Europe Investment Report 2008 — One Banking System, Twenty-Seven Responses

Europe's 2008 was not principally a story about which banks held which assets. It was a story about an incomplete institutional design meeting its first test.

The design, as it stood:

• A single market in banking. Institutions could operate across the union, take deposits across borders and lend anywhere.
• National supervision. Each bank was supervised principally by its home country's authority.
• National deposit insurance, with coverage limits set by each member state.
• And national rescue money. If a bank needed capital, its home government provided it.

This works while nothing goes wrong. The moment something does, the incentives of twenty-seven national authorities diverge from the interest of the system they jointly supervise.

What happened in practice:

Guarantees became competitive. When one member state raised or removed the limit on its deposit guarantee, deposits in neighbouring states became relatively less safe — for reasons of national fiscal capacity rather than bank quality. Others had to follow, quickly, to prevent deposits moving. The sequence was driven by competitive necessity rather than by coordination.

Cross-border banks fragmented. A supervisor's duty is to depositors and stability in its own jurisdiction. Faced with a group in difficulty, the rational move is to prevent assets and liquidity leaving the local subsidiary — which is exactly what happened, and which broke the group's ability to manage capital and liquidity centrally.

And rescue capacity was national. A bank's losses fell on its home sovereign, so the size of the banking system relative to the home economy determined whether a rescue was feasible — the ratio the UK Investment Report 2008 identifies as an unbooked fiscal commitment. For several small member states with large banking sectors, it was not.

That last point is the direct link to what followed. A banking problem became a sovereign problem because there was no other balance sheet available, which is the loop the Global Investment Outlook 2010 describes opening.

Why the design fails under stress

The mechanism deserves setting out, because the flaw is structural rather than a matter of anyone behaving badly.

In calm conditions, the arrangement has real advantages:

• Supervisors know their local institutions and their local markets.
• National responsibility means clear accountability.
• And the single market delivers genuine efficiency gains through competition and cross-border services.

Under stress, three incentive problems appear simultaneously:

The supervisor's mandate is national. Its duty runs to depositors and stability in its own jurisdiction, not to the group or to the union. So when a cross-border group weakens, each host supervisor acts to secure the part in front of it — ring-fencing capital, restricting intra-group transfers, requiring local liquidity. Each action is correct under the mandate and collectively they dismantle the group.

The fiscal authority's money is national. A rescue is funded by domestic taxpayers, so a government will not fund losses that arose in another member state, and will resist any arrangement that implies it might.

And deposit insurance is national, so its credibility depends on the fiscal capacity behind it. This makes an identical deposit in an identical bank safer in one country than another — which is precisely what a single market is supposed to eliminate.

The institutions were European and the safety net was national. Under stress the safety net's boundaries reasserted themselves, and the single market stopped at exactly the moment it was most needed.

What a complete design requires — common supervision, common resolution, common deposit insurance — is the banking union agreed years later, and the Global Investment Outlook 2012 sets out which legs were eventually built and which were not.

Guarantees as a competitive weapon

The deposit guarantee sequence of late 2008 is worth following, because it demonstrates how a purely defensive national action becomes an aggressive one across a border.

The mechanism:

1. One member state raises its deposit guarantee, or removes the limit entirely, to reassure its own depositors.
2. Depositors elsewhere now face a choice. The same amount of money is fully guaranteed in one jurisdiction and partially guaranteed next door.
3. Money moves, or is credibly expected to.
4. Other member states must match, not because their banks are weaker but because their guarantee is now the weakest.
5. The floor ratchets up across the union, driven by competition rather than by any assessment of what is needed.

Why no participant is behaving unreasonably:

• The first mover is protecting its own depositors, which is its job.
• The followers are preventing an outflow that would damage otherwise sound banks.
• And nobody can decline to follow, because the cost of being the least-guaranteed jurisdiction is a deposit run.

Two consequences:

The guarantees expanded far beyond what any coordinated process would have produced, because the process was competitive rather than deliberative.

And the fiscal commitment expanded with them. A guarantee is a contingent liability, per the UK Investment Report 2008. Raising it raises the state's exposure — and it was raised most urgently by the states with the least room, since they were the ones losing deposits.

The eventual response was a harmonised minimum coverage level across the union, which removes the competitive dynamic by removing the differential. This is the correct fix and it took the crisis to produce it.

Fragmenting a bank along borders

The behaviour of cross-border banking groups under stress is the clearest illustration of the design flaw, and its consequences persisted for a decade.

How a cross-border group normally operates:

• Capital and liquidity are managed centrally, moved to where they are needed.
• Funding is raised where it is cheapest and deployed where returns are best.
• And the group is analysed and managed as one balance sheet.

This is the efficiency the single market was designed to permit.

What host supervisors did under stress:

• Required local subsidiaries to hold their own capital, unavailable to the group.
• Restricted liquidity transfers out of the local entity.
• Applied local liquidity requirements independently of the group's position.
• And in some cases required conversion of branches into separately capitalised subsidiaries.

Each was a reasonable response to a real risk — that a parent in difficulty would pull resources out of a local subsidiary, leaving local depositors exposed.

The aggregate effect:

• Capital became trapped. A group could be adequately capitalised overall while unable to move capital to where losses were occurring.
• The efficiency gains reversed. The group became a collection of nationally-constrained entities that happened to share a brand.
• And the cost of operating cross-border rose permanently, since the trapped capital had to be held in each jurisdiction.

Ring-fencing is what a supervisor does when it cannot rely on someone else's rescue. It is the rational response to an incomplete union, and it undoes the single market one jurisdiction at a time.

This did not reverse when the crisis passed. Trapped capital and local liquidity requirements largely persisted, which is one reason European banking remained more nationally fragmented than the single market implies — a point the Europe Investment Report 2016 develops as a profitability problem.

Recognising losses slowly

A comparative point that determined how long the damage lasted, and it is one of the clearest natural experiments available.

The US approach, per the UK Investment Report 2008's description of the recapitalisation template: force recognition, apply a common stress assessment to all major institutions, require capital to a specified level, and provide state capital where private capital was unavailable.

The European approach was slower and more fragmented:

• Assessments were national and used different methodologies, so results were not comparable.
• Recognition was gradual, with losses emerging over years rather than being crystallised.
• Recapitalisation was case-by-case rather than system-wide, which reintroduced the stigma problem — accepting capital identified an institution as weak.
• And the fiscal capacity to force it was national, so a government facing a large banking sector had every incentive not to look too hard.

Why slow recognition is costly, per the Global Investment Outlook 2010:

• A bank carrying unrecognised losses does not lend. It preserves capital against a loss it has not yet booked.
• It also does not fail, so the capacity is neither used nor released.
• The uncertainty deters funding, since counterparties cannot assess the true position.
• And the delay compounds, because a weak economy generates more losses.

The observable consequence was that European bank lending remained impaired for far longer, which is the transmission problem the Europe Investment Report 2015 identifies as central to the region's slower recovery.

The eventual fix — a common asset quality review and stress test under a single supervisor — arrived in 2014, six years later. The Europe Investment Report 2014 covers it.

When the rescue capacity is national

The arithmetic that converted a banking crisis into a sovereign one is worth stating explicitly, because it explains what followed rather than merely describing 2008.

The chain:

1. A bank needs capital, and private markets will not provide it.
2. The home state provides it, because no other authority can.
3. The state borrows to do so, adding to its debt.
4. The market assesses whether the state can afford this.
5. If the banking system is large relative to the economy, the answer may be no.
6. Doubt about the sovereign then damages the banks further, since they hold its debt and depend on its guarantee.

Step five is where the union's design mattered most. A member state facing this had no larger balance sheet to appeal to, because there was no union-level resolution fund, no common deposit insurance and no fiscal capacity at the centre.

And step six is the doom loop the Global Investment Outlook 2010 sets out, entered from the banking side rather than the fiscal side.

A national rescue of a continental banking system works only where the nation is large relative to its banks. Where it is not, the rescue transfers the problem rather than solving it — and there was nowhere further to transfer it to.

The extreme cases were small economies with banking sectors many times their output, where the guarantee was arithmetically impossible before it was tested.

The structural response — a common resolution mechanism with a common fund, and rules requiring creditors to absorb losses before taxpayers — directly addresses this, and the Europe Investment Report 2013 covers its design and its first application.

Branch or subsidiary decided who paid

A technical distinction that most depositors had never heard of turned out to determine who protected them, and it is a clean illustration of the design flaw in operation.

A bank operating across borders can do so in two forms:

A branch is part of the parent institution, not a separate legal entity. It is supervised by the parent's home authority, and its depositors are covered by the home country's guarantee scheme — even though the branch sits in another country and takes deposits from that country's residents.

A subsidiary is a separately incorporated company. It is supervised by the host country and its depositors are covered by the host country's scheme, with its own capital held locally.

Why the branch form existed: the single market's passporting arrangement was designed to let institutions operate union-wide without duplicating capital and supervision in every country. It is the mechanism that makes a single banking market possible, and it delivered real efficiency.

What it meant in 2008:

• A depositor in one country, banking with a branch of a foreign institution, was relying on a foreign government's guarantee scheme.
• Most did not know this. The branch looked like a local bank, advertised locally, and paid competitive rates.
• And the home country's ability to honour the guarantee depended on its own fiscal capacity — which, where the banking system was many times the size of the home economy, was insufficient.

The failure mode is direct:

A depositor chose a bank on its interest rate and inherited a claim on a foreign treasury they had never considered. The guarantee's value depended on a fiscal position in another country, disclosed nowhere in the account terms.

The host country then faced an impossible position — its residents had lost deposits, its own scheme did not cover them, and the responsible scheme could not pay. Several host governments compensated their residents anyway and then pursued the home state, which produced disputes lasting years.

The structural response was to require large cross-border operations to be subsidiarised in significant host markets, with local capital and local coverage. This reduces the efficiency the passport was designed to deliver and removes the mismatch — which is the same trade the ring-fencing section describes, formalised into law.

What an allocator could act on

Check whether a safety net's boundaries match the market's boundaries. A single market with national supervision, insurance and rescue money fragments under stress along the lines of the money, not the market.

Expect ring-fencing when a cross-border group weakens. Host supervisors secure what is in front of them, which traps capital and means group-level ratios overstate what is actually deployable.

Read national deposit guarantees as claims on national fiscal capacity. An identical deposit in an identical bank is not equally safe across jurisdictions if the guarantors differ in their ability to pay.

Track the pace of loss recognition, not just the size of the losses. A system that recognises slowly has impaired lending for years, and the delay is observable in comparative provisioning and capital-raising timelines.

Compare banking sector assets to home economy size for every sovereign. It determines whether a rescue is feasible, and where it is not, a banking problem becomes a sovereign one mechanically.

Note that competitive guarantee expansion is a fiscal event. The states forced to raise coverage fastest are those losing deposits, which are frequently those with the least room to stand behind the promise.

What 2008 established

• A single market with national safety nets fragments under stress, because supervisory and fiscal mandates are national.
• Deposit guarantees became competitive, ratcheting upward through necessity rather than assessment.
• Cross-border groups were ring-fenced into national entities, trapping capital permanently.
• Loss recognition was slower than in the US, prolonging impaired lending for years.
• National rescue capacity for a continental banking system converted bank losses into sovereign risk.