UK Investment Report 2008 — The Contingent Liability Nobody Booked

The UK entered 2008 with a specific structural vulnerability that had been visible for years and read as a strength.

Its banking sector's assets were several multiples of national output. This had been treated as evidence of a successful, globally competitive financial industry — a major export earner and employer.

It is also, unavoidably, a contingent liability of the state. A banking system's deposits are guaranteed by the government; its failure is a systemic event the government must prevent. So the size of the banking sector relative to the economy determines how large a commitment the state has made without recording it.

The arithmetic is unforgiving. A banking system whose assets are a few times GDP means a loss of a small percentage of those assets is a very large percentage of GDP. The sovereign's capacity to absorb it depends on a ratio nobody was monitoring as a fiscal variable.

What actually broke was funding, not assets. The first UK institution to fail did so because it funded long-dated mortgages with short-dated wholesale borrowing, per the mechanism in the Global Investment Outlook 2008. When that market closed, it could not refinance — while its mortgage book was still performing. It was a liquidity failure at a solvent-looking institution, which is why it surprised nearly everyone watching credit quality.

Two things worked in the UK's favour:

The currency floated and fell sharply, which is a real adjustment mechanism. It improved competitiveness immediately and automatically, with no negotiation and no legislation — the option the Global Investment Outlook 2010 shows euro-area members did not have.

And the policy response was decisive on a specific point: rather than buying bad assets, the state injected capital directly in exchange for equity, on terms designed to be unattractive enough that banks would want to repay. This template was adopted internationally within weeks and is the period's most consequential British policy contribution.

When a banking sector is a fiscal position

The framing deserves development because it changes what should be monitored.

A country's banking sector is conventionally analysed as an industry — its profitability, employment, contribution to output and exports.

It is simultaneously a set of state commitments:

• Deposits are guaranteed up to a limit, explicitly.
• Systemically important institutions cannot be allowed to fail disorderly, implicitly but reliably — the implicit guarantee structure the Asia-Pacific Investment Report 2013 examines.
• And the payment system must keep functioning, which means the institutions operating it must be supported.

So the state has written an option whose value depends on the banking system's size and riskiness, and which appears in no budget.

The critical ratio is banking assets to GDP, because it determines feasibility:

• A banking system roughly the size of the economy produces losses the state can absorb. A ten percent loss on assets equal to one year's output is large but survivable.
• A banking system several times the economy produces losses that may exceed what the state can credibly cover. The same ten percent loss is now several times what a stressed sovereign can raise.

A large banking sector is an export industry in good times and a sovereign exposure in bad ones. The ratio that determines which it is on any given day is published and was not, before 2008, treated as a fiscal statistic.

The extreme demonstrations came from smaller economies whose banking systems had grown to many multiples of national output. There, the guarantee was arithmetically impossible — the state could not have covered the losses at any plausible tax rate, so the guarantee was void before it was tested.

The UK's position was less extreme and the same in kind, and the response — forcing recapitalisation quickly, before the losses grew — was partly a recognition that the exposure had a limit.

The durable policy consequence was that bank size relative to the home economy became a supervisory concern in its own right, and the resolution frameworks developed afterwards were designed specifically so that failure need not mean a sovereign commitment.

Funding, not assets

The sequence of UK failures follows funding structure almost exactly, and it is worth being precise because the lesson is widely misremembered as being about lending standards.

The institution that failed first had, by the standards of the period, a reasonable asset book. Its mortgages were mostly performing and mostly domestic.

Its funding was the problem:

• A relatively small deposit base relative to its lending.
• The gap funded by wholesale markets — borrowing from other financial institutions and through securitisation, at short maturities.
• Which meant continuous refinancing. Every few months, a large share of its funding had to be replaced.

When wholesale markets closed in 2007, the refinancing failed — for reasons entirely unrelated to its mortgages. The lenders were not making a judgment about UK housing. They were conserving their own liquidity.

The retail run that followed was a consequence, not a cause. Depositors queued after the institution had already sought emergency support, which is the correct order and is usually reported backwards.

The general principle, which the archive documents in five separate settings:

Institutions with maturity mismatches fail on the funding side. The assets are usually still performing when the institution dies. Screening credit quality will not detect this, because credit quality is not what breaks.

The screening implication is straightforward and was available in public accounts:

• Loan-to-deposit ratio — how much lending is funded by stable deposits versus borrowed funding.
• The share of funding maturing within a year, which is the refinancing requirement.
• And the concentration of funding sources, since a small number of wholesale lenders is a fragile base.

All three were disclosed in annual reports before the failures. The Global Investment Outlook 2008, Asia-Pacific Investment Report 2013 and US Venture Capital Report 2008 describe the identical structure in three other settings.

The currency did work that policy could not

Sterling fell sharply against major currencies through 2008, and the depreciation is best understood as an automatic adjustment rather than a symptom.

What a large depreciation does for a stressed economy:

• Exports become cheaper abroad, supporting demand from external sources when domestic demand is contracting.
• Imports become more expensive, shifting spending toward domestic production.
• Foreign earnings of domestic companies rise when translated home, supporting profits and equity values in domestic currency terms.
• And it happens immediately, without legislation, negotiation or an institution's approval.

The cost is real: imported goods cost more, which reduces real incomes, and inflation rises. The adjustment is a transfer from consumers to producers of tradeable goods, which is the point — it moves resources toward the sectors that can earn foreign income.

The comparison with the euro area is the analytically important part. The Global Investment Outlook 2010 and 2011 describe economies that needed exactly this adjustment and could not have it. Their alternative was internal devaluation — falling wages and prices — which achieves a similar relative price change but:

• Takes years rather than weeks.
• Raises the real value of existing debt while it happens, worsening the balance sheet problem.
• And requires wage reductions that are individually negotiated and socially contested, rather than a price change that happens to everyone at once without anyone deciding.

A floating currency does in a fortnight, invisibly, what internal devaluation does over five years, visibly and painfully. The economic destination is similar. The path is not remotely comparable.

This is the strongest available argument for monetary independence, and it is a genuine one — with the corresponding cost that a floating currency also imports volatility and can overshoot. The UK Investment Report 2016 examines the same mechanism under a different shock.

The template that was exported

The UK's recapitalisation approach was adopted internationally with unusual speed, and its design deserves attention because the alternatives were worse for identifiable reasons.

The problem to solve: banks had insufficient capital against expected losses. Undercapitalised banks do not lend, which turns a financial problem into a recession, and they are vulnerable to funding runs.

Option one: buy the bad assets. The state purchases troubled securities, removing them from bank balance sheets.

• The difficulty is price. Buy at market and banks book losses they cannot absorb — no help. Buy above market and the state overpays, which is a transfer with no accountability and is politically indefensible.
• And it is slow, requiring valuation of complex assets one at a time.

Option two: guarantee the assets. The state insures the losses.

• This helps but does not add capital, so the bank is still constrained.
• And it leaves the exposure with the state open-ended, which is the contingent liability problem again.

Option three: inject capital directly in exchange for equity.

• It is fast, requiring no asset valuation.
• It addresses the actual constraint, since capital is what was short.
• The state receives an ownership stake, so taxpayers participate in any recovery.
• And it can be forced. Regulators can require banks to reach a capital ratio and offer state capital to those that cannot raise it privately — which avoids the stigma problem, where only the weakest banks accept help and thereby identify themselves.

The last point is subtle and was decisive. A voluntary scheme fails because accepting help signals weakness. A mandatory scheme applied to all major institutions removes the signal entirely — this is the same equilibrium logic the Global Investment Outlook 2012 describes.

The design details that made it work: the capital was expensive, with restrictions attached, so banks wanted to repay it as soon as they could raise private capital. The state's involvement was therefore temporary by construction rather than by promise.

What the crisis actually cost

The fiscal accounting of the period is frequently misstated, and the correct version matters for assessing similar interventions.

The direct rescue cost was modest and partly recovered. Capital injected in exchange for equity is an asset, not an expense. Some stakes were sold at a profit, some at a loss, and the net direct cost was a fraction of the headline sums committed.

The real cost was elsewhere and much larger:

Lost output. The recession reduced GDP, and — this is the crucial part — the level of output did not return to its previous trend. An economy that grows at the same rate from a permanently lower base has permanently lower output than the counterfactual. The cumulative loss dwarfs the rescue cost.

The automatic fiscal deterioration. A recession reduces tax receipts and increases spending on unemployment support without any policy decision. This is most of the deficit increase, and it is the mechanical consequence of the downturn rather than of the rescue.

And a permanently higher debt path. Debt accumulated during the contraction does not disappear when growth resumes; it constrains the fiscal position for a decade or more, which is the balance sheet repair the Global Investment Outlook 2010 describes operating on the public sector.

The rescue was cheap and the crisis was ruinous. Conflating the two produces the wrong conclusion about whether the rescue was worth doing.

The analytical point generalises: when assessing any financial intervention, the counterfactual is not "no cost" but "the cost of the crisis without intervention." A rescue that appears expensive against zero may be very cheap against the alternative — and the alternative is what the Global Investment Outlook 2008 describes happening in the institutions that were not rescued.

What an allocator could act on

Track banking assets relative to home GDP as a sovereign metric. It determines whether the state can credibly stand behind its banking system, and it is published without being treated as a fiscal statistic.

Screen funding structure, not credit quality, for institutional fragility. Loan-to-deposit ratio, share of funding maturing within a year, and funding source concentration are all disclosed and all predicted the sequence of failures.

Value monetary independence by the adjustment it permits. A floating currency delivers in weeks what internal devaluation takes years to achieve, without raising the real value of existing debt in the process.

Note that mandatory schemes solve stigma problems that voluntary ones cannot. When accepting help signals weakness, only universal application removes the signal.

Separate rescue cost from crisis cost. The direct fiscal outlay was small and partly recovered; the lost output and permanently higher debt path were the real bill, and they follow from the downturn rather than the intervention.

Model output losses as level effects, not growth effects. An economy that never regains its previous trend carries the loss forever, which is why the cumulative cost so far exceeds any headline rescue figure.

What 2008 established

• A banking sector large relative to its economy is an unbooked fiscal commitment, and the ratio determines its credibility.
• Funding structure, not asset quality, determined the failure sequence — the institutions died solvent on paper.
• A floating currency provided rapid, automatic adjustment unavailable inside a monetary union.
• Forced recapitalisation in exchange for equity became the international template, largely because mandatory application removed the stigma problem.
• The crisis cost far exceeded the rescue cost, principally through a permanent shortfall in the level of output.