Greece & the Periphery Report 2010 — The Doom Loop Opens

Greece is roughly two per cent of eurozone output. On any ordinary reading, its public finances should have been a contained problem. The reason it was not is that the crisis was never really about Greece. It was about what Greece revealed.

Since the introduction of the euro, the market had priced peripheral sovereign debt as though it were close to German debt. Spreads compressed dramatically and stayed compressed. That pricing embodied a belief — never stated as policy, never formally denied — that a eurozone member's debt was in some sense collectively underwritten.

Greece falsified the belief. Once it became clear that no automatic mechanism stood behind a member state's obligations, the question was not whether Greece could pay. It was what every other peripheral bond was worth, given that the assumption behind its price had been removed. A national fiscal problem became a categorical repricing, which is why contagion moved to countries whose fiscal positions bore no resemblance to Greece's.

The transmission ran through bank balance sheets, and this is the part that made it self-reinforcing. European banks held large quantities of their own sovereign's debt, encouraged by capital rules that assigned it a zero risk weight — meaning a bank could hold it without capital against it. That made sovereign exposure the cheapest asset available, and banks held it accordingly.

So a sovereign repricing was immediately a banking problem. And a banking problem in a country without a shared fiscal backstop is a sovereign problem, because the state must recapitalise its banks, which increases its debt, which worsens the repricing that started it. That circuit — the doom loop — is the defining mechanism of the European crisis, and it had no equivalent in the United States, where bank support was federal while the stressed borrowers were households and states.

Two structural absences made it unresolvable by ordinary means. Devaluation was unavailable, because there was no national currency to devalue. A lender of last resort for sovereigns was unavailable, because the central bank's mandate was contested on precisely that question. What remained was internal devaluation: reducing costs and prices directly, through wages and public spending. It works, slowly, and it is contractionary while it works — which is where the arithmetic of the debt ratio turns against the policy.

What monetary union had removed

To see why the response was so constrained, it helps to state what a country with its own currency does in this situation, and then remove those options one at a time.

The ordinary adjustment path. A country that has lost competitiveness and cannot fund itself devalues. Its exports become cheaper, imports dearer, the external balance corrects, and — critically — the adjustment happens through the exchange rate rather than through wages and employment directly. It is disruptive and it is fast.

The ordinary funding path. A country borrowing in its own currency has a central bank that can, in extremis, buy its debt. This does not make default impossible, but it makes a funding failure — an inability to roll over debt at any price — very unlikely, because there is a buyer of last resort.

Monetary union removed both, and the second removal was the more consequential:

• No devaluation. Competitiveness had to be regained by lowering domestic costs relative to trading partners — internal devaluation. This works through unemployment and wage reductions, takes years, and is politically brutal in a way an exchange-rate move is not.
• No unambiguous lender of last resort. The European Central Bank's mandate on whether it could stand behind member-state debt was genuinely contested, and that ambiguity was itself the problem. A backstop that might not exist provides almost none of the benefit of one that certainly does.
• No fiscal transfer mechanism. In a federal system, a region in difficulty receives automatic transfers. The eurozone had monetary union without the fiscal architecture that normally accompanies it, so adjustment fell entirely on the member state.

The euro removed the shock absorbers and kept the shocks. That is a defensible design if members' economies converge; it is a trap if they diverge, and nothing in the architecture forced convergence.

This is the structural point the archive returns to. The Europe Investment Report 2011 describes redenomination risk becoming a live price — the market pricing the possibility that a bond might be repaid in a different currency. That risk exists only because the union's incompleteness made exit conceivable.

Zero risk weight, and why banks held their own sovereign

The mechanism that turned a sovereign repricing into a banking crisis was a capital rule, and it was not an oversight.

The rule. Under the prevailing framework, exposures to a member state's own sovereign debt could be assigned a zero risk weight — held without capital allocated against them.

Why that is not obviously wrong. Capital exists to absorb unexpected loss. If a sovereign's own-currency debt is treated as the risk-free benchmark, requiring capital against it is incoherent, because the whole framework is calibrated relative to that asset.

Why it was wrong here, and the error is specific: eurozone members do not issue in a currency they control. They borrow in what is, functionally, a foreign currency — one whose issuance is decided collectively and which no single member can create. That makes their debt a credit instrument, not a risk-free one, and the rule treated it as the latter.

The consequences were predictable once stated:

• Sovereign debt was the cheapest asset on the balance sheet, since it consumed no capital. Banks held it in size.
• Home bias intensified, because domestic banks were the natural buyers of domestic issuance, particularly as foreign holders withdrew.
• The concentration was therefore worst precisely where the sovereign was weakest, since a stressed sovereign's bonds were increasingly held by its own banks as everyone else sold.
• No capital stood against the exposure, so a markdown went straight through to solvency.

The result was that the banking system most exposed to a sovereign was the one least able to survive its repricing — and it was the system that sovereign would have to rescue.

The doom loop, stated precisely

The circuit is worth setting out step by step, because its self-reinforcing property is the whole difficulty.

1. Sovereign spreads widen on a reassessment of credit risk.
2. Domestic banks mark down their sovereign holdings, reducing capital — with nothing set aside against it.
3. Bank funding costs rise, because a bank is not a better credit than the state standing behind it. Wholesale funding shortens, then withdraws.
4. Banks contract lending to preserve capital, which reduces domestic activity.
5. Weaker activity reduces tax revenue and raises the deficit, worsening the fiscal position.
6. The state must support its banks, adding contingent or actual liabilities to its own balance sheet.
7. Return to step one, worse.

Two properties make this different from an ordinary banking crisis.

The rescuer is the source of the loss. In a normal crisis, the sovereign is external to the banking problem and can absorb it. Here the sovereign is the impaired asset, so its support is worth less exactly as it is needed more. The guarantee is correlated with what it guarantees, which is the same structural flaw as insurance written by a counterparty exposed to the insured event.

There is no external anchor. With a national central bank, the loop can be broken by a credible commitment to fund the sovereign. Without one, each turn of the circuit is a genuine deterioration with nothing arresting it.

A guarantee is only worth the independence of the guarantor from the event. The eurozone's design made the guarantor and the event the same thing.

This is why the resolution, when it came, was verbal rather than fiscal. The Europe Investment Report 2012 describes a credible commitment repricing a continent without being spent — because what the loop needed was not money but the removal of the possibility that money would be unavailable.

Why liquidity and solvency could not be told apart

A great deal of contemporary argument concerned whether the periphery faced a liquidity problem or a solvency problem. The question was less separable than it appeared, and the reason is mechanical.

Solvency for a sovereign is a statement about the debt path — whether debt as a share of output stabilises, given growth, primary balance and the interest rate.

The interest rate is one of the inputs, and the market sets it. So:

• At a low borrowing cost, a given fiscal position is sustainable.
• At a high borrowing cost, the same fiscal position is not, because interest costs compound faster than the economy grows.
• The borrowing cost reflects the market's judgement about sustainability.

The judgement therefore determines its own truth. A market that believes a sovereign is solvent lends cheaply, which makes it solvent. A market that doubts it demands a premium, which makes the doubt correct. Both are self-consistent equilibria over identical fundamentals.

This has a direct policy implication and it is not obvious. If multiple equilibria exist, the task is not primarily to change the fundamentals — it is to move the market from the bad equilibrium to the good one. A credible commitment to lend at a ceiling can do that without any lending occurring, because the commitment removes the bad equilibrium from the set of possibilities.

It also explains why the crisis appeared to resolve abruptly rather than gradually. A fundamentals problem improves slowly as fundamentals improve. An equilibrium problem ends the moment the alternative equilibrium is removed, which is the same asymmetry the Global Investment Outlook 2008 identifies between credit problems and funding runs.

The arithmetic that made austerity self-undermining

Fiscal consolidation was the condition attached to support. Its results were frequently worse than projected, and the reason is arithmetic rather than ideology.

The objective is the ratio of debt to output. Consolidation reduces the numerator by cutting the deficit. It also reduces the denominator, because fiscal contraction reduces output in the short run.

Whether the ratio falls depends on the fiscal multiplier — how much output falls per unit of consolidation:

• With a low multiplier, output falls little, the numerator improvement dominates, and the ratio falls. This is the assumption under which the programmes were designed.
• With a high multiplier, output falls substantially, and the ratio can rise despite genuine deficit reduction.

Multipliers are not constant, and they are highest under exactly these conditions:

• When interest rates are at their lower bound, monetary policy cannot offset fiscal contraction.
• When the banking system is impaired, credit cannot expand to fill the gap.
• When trading partners consolidate simultaneously, the export channel that normally cushions adjustment is closed — and the periphery consolidated together.
• When the exchange rate cannot move, the most powerful offset of all is unavailable.

Every one of those held. The programmes assumed multipliers estimated from ordinary conditions and applied them to the least ordinary conditions available. The forecasts were not merely optimistic; they were mis-specified in a direction that consolidation itself made worse.

The honest statement of the dilemma: with market access closed, deficits must be financed by someone, and if no one will, contraction is not a choice. The critique is not that consolidation was avoidable but that its pace was calibrated on the wrong parameter, and the error compounded.

What an allocator could act on

Ask what a spread is pricing, not what it is. Peripheral spreads before 2010 priced a belief about collective support, not an assessment of national finances. When a price rests on an unstated assumption, the repricing when it is tested is discontinuous rather than gradual.

Check whether the guarantor is independent of the event. A backstop correlated with what it backs is not a backstop. The doom loop is the general case: whenever the party absorbing a loss is impaired by the same loss, the protection fails when it is needed.

Watch for regulatory treatment creating concentration. Zero risk weights made sovereign debt the rational holding for banks, and the concentration was worst where the sovereign was weakest. A rule that makes an exposure free will produce a lot of it, and that is a position built by the rule rather than by judgement.

Distinguish fundamentals problems from equilibrium problems. Where the borrowing cost feeds back into solvency, multiple equilibria exist and the resolution can be abrupt and cheap — a commitment rather than a transfer. That changes both the timing and the shape of any recovery trade.

Model the multiplier as conditional. Fiscal consolidation at the lower bound, with an impaired banking system, a fixed exchange rate and synchronised partners is a different exercise from consolidation in ordinary conditions. Using an unconditional estimate understates the contraction in exactly the scenario being managed.

Treat contagion as category repricing. Markets sold peripheral debt as a class because the assumption being repriced was categorical. Assessing individual national fundamentals will mislead when the thing being reassessed is the category itself.

What 2010 established

• Eurozone sovereign debt is a credit instrument, because members borrow in a currency they do not control — a fact obscured for a decade by compressed spreads.
• The sovereign-bank doom loop, in which the rescuer is the source of the loss and each turn of the circuit tightens the next.
• Zero risk weights concentrated sovereign exposure in the banks least able to bear it, and produced that concentration by rule rather than by choice.
• Liquidity and solvency are not separable for a sovereign when the borrowing cost determines the debt path, which admits multiple equilibria over identical fundamentals.
• Fiscal multipliers are conditional, and consolidation can raise the debt ratio when the offsets that normally cushion it are all unavailable.