One of investors' greatest fears today is that the world is on the edge of an economic downturn that may be deeper and longer than anything experienced in modern times.

Spurring this concern is a prolonged liquidity crisis that has affected virtually every corner of the global financial system. Lenders have tightened the spigots of credit to a trickle, while individuals and businesses are loath to spend or to take risks. In this environment, doomsday scenarios continue to proliferate: a wide-ranging and catastrophic “unwind,” or deleveraging; a further stock market collapse; or, at best, a global economy that limps along, if not in outright depression, then in a long and bitter recession, perhaps resembling Japan's “lost decade.”

To assess these risks, we studied a number of historical banking crises and their impact on economic growth and financial markets. A clear-eyed analysis of these crises lends important perspective on our current situation, providing us with a better-informed view of the risks and potential opportunities in the market today.

Our study leads us to conclude:

  • Understanding the current crisis requires a long-term global perspective. Severe shocks to a financial system do not occur often; to find comparable events we need to look beyond the post-World War II experience in the United States. By casting a broader net, we can capture and analyze a number of prior episodes with striking similarities to the current environment.

  • Every prior episode followed the pattern of a multi-year credit expansion culminating in a banking system crisis, followed by a corresponding credit contraction. The magnitude of the credit contractions bore a direct relationship to the depth of stock market and economic losses. While it is important not to draw overly fine conclusions from a limited number of episodes, benchmarking the recent credit expansion to past banking crises provides useful comparison: It suggests that the economy may experience an even more severe downturn than most expect, but that United States and global stock market valuations already may reflect such a scenario.

  • The size and scope of the U.S. government actions to ensure continued credit supply and help stave off a collapse in credit demand are unprecedented. Their efficacy will be critical in determining whether the outcome of this crisis is better or worse than those of previous crises. Our analysis suggests that tracking the aggregate amount of credit in the system provides the best measure of the success of government intervention.

An Enduring Cycle

To begin our research, we identified nearly 200 systemic banking crises that we believe share important similarities with the current crisis.1 These episodes, which occurred in more than 100 different countries during the past 130 years, involved the simultaneous failure or near-failure of multiple banks, impairing the ability and willingness of credit intermediaries to channel money from savers to borrowers. It is this characteristic — a severe shock to a country's financial system — that differentiates the current crisis from more typical business cycle-related recessions and bear markets.

In narrowing our sample set to only the most relevant systemic banking crises, we applied the following logic: Arguably, developed-market banking crises during the last 20 years provided the closest parallels, but would have left us with too narrow a sample size; so we included certain emerging-market crises during the same period. We determined their relevance by measuring factors such as the country's wealth and the maturity of its banking system.2 Finally, because of the uniquely large and diversified nature of the U.S. economy, we looked back in history to include five major pre-World War II U.S. crises. This screening process resulted in 15 crises. (See “Comparable Crises,” this page.)

While we recognize that a sample of 15 banking crises is by no means a large data set and that each situation was unique to its time and place in history, there are commonalities across the events. Indeed, each banking crisis we studied had at its core the hallmarks of a classic credit cycle: A self-reinforcing but unsustainable relationship blossoms between borrowers emboldened to increase their indebtedness and lenders prepared to stretch their balance sheets to meet the escalating demand for credit.

We looked at the run-up in lending before each crisis and the subsequent deleveraging, as measured by the loan-to-deposit ratios of banks in each of the 15 examples. The ratio in each case increased for a varying number of years prior to the crisis, peaked around the crisis, and then decreased for years afterward. Underneath the rise in the amount of loans extended per dollar of deposits lay deteriorating underwriting standards and the willingness of lenders to rely more heavily on riskier sources of funding.

While the “unwind” trigger was different every time (when we could identify one at all), the onset of a crisis not surprisingly tended to mark a peak in lending activity. A crisis event usually coincided with loan impairments and defaults that in turn damaged bank capital and the ability to lend. Simultaneously, changes in sentiment reduced the desire to lend and, in many cases, to borrow.

In virtually all cases, the disinclination of banks to lend created ripple effects that threatened to bring the wheels of commerce to a near standstill. Most of the crises we studied share many similarities with today's situation, including major real estate price declines, equity bear markets and heightened equity volatility. The fact that these traits are shared across the episodes in our sample set helps establish them as reasonable benchmarks against which we can compare our current situation.


A starting point for analyzing the implications of a systemic banking crisis for economic growth and stock market performance is simply to observe the average outcomes of past crises. What we find is that the average peak-to-trough downturns in both the real economy and the financial markets were indeed severe. (See “Benchmarking the Pain,” this page.) On average, corporate earnings turned negative, GDP fell 10 percent, and the unemployment rate rose by more than seven percentage points. The worst of these crises, the Great Depression, saw GDP contract by more than 30 percent and unemployment spike by 21 percentage points. That's an outcome more than three times the magnitude of the average crisis in our sample.

Local stock markets in the affected countries during these crises saw average real price declines of 52 percent from peak to trough over a 2.6-year period. On average, they took 9.2 years to regain their peaks. Again, the Great Depression took the dubious honor of being the worst, posting an inflation-adjusted stock decline of 81 percent, with a painful 26.5 years to recover.

What might a simplistic application of these data imply for the current crisis?

  • If today's crisis turns out to be “average,” then we might expect a dismal real economy but a stock market with a peak-to-trough decline equal to what already has occurred.

  • The typical recovery period of nine or so years suggests that prices could return to their inflation-adjusted peaks by the year 2016. While that may sound like a long time to wait, a recovery by then would imply eight years of real double-digit stock returns.

But can we say this is an average crisis? And is the economic impact likely to be in line with past banking crises?

Consequences of Deleveraging

Historically, the magnitude of a credit expansion before a crisis has been reflected in the ensuing credit contraction. In plain English: the bigger the boom, the bigger the bust. This relationship becomes meaningful when we consider that the magnitude of credit contractions in the aftermath of banking crises has been related to the depth of economic and stock market declines.

In our sample of crises, outstanding bank loans tended to decline in the ensuing contraction by an amount roughly equal to real loan growth in the final three years of expansion. (See “Run-ups Forecast Contractions,” this page.) This three-year window makes intuitive sense based on deteriorating lending standards in the run-ups to nearly all the crises we studied. Anecdotal examples include Japanese banks accepting artwork as loan collateral in the late 1980s and U.S. lenders issuing NINJA (No Income, No Job, No Assets) mortgages during the recent cycle.

One challenge in applying this relationship to the current cycle lies in tallying aggregate loans outstanding. During much of the historical record and in many countries still, banks provide the vast majority of outstanding private credit. In the United States, however, non-bank lending plays an important role, and never has that role been more prominent than in the recent cycle with the emergence of the so-called shadow banking system. This system includes a host of largely unregulated entities, including private finance companies and special purpose vehicles set up by banks to move loans off their balance sheets.

To capture loans enabled by this shadow system as well as by more traditional forms of nonbank lending such as the corporate bond market, we added all borrowing by households and non-financial businesses to derive our estimate of total U.S. loans outstanding. (See “How Big Was the U.S. Run-up, Really?” p. 35.)3

Under this definition of loans outstanding, the inflation-adjusted credit growth in the United States during the three years leading up to the first quarter 2008 peak totaled 20 percent.

If the U.S. credit market undergoes a contraction that follows the pattern of prior crises — and as of the third quarter of 2008 this contraction had barely begun — we would expect $5 trillion (20 percent) of the $25 trillion peak loans outstanding to vanish through roll-offs, write-downs, and defaults in the coming years, even as new loans are made.

What would a contraction of this magnitude imply for the U.S. stock market? With the exception of Japan in the 1990s (see “Has the U.S. Equity Market Hit Bottom?” this page), the magnitude of the stock market declines in our sample bore a reasonable relationship to the magnitude of the ultimate declines in loans outstanding. This relationship reflects the higher risk premiums and slower profit growth generally associated with contracting credit.

If loans in the United States in fact contract 20 percent and the historical relationship to equity prices holds, we would expect a nearly 50 percent real decline in equity prices. That is almost exactly the loss experienced by the S&P 500 Index from its October 2007 peak through its recent lows.

Of course, credit contractions ultimately impact equity markets because of their effects on the real economy. What would a 20 percent decline in outstanding credit imply for U.S. GDP? Following other systemic banking crises, a 20 percent contraction in the domestic credit market would be consistent with full-year U.S. GDP shrinking by 7 percent — an almost unimaginable figure. (See “Forecasting U.S. Economy Shrinkage,” this page.) This would represent an outcome unprecedented in the post-World War II era and far exceeds economists' consensus forecast. Indeed, we recognize there are legitimate reasons why this relationship may not hold as strongly in today's economy.

Of the crises we examined, those occurring before World War II experienced the greatest economic contraction, on average. Modern developed economies have tended to be less severely affected by crisis-induced credit crunches, in part reflecting the powerful role now played by government fiscal and monetary authorities in responding to financial and economic emergencies. (See “Failure and Success,” p. 39.) Such involvement is likely to be an important factor in how this crisis plays out.

Government Intervention

Will the current crisis follow the patterns of past systemic banking crises? If the United States can avoid a 20 percent deleveraging cycle, then perhaps the equity market has overshot and GDP need not fall by 7 percent.

The main reason for optimism, in our opinion, is the unprecedented government response to date.

Even before the current crisis, government entities played a key role in the funding channels that linked suppliers of credit to borrowers. Indeed, government programs implicitly or explicitly supported about 50 percent of financial intermediaries' deposit and credit market liabilities, thanks to FDIC deposit insurance, loans made to financial institutions by the Federal Home Loan Banks (FHLBs), and government-sponsored enterprises (GSEs) such as Fannie Mae and Freddie Mac. In the wake of the crisis, the government has dramatically expanded existing programs to ensure the stability of critical sources of credit. Fannie and Freddie have been placed into conservatorship, effectively nationalizing their operations, and the FDIC has expanded its deposit insurance guarantee by about $1 trillion.

Furthermore, between the U.S. Treasury, the FDIC and the Federal Reserve, an alphabet soup of entirely new programs has been introduced, extending the government's buttressing of the lending markets well beyond the traditional avenues of deposit insurance and agency backing. (See “Here Comes the Fed,” p. 38.) The new programs loosely target previously unsupported segments of the funding liabilities of the U.S. financial system. We estimate that announced efforts to shore up the commercial paper and asset-backed securities markets, as well as financial institutions' bond market liabilities (in which partial relief has come from Troubled Asset Relief Program (TARP)-funded equity recapitalization), could provide a minimum of $3 trillion in additional credit market support, on top of the $1 trillion in expanded FDIC insurance.

In other words, through new, expanded and existing programs, the government has added about $4 trillion in lending market support, effectively backstopping more than 70 percent of the deposit and debt liabilities of the financial system's primary credit intermediaries. And while most of this support comes in the form of broad guarantees, the impact of various direct interventions already can be seen on the Fed's balance sheet, which has ballooned from a little more than $800 billion to more than $2 trillion and is likely to hit $3 trillion by spring.

What's more, should the situation continue to worsen, the government authorities are clearly prepared, as Federal Reserve chairman Ben Bernanke recently declared, “to do what we need to do to keep the U.S. credit system working and to try and create a recovery in the financial system.”4

Are such efforts likely to meet their goal of breaking the destructive cycle of forced deleveraging characteristic of past crises?

In our view, the depth and breadth of government intervention mark a promising break from many previous episodes, when the response was often either too tepid or too narrowly targeted, or both.

But we recognize it's too early to sound the “all clear.” Indeed, some retrenchment on the part of lenders will be both necessary and healthy. And in fact, the programs enacted to date do not appear geared to prevent deleveraging altogether. Rather, they aim, again in Bernanke's words, to slow it by “allowing a more orderly process of asset sales and the necessary deleveraging by financial institutions.”5

Regardless of the success of government efforts to stabilize the supply of credit, the ability and willingness to lend does not necessarily lead to a desire to borrow. Perhaps not surprisingly, we recently have witnessed aggregate residential mortgage debt — and with it total household debt — decline for the first time in the post-World War II era.6 Should the recession deepen, it's possible that uncertainty and pessimism about the future could become so acute that firms and households simply refuse to borrow even when credit is available. This raises the specter of the so-called paradox of thrift, when too much saving (that is to say, too little spending) actually threatens economic growth. Additional fiscal stimulus from Washington is meant to help address this risk by directly stimulating demand and by indirectly buoying consumer confidence.

In the near term, it appears the biggest risk to mitigating the downside of the current deleveraging cycle is that fiscal stimulus may not stave off a collapse in credit demand. Over the longer term, the key risk may come from the law of unintended consequences: While the size and character of the current crisis are not unprecedented, the magnitude and diversity of the government's response are. Keeping a watchful eye on the potential side effects stemming from these monetary and fiscal support programs will help us gauge the success of the government's efforts.

Looking Ahead

The current financial crisis is different from other post-World War II U.S. recessions and bear markets. At its heart is an enormous disruption to the financial system rather than a typical business cycle or exogenous shock. But this situation is not that different in character or magnitude from past systemic banking crises, from both qualitative and quantitative perspectives.

While no one can predict what lies ahead, the similarities of recent events to past credit crises suggest that we can look to credit metrics to understand and monitor the current situation. Applying these to the patterns of similar past crises points to a real economy that faces a deep recession but a stock market whose prices may already anticipate such an outcome.

Furthermore, if today's government remedies prove more effective than those of the past, we may navigate this storm with a less painful deleveraging than might otherwise be expected.

— Bernstein Global Wealth Management, a unit of AllianceBernstein L.P., does not offer tax, legal or accounting advice. In considering this material, you should discuss your individual circumstances with professionals in those areas before making any decisions.


  1. Our original list came from events categorized as systemic banking crises in these studies: Luc Laeven and Fabian Valencia, “Systemic Banking Crises: A New Database,” International Monetary Fund (IMF), November 2008; and Carmen Reinhart and Kenneth Rogoff, “This Time Is Different: A Panoramic View of Eight Centuries of Banking Crises,” National Bureau of Economic Research (NBER), April 2008.
  2. Regarding the level of wealth, we limited the sample set to those countries with a real gross domestic product (GDP) per capita that was greater than 20 percent of that of the United States. For the maturity of the banking system, we based our analysis on the credit market regulation score of the Economic Freedom of the World Index, which is co-published by the Cato Institute in Washington, D.C., the Fraser Institute in Vancouver and more than 70 think tanks around the world. That component of the index measures bank deposits held in privately held banks, foreign bank competition, the percentage of domestic credit consumed by the private sector, and the level of intervention in credit markets. We included only those past crises in which the country involved had a score on the index that was greater than 70 percent of the U.S. score as of the late 1980s.
  3. Categories of household and business debt outstanding include residential and commercial mortgages (including mortgage debt owed by the financial sector); consumer credit; nonfinancial corporate bonds and commercial paper; bank loans to nonfinancial and financial sectors (excluding loans from the Federal Reserve to the financial institutions); and other loans and advances to the non-financial sector (excluding U.S. government loans to state and local governments and government-sponsored enterprises).
  4. In response to questioning after Congressional testimony, Nov. 18, 2008.
  5. U.S. Federal Reserve chairman Ben Bernanke's prepared remarks at the Greater Austin Chamber of Commerce, Dec. 1, 2008.
  6. Though by less than 1 percent through the third quarter of 2008, per the seasonally adjusted Fed Flow of Funds on an annualized basis.

Jon Ruff, far left, is a director and Vincent L. Childers is a research analyst in the Wealth Management Group at Bernstein Global Wealth Management in New York

Comparable Crises

The current situation is not without precedent

We looked beyond U.S. bear markets and recessions to analyze systemic financial crises around the globe and over time, narrowing our sample set to 15 episodes with relatively common characteristics.

Group I: Developed Countries, Post-WWII

Relative Wealth* Banking System Sophistication**
1 United States — late 1980s 100% 9.5
2 Japan — 1990s 85 8.0
3 Norway — late 1980s 80 8.5
4 Sweden — early 1990s 76 8.4
5 Finland — late 1980s 73 9.3
Average 83% 8.7
Median 80% 8.5

Group II: Open Developing Countries, 1990+

Relative Wealth* Banking System Sophistication**
1 South Korea — late 1990s 50% 6.9
2 Argentina — early 2000s 30 7.7
3 Thailand — late 1990s 27 7.3
4 Turkey — early 2000s 23 7.6
5 Colombia — late 1990s 21 7.6
Average 30% 7.4
Median 27% 7.6

Group III: United States, 1870-WWII

Relative Wealth*
1 United States — 1930s 100%
2 United States — mid-1910s 100
3 United States — early 1900s 100
4 United States — 1890s 97
5 United States — 1870s 76
Average 95%
Median 100%

*Relative Wealth=Real gross domestic product (GDP) per capita of country divided by the greater of U.S. or U.K. real GDP per capita

**Banking System Sophistication=For the maturity of the banking system, we based our analysis on the credit market regulation score of the Economic Freedom of the World Index, published by the Fraser Institute. That component of the index measures bank deposits held in privately held banks, foreign bank competition, the percentage of domestic credit consumed by the private sector and the level of intervention in credit markets. We included only those past crises in which the country involved had a score on the index that was greater than 70 percent of the U.S. score as of the late 1980s
— AllianceBernstein, 2009

Benchmarking the Pain

The economic and market impact of the crises can be measured

Drops in both economic activity and the stock market were, on average, severe, but the Great Depression was an outlier in terms of GDP reduction.

Averages of past Crises

What Happens to Real Economy What Happens to Real Equity Prices
Corporate Earnings (EPS) GDP* Unemployment (Percentage Points) Peak to Trough Drop Peak to Trough (Years) Years to Regain Peak
Developed -137% -4.4% Up 5.6 -53% 4.1 7.0
Emerging -213 -11.6 Up 4.5 -60 1.0 5.4
Pre-WWII -42 -15.1 Up 11.4 -45 2.8 15.2
Great Depression** -67 -30.8 Up 21.0 -81 2.8 26.5
Average of All -130% -10.0% Up 7.2 -52% 2.6 9.2
*Real, per capita, annualized
**Also included in Pre-WWII data
— Global Financial Data, IMF, Morgan
Stanley Capital International, NBER, OECD &
AllianceBernstein, 2009

Failure and Success

Three very different experiences provide insight into the role government can play in resolving — or prolonging — a systemic financial crisis.

The United States in the 1930s

The Great Depression differed from other crises in many ways, but perhaps the most insidious was the 25 percent decline in the consumer price level seen from 1929 to 1933 in the United States.1 Such a deflation distorts many economic relationships, not least between lenders and borrowers as the real value of debt rises. One can see the heightened pain for a highly levered economy from extreme deflation in the experience of the United States in the 1930s, which fell well outside the normal relationship between real loan and real gross domestic product (GDP) contraction. (See “Forecasting U.S. Economy Shrinkage,” p. 36.)

The most significant reason for the magnitude of this price decline was continued adherence to the gold standard. The more quickly countries left the gold standard in the 1930s, the more quickly their economies “reflated.” Under the gold standard, governments were virtually powerless to influence the money supply and therefore to undertake any action that would reflate their ailing economies. Today, governments strongly influence the money supply, so we view a repeat of this aspect of the Great Depression as a virtual impossibility.

Japan in the 1990s

The Japanese government's failure to address its banking problems for most of the 1990s impaired its ability to stimulate the economy and resulted in the maintenance of an abnormally high level of loans outstanding. Rather than write down dubious loans, Japanese banks restructured them and so kept credit flowing to weak and otherwise insolvent “zombie” firms. The survival of these zombie firms pressured the profits of healthy firms and hence weighed heavily on equity prices. The repercussions of such distorted credit flows meant that Japan in the 1990s fell outside the normal relationship between real loan and real equity price contraction. Structural disincentives to write down bad loans do not exist in the United States, and as such we would expect a tighter relationship between equity declines and eventual loan declines than that experienced by Japan. (See “Has the U.S. Equity Market Hit Botton?” p. 35.)

Sweden in the Early 1990s

The Swedish experience presents a more positive outcome than either the United States in the 1930s or Japan in the 1990s.

Financial reforms in the 1980s led to a credit boom and, with it, overheated consumer spending and real estate prices. A tighter monetary and fiscal environment combined with a currency crisis to reverse the boom by the early 1990s.

The resulting carnage in the banking system ended only after the government had nationalized 22 percent of all banking-system assets.2 The government then transferred the bad assets of the failed banks to two specially created companies for disposal. These companies operated very effectively and completed their task ahead of schedule. The benefit of this approach was that GDP and equity declines were less than would be expected given the magnitude of loan contraction. (See “Forecasting U.S. Economy Shrinkage,” p. 36 and “Has the U.S. Equity Market Hit Botton?” p. 35.)

Furthermore, equity prices returned to their prior peak much faster — 2.2 years — than in most other crises.


  1. In the United States, the Long Depression of the 1870s following the Panic of 1873 also exhibited a 25 percent price decline, but because we do not have reliable gross domestic product (GDP) data, it is not in our loan-GDP analysis.
  2. Federal Reserve Bank of Cleveland, “On the Resolution of Financial Crises: The Swedish Experience,” O. Emre Ergungor, June 2007.


War of the Rose — Gregory Crewdson's Digital C-Print “Untitled,” from his 2001 series “Twilight,” belongs to Brandeis University's Rose Art Museum. The university's trustees approved a plan to close the museum and sell its entire collection. See “Driven To Defy Donor Intent?” p. 58 for story. Photograph, courtesy of the artist and Luhring Augstine, a New York-based art gallery.