- 14 -

THE AFTERMATH OF FINANCIAL CRISES

In the preceding chapter we presented a historical analysis comparing the run-up to the 2007 U.S. subprime financial crisis with the antecedents of other banking crises in advanced economies since World War II. We showed that standard indicators for the United States, such as asset price inflation, rising leverage, large sustained current account deficits, and a slowing trajectory of economic growth, exhibited virtually all the signs of a country on the verge of a financial crisis—indeed, a severe one. In this chapter we engage in a similar comparative historical analysis focused on the aftermath of systemic banking crises. Obviously, as events unfold, the aftermath of the U.S. financial crisis may prove better or worse than the benchmarks laid out here. Nevertheless, the approach is valuable in itself, because in analyzing extreme shocks such as those affecting the U.S. economy and the world economy at the time of this writing, standard macroeconomic models calibrated to statistically “normal” growth periods may be of little use.

In the previous chapter we deliberately excluded emerging market countries from the comparison set in order not to appear to engage in hyperbole. After all, the United States is a highly sophisticated global financial center. What can advanced economies possibly have in common with emerging markets when it comes to banking crises? In fact, as we showed in chapter 10, the antecedents and aftermath of banking crises in rich countries and in emerging markets have a surprising amount in common. They share broadly similar patterns in housing and equity prices, unemployment, government revenues, and debt. Furthermore, the frequency or incidence of crises does not differ much historically, even if comparisons are limited to the post-World War II period (provided that the ongoing global financial crisis of the late 2000s is taken into account). Thus, in this chapter, as we turn to characterizing the aftermath of severe financial crises, we include a number of recent emerging market cases so as to expand the relevant set of comparators.¹

Broadly speaking, financial crises are protracted affairs. More often than not, the aftermath of severe financial crises share three characteristics:

•  First, asset market collapses are deep and prolonged. Declines in real housing prices average 35 percent stretched out over six years, whereas equity price collapses average 56 percent over a downturn of about three and a half years.

•  Second, the aftermath of banking crises is associated with profound declines in output and employment. The unemployment rate rises an average of 7 percentage points during the down phase of the cycle, which lasts on average more than four years. Output falls (from peak to trough) more than 9 percent on average, although the duration of the downturn, averaging roughly two years, is considerably shorter than that of unemployment.²

•  Third, as noted earlier, the value of government debt tends to explode; it rose an average of 86 percent (in real terms, relative to precrisis debt) in the major post-World War II episodes. As discussed in chapter 10 (and as we reiterate here), the main cause of debt explosions is not the widely cited costs of bailing out and recapitalizing the banking system. Admittedly, bailout costs are difficult to measure, and the divergence among estimates from competing studies is considerable. But even upper-bound estimates pale next to actual measured increases in public debt. In fact, the biggest driver of debt increases is the inevitable collapse in tax revenues that governments suffer in the wake of deep and prolonged output contractions. Many countries also suffer from a spike in the interest burden on debt, for interest rates soar, and in a few cases (most notably that of Japan in the 1990s), countercyclical fiscal policy efforts contribute to the debt buildup. (We note that calibrating differences in countercyclical fiscal policy across countries can be difficult because some countries, such as the Nordic countries, have powerful built-in fiscal stabilizers through high marginal tax rates and generous unemployment benefits, whereas other countries, such as the United States and Japan, have automatic stabilizers that are far weaker.)

In the last part of the chapter, we will look at quantitative benchmarks from the period of the Great Depression, the last deep global financial crisis prior to the recent one. The depth and duration of the decline in economic activity were breathtaking, even by comparison with severe postwar crises. Countries took an average of ten years to reach the same level of per capita output as they enjoyed in 1929. In the first three years of the Depression, unemployment rose an average of 16.9 percentage points across the fifteen major countries in our comparison set.

Historical Episodes Revisited

The preceding chapter included all the major postwar banking crises in the developed world (a total of eighteen) and put particular emphasis on the ones dubbed the “Big Five” (those in Spain, 1977; Norway, 1987; Finland, 1991; Sweden, 1991; and Japan, 1992). It is quite clear from that chapter, as well as from the subsequent evolution of the 2007 U.S. financial crisis, that the crisis of the late 2000s must be considered a severe Big Five-type crisis by any metric. As a result, in this chapter we will focus on severe systemic financial crises only, including the Big Five crises in developed economies plus a number of famous episodes in emerging markets: the 1997-1998 Asian crises (in Hong Kong, Indonesia, Korea, Malaysia, the Philippines, and Thailand); that in Colombia in 1998; and Argentina's 2001 collapse. These are cases for which we have all or most of the relevant data to allow for meaningful quantitative comparisons across key indicator variables, such as equity markets, housing markets, unemployment, growth, and so on. Central to the analysis are historical housing price data, which can be difficult to obtain and are critical for assessing the recent episode.³ We also include two earlier historical cases for which we have housing prices: those of Norway in 1899 and the United States in 1929.

The Downturn after a Crisis:
Depth and Duration

In figure 14.1, based on the same data as table 10.8, we again look at the bust phase of housing price cycles surrounding banking crises in the expanded data set. We include a number of countries that experienced crises from 2007 on. The latest crises are represented by bars in dark shading, past crises by bars in light shading. The cumulative decline in real housing prices from peak to trough averages 35.5 percent.⁴ The most severe real housing price declines were experienced by Finland, Colombia, the Philippines, and Hong Kong. Their crashes amounted to 50 to 60 percent, measured from peak to trough. The housing price decline experienced by the United States during the latest episode at the time of this writing (almost 28 percent in real terms through late 2008 according to the Case-Shiller index) is already more than twice that registered in the United States during the Great Depression.

Notably, the duration of housing price declines has been quite long lived, averaging roughly six years. Even excluding the extraordinary experience of Japan (with its seventeen consecutive years of real housing price declines), the average remains more than five years. As figure 14.2 illustrates, the equity price declines that accompany banking crises are far steeper than are housing price declines, albeit shorter lived. The shorter duration of a downturn compared with real estate prices is perhaps unsurprising given that equity prices are far less inertial. The average historical decline in equity prices has been 55.9 percent, with the downturn phase of the cycle lasting 3.4 years. As of the end of 2008, Iceland and Austria had already experienced peak-to-trough equity price declines far exceeding the average of the historical comparison group.

Figure 14.1. Cycles of past and ongoing real house prices and banking crises.
Sources: Appendixes A.1 and A.2 and sources cited therein.
Notes: Each banking crisis episode is identified by country and the beginning year of the crisis. Only major (systemic) banking crisis episodes are included, subject to data limitations. The historical average reported does not include ongoing crisis episodes. For the ongoing episodes, the calculations are based on data through the following periods: October 2008, monthly, for Iceland and Ireland; 2007, annual, for Hungary; and Q3, 2008, quarterly, for all others. Consumer price indexes are used to deflate nominal house prices.

In figure 14.3 we look at increases in unemployment rates across the historical comparison group. (Because the unemployment rate is classified as a lagging indicator, we do not include the most recent crisis, although we note that the U.S. unemployment rate has already risen by 5 percentage points from its bottom value of near 4 percent.) On average, unemployment rises for almost five years, with an increase in the unemployment rate of about 7 percentage points. Although none of the postwar episodes has rivaled the rise in unemployment of more than 20 percentage points experienced by the United States during the Great Depression, the employment consequences of financial crises are nevertheless strikingly large in many cases. For emerging markets the official statistics likely underestimate true unemployment.

Figure 14.2. Cycles of past and ongoing real equity prices and banking crises.
Sources: Appendixes A.1 and A.2 and sources cited therein.
Notes: Each banking crisis episode is identified by country and the beginning year of the crisis. Only major (systemic) banking crisis episodes are included, subject to data limitations. The historical average reported does not include ongoing crisis episodes. For the ongoing episodes, the calculations are based on data through December 2, 2008. Consumer price indexes are used to deflate nominal equity prices.

Interestingly, figure 14.3 reveals that when it comes to banking crises, the emerging markets, particularly those in Asia, seem to do better in terms of unemployment than the advanced economies. (An exception was seen in the deep recession experienced by Colombia in 1998.) Although there are well-known data issues involved in comparing unemployment rates across countries,⁵ the relatively poor performance in advanced countries suggests the possibility that greater (downward) wage flexibility in emerging markets may help cushion employment during periods of severe economic distress. The gaps in the social safety net in emerging market economies, compared to industrial ones, presumably also make workers more anxious to avoid becoming unemployed.

Figure 14.3. Cycles of past unemployment and banking crises.
Sources: Organisation for Economic Co-operation and Development; International Monetary Fund (various years), International Financial Statistics; Carter et al. (2006); various country sources; and the authors’ calculations.
Notes: Each banking crisis episode is identified by country and the beginning year of the crisis. Only major (systemic) banking crisis episodes are included, subject to data limitations. The historical average reported does not include ongoing crisis episodes.

In figure 14.4 we look at the cycles in real per capita GDP around severe banking crises. The average magnitude of declines, at 9.3 percent, is stunning. Admittedly, as we noted earlier, for the post-World War II period, the declines in real GDP have been smaller for advanced economies than for emerging market economies. A probable explanation for the more severe contractions in emerging market economies is that they are prone to abrupt reversals in the availability of foreign credit. When foreign capital comes to a “sudden stop,” to use the phrase popularized by Rudiger Dornbusch and Guillermo Calvo, economic activity heads into a tailspin.⁶

Figure 14.4. Cycles of past real per capita GDP and banking crises.
Sources: Total Economy Database (TED), Carter et al. (2006), and the authors’ calculations.
Notes: Each banking crisis episode is identified by country and the beginning year of the crisis. Only major (systemic) banking crisis episodes are included, subject to data limitations. The historical average reported does not include ongoing crisis episodes. Total GDP in millions of 1990 U.S. dollars (converted at Geary Khamis PPPs) divided by midyear population.

Compared to unemployment, the cycle from peak to trough in GDP is much shorter, only two years. Presumably this is partly because potential GDP growth is positive and we are measuring only absolute changes in income, not gaps relative to potential output. Even so, the recessions surrounding financial crises are unusually long compared to normal recessions, which typically last less than a year.⁷ Indeed, multiyear recessions usually occur only in economies that require deep restructuring, such as that of Britain in the 1970s (prior to the advent of Prime Minister Margaret Thatcher), Switzerland in the 1990s, and Japan after 1992 (the last due not only to its financial collapse but also to the need to reorient its economy in light of China's rise). Banking crises, of course, usually require painful restructuring of the financial system and so are an important example of this general principle.

The Fiscal Legacy of Crises

Declining revenues and higher expenditures, owing to a combination of bailout costs and higher transfer payments and debt servicing costs, lead to a rapid and marked worsening in the fiscal balance. The episodes of Finland and Sweden stand out in this regard; the latter went from a precrisis surplus of nearly 4 percent of GDP to a whopping 15 percent deficit-to-GDP ratio. See table 14.1.

figure 14.5 shows the increase in real government debt in the three years following a banking crisis. The deterioration in government finances is striking, with an average debt increase of more than 86 percent. The calculation here is based on relatively recent data from the past few decades, but recall that in chapter 10 of this book we take advantage of our newly unearthed historical data on domestic debt to show that a buildup in government debt has been a defining characteristic of the aftermath of banking crises for over a century. We look at the percentage increase in debt rather than in debt relative to GDP because sometimes steep output drops complicate the interpretation of debt-to-GDP ratios. We have already emphasized but it bears being stated again, the characteristically huge buildup in government debt is driven mainly by a sharp falloff in tax revenue due to the deep recessions that accompany most severe financial crises. The much-ballyhooed bank bailout costs have been, in several cases, only a relatively minor contributor to the postcrisis increase in debt burdens.

TABLE 14.1
Fiscal deficits (central government balance) as a percentage of GDP

    Sources: International Monetary Fund (various years), Government Financial Statistics and World Economic Outlook, and the authors’ calculations.
    ^aAs shown in figure 14.4, Spain was the only country in our sample to show a (modest) increase in per capita GDP growth during the postcrisis period.

Figure 14.5. The cumulative increase in real public debt in the three years following past banking crises.
Sources: Appendixes A.1 and A.2 and sources cited therein.
Notes: Each banking crisis episode is identified by country and the beginning year of the crisis. Only major (systemic) banking crisis episodes are included, subject todata limitations. The historical average reported does not include ongoing crisis episodes, which are omitted altogether, because these crises began in 2007 or later, and the debt stock comparison here is with three years after the beginning of the banking crisis. Public debt is indexed to equal 100 in the year of the crisis.

Sovereign Risk

As shown in figure 14.6, sovereign default, debt restructuring, and/or near default (avoided by international bailout packages) have been a part of the experience of financial crises in many emerging markets; therefore, a decline in a country's credit rating during a crisis hardly comes as a surprise. Advanced economies, however, do not go unscathed. Finland's sovereign risk rating score went from 79 to 69 in the space of three years, leaving it with a score close to those of some emerging markets! Japan suffered several downgrades from the more famous rating agencies as well.

Figure 14.6. Cycles of Institutional Investor sovereign ratings and past banking crises.
Sources: Institutional Investor (various years) and the authors’ calculations.
Notes: Institutional Investor's ratings range from 0 to 100, rising with increasing creditworthiness.

Comparisons with Experiences from the
First Great Contraction in the 1930s

Until now, our comparison benchmark has consisted of postwar financial crises. The quantitative similarities of those crises with the recent crisis in the United States, at least for the run-up and early trajectory, have been striking. Yet, in many ways this “Second Great Contraction” is a far deeper crisis than others in the comparison set, because it is global in scope, whereas the other severe post-World War II crises were either country-specific or at worst regional. Of course, as we will discuss in more detail in chapter 17, policy authorities reacted somewhat hesitantly in the 1930s, which may also explain the duration and severity of the crisis. Nevertheless, given the lingering uncertainty over the future evolution of the crisis of the late 2000s (the Second Great Contraction), it is useful to look at evidence from the 1930s, the First Great Contraction.

Figure 14.7 compares the crises of the 1930s with the deep post-World War II crises in terms of the number of years over which output fell from peak to trough. The upper panel shows postwar crises including those in Colombia, Argentina, Thailand, Indonesia, Sweden, Norway, Mexico, the Philippines, Malaysia, Japan, Finland, Spain, Hong Kong, and Korea—fourteen in all. The lower panel shows fourteen Great Depression crises, including those in Argentina, Chile, Mexico, Canada, Austria, France, the United States, Indonesia, Poland, Brazil, Germany, Romania, Italy, and Japan.

Figure 14.7. The duration of major financial crises: Fourteen Great Depression episodes versus fourteen post-World War II episodes (duration of the fall in output per capita).
Sources: Appendix A.3 and the authors’ calculations.
Notes: The fourteen postwar episodes were those in Spain, 1977; Norway, 1987; Finland, 1991; Sweden, 1991; Japan, 1992; Mexico, 1994; Indonesia, Thailand, and (grouped as Asia-4 in the figure) Hong Kong, Korea, Malaysia, and Philippines, all 1997; Colombia, 1998; and Argentina, 2001. The fourteen Great Depression episodes were comprised of eleven banking crisis episodes and three less systemic but equally devastating economic contractions in Canada, Chile, and Indonesia during the 1930s. The banking crises were those in Japan, 1927; Brazil, Mexico, and the United States, all 1929; France and Italy, 1930; and Austria, Germany, Poland, and Romania, 1931.

Each half of the diagram forms a vertical histogram. The number of years each country or several countries were in crisis is measured on the vertical axis. The number of countries experiencing a crisis of any given length is measured on the horizontal axis. One sees clearly from the diagram that the recessions accompanying the Great Depression were of much longer duration than the postwar crises. After the war, output typically fell from peak to trough for an average of 1.7 years, with the longest downturn of four years experienced by Argentina and Finland. But in the Depression, many countries, including the United States and Canada, experienced a downturn of four years or longer, with Mexico and Romania experiencing a decrease in output for six years. Indeed, the average length of time over which output fell was 4.1 years in the Great Depression.⁸

It is important to recognize that standard measures of the depth and duration of recessions are not particularly suitable for capturing the epic decline in output that often accompanies deep financial crises. One factor is the depth of the decline, and another is that growth is sometimes quite modest in the aftermath as the financial system resets. An alternative perspective is provided in figure 14.8, which measures the number of years it took for a country's output to reach its precrisis level. Of course, after a steep fall in output, just getting back to the starting point can take a long period of growth. Both halves of the figure are stunning. For the postwar episodes, it took an average of 4.4 years for output to claw its way back to precrisis levels. Japan and Korea were able to do this relatively quickly, at only 2 years, whereas Colombia and Argentina took 8 years. But things were much worse in the Depression, and countries took an average of 10 years to increase their output back to precrisis levels, in part because no country was in a position to “export its way to recovery” as world aggregate demand imploded. The figure shows, for example, that the United States, France, and Austria took 10 years to rebuild their output to its initial pre-Depression level, whereas Canada, Mexico, Chile, and Argentina took 12. Thus, the Great Depression era sets far more daunting benchmarks for the potential trajectory of the financial crisis of the late 2000s than do the main comparisons we have been making to severe postwar crises.

Figure 14.8. The duration of major financial crises: Fourteen Great Depression episodes versus fourteen post-World War II episodes (number of years for output per capita to return to its precrisis level).
Sources: Appendix A.3 and the authors’ calculations.
Notes: The fourteen postwar episodes were those in Spain, 1977; Norway, 1987; Finland, 1991; Sweden, 1991; Japan, 1992; Mexico, 1994; Hong Kong, Indonesia, Korea, Malaysia, the Philippines, and Thailand, all 1997; Colombia, 1998; and Argentina, 2001. The fourteen Great Depression episodes were comprised of eleven banking crisis episodes and three less systemic but equally devastating economic contractions in Canada, Chile, and Indonesia. The banking crises were those in Japan, 1927; Brazil, Mexico, and the United States, all 1929; France and Italy, 1930; and Austria, Germany, Poland, and Romania, 1931. The precrisis level for the Great Depression was that of 1929.

As we will show in chapter 16, the unemployment increases in the Great Depression were also far greater than those in the severe post-World War II financial crises. The average rate of unemployment increase was about 16.8 percent. In the United States, unemployment rose from 3.2 percent to 24.9 percent.

Finally, in figure 14.9 we look at the evolution of real public debt during the crises of the Great Depression era. Interestingly, public debt grew more slowly in the aftermath of these crises than it did in the severe postwar crises. In the Depression, it took six years for real public debt to grow by 84 percent (versus half that time in the postwar crises). Some of this difference reflects the very slow policy response that occurred in the Great Depression. It is also noteworthy that public debt in emerging markets did not increase in the later stages (three to six years) following the crises. Some of these emerging markets had already drifted into default (on both domestic and external debts); others may have faced the kind of external constraints that we discussed in connection with debt intolerance and, as such, had little capacity to finance budget deficits.

Figure 14.9. The cumulative increase in real public debt three and six years following the onset of the Great Depression in 1929: Selected countries.
Sources: Reinhart and Rogoff (2008b) and sources cited therein.
Notes: The beginning years of the banking crises range from 1929 to 1931. Australia and Canada did not have a systemic banking crisis but are included for comparison purposes, because both also suffered severe and protracted economic contractions. The year 1929 marks the peak in world output and hence is used as the marker for the beginning of the Depression episode.

Concluding Remarks

An examination of the aftermath of severe postwar financial crises shows that these crises have had a deep and lasting effect on asset prices, output, and employment. Unemployment increases and housing price declines have extended for five and six years, respectively. Real government debt has increased by an average of 86 percent after three years.

How relevant are historical benchmarks in assessing the trajectory of a crisis such as the global financial crisis of the late 2000s, the Second Great Contraction? On the one hand, authorities now have arguably more flexible monetary policy frameworks, thanks particularly to a less rigid global exchange rate regime. And some central banks showed an aggressiveness early on by acting in a way that was notably absent in the 1930s or in the latter-day Japanese experience. On the other hand, we would be wise not to push too far the conceit that we are smarter than our predecessors. A few years back, many people would have said that improvements in financial engineering had done much to tame the business cycle and limit the risk of financial contagion. And as we saw in the final section of this chapter, the Great Depression crises were far more traumatic events than even the more severe of the post-World War II crises. In the Depression, it took countries in crisis an average of ten years for real per capita GDP to reach its precrisis level. Still, in the postwar crises it has taken almost four and a half years for output to reach its precrisis level (though growth has resumed much more quickly, it has still taken time for the economy to return to its starting point).

What we do know is that after the start of the recent crisis in 2007, asset prices and other standard crisis indicator variables tumbled in the United States and elsewhere along the tracks laid down by historical precedent. It is true that equity markets have since recovered some ground, but by and large this is not out of line with the historical experience (already emphasized in chapter 10) that V-shaped recoveries in equity prices are far more common than V-shaped recoveries in real housing prices or employment. Overall, this chapter's analysis of the postcrisis outcomes for unemployment, output, and government debt provides sobering benchmark numbers for how deep financial crises can unfold. Indeed, our post-World War II historical comparisons were largely based on episodes that were individual or regional in nature. The global nature of the recent crisis has made it far more difficult, and contentious, for individual countries to grow their way out through higher exports or to smooth the consumption effects through foreign borrowing. As noted in chapter 10, historical experience suggests that the brief post-2002 lull in sovereign defaults is at risk of coming to an abrupt end. True, the planned quadrupling of International Monetary Fund (IMF) resources, along with the apparent softening of IMF loan conditions, could have the effect of causing the next round of defaults to play out in slow motion, albeit with a bigger bang at the end if the IMF itself runs into broad repayment problems. Otherwise, as we have mentioned repeatedly, defaults in emerging market economies tend to rise sharply when many countries are simultaneously experiencing domestic banking crises.