro inside the restricted jurisdiction could not freely become a euro outside it.
• So the two were not the same asset, and they traded at different effective values.
• Which means the currency union had, for that jurisdiction and that period, partially suspended.

Why it was necessary anyway: without controls, the resolution would have triggered exactly the deposit flight it was designed to manage, and the flight would have destroyed the institutions being resolved.

The precedent that was set:

Capital controls inside a currency union are possible, were used, and were accepted by the union's institutions. That is a permanent change to what participants can assume — and it is the mechanical form of redenomination risk, which the Global Investment Outlook 2011 describes as the thing markets had begun pricing.

The practical consequence for anyone holding assets in a currency union member: the fungibility of the currency across borders is a policy that has been suspended before. It is not a property of the currency itself.

This informed how the subsequent framework was designed — with more emphasis on ensuring resolution could occur without controls, and on making loss-absorbing capacity sufficient in advance.

Supervision was the achievable leg

The banking union agreed in this period delivered one leg substantially and the others partially, and the pattern of what was achievable is instructive.

Common supervision — a single supervisor for the largest institutions:

• Transfers authority, not money. No member state is asked to fund another's banks.
• Addresses the ring-fencing problem from the Europe Investment Report 2008 by giving one authority a union-wide view.
• And it removes the home-supervisor incentive to be lenient with national champions.
• This was agreed and implemented within about two years, which is rapid by the standards of the period.

Common resolution — a single mechanism and fund:

• Requires money, so it was harder.
• The fund was built through contributions accumulated over years, meaning the mechanism existed in law well before it existed in capacity.
• And the decision-making structure was complex, reflecting the difficulty of agreeing who decides to impose losses.

Common deposit insurance:

• Requires the most mutualisation, since it means one member state's taxpayers potentially standing behind another's depositors.
• It was not agreed, and remained under discussion for more than a decade.

The pattern is consistent and worth extracting:

The legs of the banking union were achieved in inverse order of how much money they required. Supervision transfers authority and was fast; resolution transfers money slowly and was slow; deposit insurance transfers money immediately and did not happen.

The consequence, per the Global Investment Outlook 2012, is that the leg most directly addressing the run mechanism was the one not built. Redenomination risk transmits through deposits, and common deposit insurance is the instrument that would remove the incentive to move them.

So the architecture reduces the probability of bank failure and does less about the probability of a deposit run — which is a coherent political sequencing and leaves the original fragility partially intact.

What bail-in does to bank funding

The pricing consequences were immediate and are the practical investment content of the change.

Before credible bail-in:

• Senior unsecured bank debt was priced as though implicitly guaranteed, so spreads reflected the sovereign more than the bank.
• Differences between strong and weak banks in the same country were small.
• And the analysis that mattered was the sovereign's, not the institution's.

After:

• Senior debt prices reflect the bank's own loss-absorbing capacity and asset quality.
• Dispersion between institutions widens, which is the market discipline the change was intended to restore.
• And the capital structure's layers price differently, since each has a defined position in the hierarchy.

The practical analysis required:

• How much loss-absorbing capacity sits below your claim? Equity, subordinated debt and any dedicated bail-in layer. This is the buffer, and it is disclosed.
• What is the asset quality, and how large would losses plausibly be relative to that buffer?
• And where exactly does your instrument sit in the hierarchy, including under the resolution law of the relevant jurisdiction — which varies and matters.

The transitional problem worth noting: instruments issued before the framework existed were priced on the old assumption. Repricing them was a real transfer from holders, and the framework's phase-in was partly designed to manage that.

The Europe Investment Report 2016 covers how the resulting funding costs affected bank profitability, which is the second-order consequence that shaped the sector for a decade.

Resolution needs somewhere for the bank to go

A practical constraint that the framework's design tends to understate: imposing losses is the easy half. The hard half is what the institution becomes the next morning.

What resolution has to achieve simultaneously:

• Allocate losses to creditors in the correct order.
• Keep critical functions operating — payments, deposits, access to accounts — because a bank that stops working stops working for its customers, whatever the creditors' position.
• And produce a viable entity or an orderly wind-down that does not destroy value.

The available routes and what each requires:

Sale to another institution. The cleanest outcome — the business continues under a stronger owner. It requires a willing buyer with capital, which is exactly what is scarce in a systemic crisis. A buyer that exists in calm conditions may not exist when several banks need one at once.

A bridge institution, where the viable business is transferred to a temporary entity while the losses stay behind. This requires public capital for the bridge, at least temporarily, and an eventual buyer.

Bail-in and recapitalisation in place, where creditors are converted to equity and the institution continues under new ownership. This requires sufficient loss-absorbing liabilities — the layer described above — and creditors capable of becoming owners.

And liquidation, which is orderly only for small institutions and destroys substantial value for large ones.

Why the sale route is less available than it appears:

A resolution framework implicitly assumes a buyer. In a systemic crisis the potential buyers are the other banks, and they are dealing with their own problems. The tool that works best is the one least available when it is most needed.

This is why the pre-positioned loss-absorbing layer matters so much. It is the route that requires no counterparty — the creditors are already there, and converting them needs no willing third party. The other routes all depend on someone else being able to act.

The remaining constraint is time. Resolution has to be executed between a Friday close and a Monday open, because an institution that is visibly in resolution during business hours experiences a run. That timetable rules out negotiation, due diligence and price discovery — which is why so much of the framework is about pre-positioning: recovery plans, resolution plans, and loss-absorbing capacity all agreed years in advance, precisely so that nothing has to be decided in the weekend available.

What an allocator could act on

Locate any bank claim precisely in the creditor hierarchy. The position determines the outcome entirely, and it varies by jurisdiction and by instrument.

Measure the loss-absorbing capacity beneath your claim. Equity, subordinated debt and dedicated bail-in instruments are the buffer, they are disclosed, and their size relative to plausible losses is the analysis.

Assess whether a bail-in can stop before reaching deposits. Where loss-absorbing capacity above run-prone liabilities is insufficient, bail-in triggers the run it was designed to avoid.

Treat currency fungibility within a union as a policy, not a property. Capital controls were imposed inside the euro area and accepted, which means a euro in one member state is not unconditionally a euro in another.

Expect dispersion between banks to widen once implicit guarantees are removed. That dispersion is the investable consequence of the framework change.

Note which union legs required money. Supervision transferred authority and was built; deposit insurance required mutualisation and was not, leaving the run mechanism least addressed.

What 2013 established

• Bail-in reallocates losses to break the bank-sovereign loop, at the cost of more expensive bank funding and a run incentive.
• The creditor hierarchy is the whole mechanism, and reaching deposits is qualitatively different from reaching bondholders.
• Capital controls inside a currency union were used and accepted, making currency fungibility conditional.
• Common supervision was achievable because it transfers authority rather than money.
• Common deposit insurance was not built, leaving the leg most relevant to deposit runs absent.