Banking crises around the world

Having rejected Henry Paulson’s rescue plan, it’s
not clear what Congress—or those in the broad population
opposed to a “bailout”—propose to do to
keep the financial system from imploding. But a database of systemic
banking crises recently assembled by IMF economists Luc Laevan and
Fabian Valencia (  )
provides a useful map of how crises play out and what does and
doesn’t work.

Laevan and Valencia identify 124 systemic banking crises between 1970
and 2007, and assemble detailed information on 42 of them, representing
37 countries. (Some countries, like Argentina, appear multiple times.)

In almost every case, governments took active measures to mitigate the
crisis, so there is no real test of whether rescue schemes actually
work; no politician seems willing to face the consequences of letting
the chips fall where they may. But the work of Laevan and Valencia does
offer some guidance as to what works best.

Dithering Costs
One crucial lesson stands out: speed matters. This is obvious to anyone
who followed Japan’s dithering in the 1990s; standing aside
and hoping the problem goes away is not a good idea. Relatedly,
“forbearance”—regulatory indulgence, such
as permitting insolvent banks to continue in business—does
not work, as has been established in earlier research. As the authors
say, “The typical result of forbearance is a deeper hole in
the net worth of banks, crippling tax burdens to finance bank bailouts,
and even more severe credit supply contraction and economic decline
than would have occurred in the absence of forbearance.” This
suggests that suspending mark-to-market requirements is not a good idea.

Since forbearance does not work, some sort of systemic restructuring is
a key component of almost every banking crisis, meaning forced
closures, mergers, and nationalizations. Shareholders frequently lose
money in systemic restructuring, often lots of it, and are even forced
to inject fresh capital. The creation of asset management companies to
handle distressed assets is a frequent feature of restructurings, but
they do not appear to be terribly successful. More successful are
recapitalizations using public money (which can often be partly or even
fully recouped through privatization after the crisis passes); recaps
seem to result in smaller hits to GDP. But they’re not cheap:
they average 6% of GDP, which for the U.S. would be about $850 billion.

Total fiscal costs, net of eventual asset recoveries, average 13% of
GDP (over $1.8 trillion for the U.S.); the average recovery of public
outlays is around 18% of the gross outlay.

But those who don’t want to spend that kind of taxpayer money
should consider this: Laevan and Valencia find that “[t]here
appears to be a negative correlation between output losses and fiscal
costs, suggesting that the cost of a crisis is paid either through
fiscal costs or larger output losses.” And if the economy
goes into the tank, government revenues take a big hit, so
what’s saved on the expenditure side could well be lost on
the revenue side.

Oh, and about half the countries that have experienced crises have had
some form of deposit insurance. So merely expanding the
FDIC’s coverage is not likely to do the trick—and,
in any case, it’s going to be hard to escape the huge expense
of a systemic recapitalization, though using the FDIC might simplify
the politics of the rescue.

(A note on the politics of the rescue: an ABC poll shows the public to
be far more worried about the economic consequences of the
bailout’s defeat than Congress seems to be. There’s
not a lot of enthusiasm for what’s seen as handing money over
to Wall Street—but if properly structured and sold, say with
more cost recovery prospects for the government, more relief for
debtors, a rescue is not as unpopular as some would have it.)

Relevant Examples
Most of the countries in the Laevan/Valencia database are in the
developing world, and are of questionable relevance to the U.S. But TLR
has taken a closer look at four countries that offer more relevant
models: Japan, Korea, Norway, and Sweden. Some major stats for the four and the U.S. are in the table at the end of this entry, as are  graphs of some important indicators.

Sweden, now widely seen as a model of swift, bold action, kept its
ultimate fiscal costs relatively low—3.6% of GDP at first,
almost all of which was recovered through stock and asset
sales—but was unable to avoid a deep recession. At the other
end of the spectrum, Japan, the model of foot-dragging half-measures,
saved no money through its procrastination; its fiscal outlay was 24%
of GDP, almost none of which was recovered. And it was unable to avoid

Note, though, that some of the worried talk surrounding the financial
market impact of bank bailouts looks misplaced, at least on these
models. Three years after the outbreak of crisis, inflation was lower
and stock prices higher in all four countries, and government bond
yields were lower in all but Japan. It’s likely that the
deflationary effects of a credit crunch outweigh the inflationary
effects of debt finance.

Although the U.S. in 2007 had a lot in common with other countries on
the brink of a banking crisis, one thing stands out: the depth of the
current account deficit. Of the four comparison countries, only Korea
comes close to the U.S. level of red ink. The unweighted average
current account deficit of the 42 countries in the Laevan/Valencia
database was 3.9% of GDP—compared with 6.2% for the U.S. That
suggests that the U.S. has more to deal with than just resolving a
banking crisis.

A Better Bailout
So, with the modified Paulson plan dead for now, what might a better
bailout scheme look like in light of the Laevan/Valencia historical

First, it must be adopted quickly. Perhaps operating through the FDIC
would be a way to accomplish that, though the FDIC will almost
certainly need to have its coffers copiously refilled.

Second, forbearance would be a bad idea; it does no one any good not to
face reality.

Third, purchasing bad assets and turning them over to an asset
management corporation is not a promising strategy.

Fourth, recapitalizing the banks should be the heart of any policy; as
the authors say, it should be selective, meaning supporting those
institutions with hope of revival, and letting the terminal go down.

And fifth, targeted relief for distressed debtors, supported with
public funds, has also shown success in earlier banking crises, and
should be part of any rescue scheme in the U.S. as well.
Crises like this are manageable. They’re expensive and
painful to resolve, but even more expensive and painful when left to

—Philippa Dunne & Doug Henwood


banking crises: some stats

Japan Korea Norway Sweden U.S.
start 1997 1997 1991 1991 2007
fiscal cost
gross 24.0% 31.2% 2.7% 3.6%
net 23.9% 23.2% 0.6% 0.2%
output loss 17.6% 50.1% 0.0% 30.6%  
minimum growth -2.0% -6.9% 2.8% -1.2%
fiscal balance -5.1% 0.2% 2.5% 3.4% -2.6%
public debt 100.5% 8.8% 28.9% 60.1%
inflation 0.6% 4.9% 4.4% 10.9% 2.6%
GDP growth 2.8% 7.0% 1.9% 1.0% 2.9%
current account 1.4% -4.1% 2.5% -2.6% -6.2%
recap costs
gross 6.6% 19.3% 2.6% 1.9%
net 6.5% 15.8% 0.6% 1.5%
three years later
CPI -2.5% -2.1% -2.0% -7.3%
gov bonds 0.1% -3.2% -2.7% -1.0%
stocks 10.9% 12.1% 53.0% 41.0%
employment -1.7% 0.2% 1.3% -10.6%

All percentage figures except inflation, GDP growth, bond yields, stock prices, and employment are percent of GDP. Gross fiscal cost is total outlays for banking system support; net is after equity and asset sales. Output loss is total deviation from trend growth rate in GDP from the crisis year through three years after crisis onset, expressed as a percent of trend GDP. Minimum growth is the lowest level of GDP growth reached during the crisis. Pre-crisis stats are for the year before the onset of the crisis. Recap costs are costs of cash, equity, or debt injections or asset purchases to recapitalize the banking system; net is after recovery of these costs through asset sales and the like. Stats labeled “three years later” are changes in indicators three years after crisis onset. “Gov bonds” are bond yields reported by the IMF in its International Financial Statistics database. “Stocks” are the based on the stock indexes published in IFS. Stats in the first eleven rows come from “Systemic Banking Crises: A New Database,” by Luc Laeven and Fabian Valencia (IMF Working Paper 08/224) and the associated spreadsheets available from

. The next four rows are computed by TLR from the IFS database.