- 16 -

COMPOSITE MEASURES OF
FINANCIAL TURMOIL

In this book we have emphasized the clustering of crises at several junctures both across countries and across different types of crises. A country experiencing an exchange rate crisis may soon find itself in banking and inflation crises, sometimes with domestic and external default to follow. Crises are also transmitted across countries through contagion or common factors, as we discussed in the previous chapter.

Until now, however, we have not attempted to construct any quantitative index that combines crises regionally or globally. Here, in keeping with the algorithmic approach we have applied to delineating individual financial crisis events, we will offer various types of indexes of financial turbulence that are helpful in assessing the global, regional, and national severity of a crisis.

Our financial turbulence index reveals some stunning information. The most recent global financial crisis—which we have termed the “Second Great Contraction”—is clearly the only global financial crisis that has occurred during the post–World War II period. Even if the Second Great Contraction does not evolve into the Second Great Depression, it still surpasses other turbulent episodes, including the breakdown of Bretton Woods, the first oil shock, the debt crisis of the 1980s in the developing world, and the now-famous Asian crisis of 1997–1998. The Second Great Contraction is already marked by an extraordinarily global banking crisis and by spectacular global exchange rate volatility. The synchronicity of the collapses in housing markets and employment also appears unprecedented since the Great Depression; late in this chapter we will show little-used data from the Great Depression to underscore this comparison.

The index of financial turbulence we develop in this chapter can also be used to characterize the severity of regional crisis, and here we compare the experiences of different continents. The index shows how misinformed is the popular view that Asia does not have financial crises.

This chapter not only links crises globally but also takes on the issue of how different varieties of crisis are linked within a country. Following Kaminsky and Reinhart, we discuss how (sometimes latent) banking crises often lead to currency crashes, outright sovereign default, and inflation.¹

Finally, we conclude by noting that pulling out of a global crisis is, by nature, more difficult than pulling out of a multicountry regional crisis (such as the Asian financial crisis of 1997–1998). Slow growth in the rest of the world cuts off the possibility that foreign demand will compensate for collapsing domestic demand. Thus, measures such as our index of global financial turbulence can potentially be useful in designing the appropriate policy response.

Developing a Composite Index of Crises:
The BCDI Index

We develop our index of crisis severity as follows. In chapter 1 we defined five “varieties” of crises: external and domestic sovereign default, banking crises, currency crashes, and inflation outbursts.² Our composite country financial turbulence index is formed by simply summing up the number of types of crises a country experiences in a given year. Thus, if a country did not experience any of our five crises in a given year, its turbulence index for that year would be zero, while in a worst-case scenario (as in Argentina in 2002, for instance) it would be five. We assign such a value for each country for each year. This is what we dub the BCDI index, which stands for banking (systemic episodes only), currency, debt (domestic and external), and inflation crisis index.

Although this exercise captures some of the compounding dimensions of the crisis experience, it admittedly remains an incomplete measure of its severity.³ If inflation goes to 25 percent per annum (meeting the threshold for a crisis by our definition), it receives the same weight in the index as if it went to 250 percent, which is obviously far more serious.⁴ This binary treatment of default is similar to that of the rating agency Standard and Poor's (S&P), which lists countries as either in default or not in default. The S&P index (and ours) take account of debt crisis variables. For example, Uruguay's relatively swift and “market-friendly” restructuring in 2003 is assigned the same value as the drawn-out outright default and major “haircut” successfully imposed on creditors by its larger neighbor, Argentina, during its 2001–2002 default. Nevertheless, indexes such as S&P's have proven enormously useful over time precisely because default tends to be such a discrete event. Similarly, a country that reaches our crisis markers across multiple varieties of crises is almost surely one undergoing severe economic and financial duress.

Where feasible, we also add to our five-crises composite a “Kindleberger-type” stock market crash, which we show separately.⁵ In this case, the index runs from zero to six.⁶ Although Kindleberger himself did not provide a quantitative definition of a crash, Barro and Ursúa have adopted a reasonable benchmark for defining asset price collapses, which we adopt here. They define a stock market crash as a cumulative decline of 25 percent or more in real equity prices.⁷ We apply their methods to the sixty-six countries covered in our sample; the starting dates for equity prices are determined by data availability, as detailed on a country-by-country basis in the data appendixes. Needless to say, our sample of stock market crashes ends with a bang in the cross-country megacrashes of 2008. As in the case of growth collapses, many (if not most) of the stock market crashes have coincided with the crisis episodes described here (chapters 1 and 11). “Most” clearly does not mean all; the Black Monday crash of October 1987 (for example) is not associated with a crisis of any other stripe. False signal flares from the equity market are, of course, familiar. As Samuelson famously noted, “The stock market has predicted nine of the last five recessions.”⁸ Indeed, although global stock markets continued to plummet during the first part of 2009 (past the end date of our core data set), they then rose markedly in the second quarter of the year, though they hardly returned to their precrisis level.

Beyond sovereign events, there are two other important dimensions of defaults that our crisis index does not capture directly. First, there are defaults on household debt. These defaults, for instance, have been at center stage in the unfolding subprime saga in the United States in the form of the infamous toxic mortgages. Household defaults are not treated separately in our analysis owing to a lack of historical data, even for advanced economies. However, such episodes are most likely captured by our indicator of banking crises. Banks, after all, are the principal sources of credit to households, and large-scale household defaults (to the extent that these occur) impair bank balance sheets.

More problematic is the incidence of corporate defaults, which are in their own right another “variety of crisis.” This omission is less of an issue in countries where corporations are bank-dependent. In such circumstances, the same comment made about household default applies to corporate debt. For countries with more developed capital markets, it may be worthwhile to consider widespread corporate default as yet another variety of crisis. As shown in figure 16.1, the United States began to experience a sharp run-up in the incidence of corporate default during the Great Depression well before the government defaulted (the abrogation of the gold clause in 1934). However, it is worth noting that corporate defaults and banking crises are indeed correlated, so our index may partially capture this phenomenon indirectly. In many episodes, corporate defaults have also been precursors to government defaults or reschedulings as governments have tended to shoulder private sector debts.

An Illustration of the Composite at a Country Level

The Argentine crisis of 2001–2002 illustrates how crises may potentially reinforce and overlap one another. The government defaulted on all its debts, domestic and foreign; the banks were paralyzed in a “banking holiday” when deposits were frozen indefinitely; the exchange rate for pesos to U.S. dollars went from one to more than three practically overnight; and prices went from declining (with deflation running at an annual rate of -1 percent or so) to inflating at a rate of about 30 percent (by conservative official estimates). We might add that this episode qualifies as a Barro-Ursúa growth collapse (per capita GDP fell by about 20 to 25 percent), and real stock prices crashed by more than 30 percent, along the lines of a Kindlebergertype crash episode.

Figure 16.1. The proportion of countries with systemic banking crises (weighted by their share of world income) and U.S. corporate speculative-grade default rates, 1919–2008.
Sources: Kaminsky and Reinhart (1999), Bordo et al. (2001), Maddison (2004), Caprio et al. (2005), Jácome (2008), Moody's Magazine (various issues), and additional sources listed in appendix A.3, which provides banking crises dates.
Notes: The sample includes all sixty-six countries listed in table 1.1 that were independent states in the given year. Three sets of GDP weights are used, 1913 weights for the period 1800–1913, 1990 weights for the period 1914–1990, and finally 2003 weights for the period 1991–2008. The entries for 2007–2008 list crises in Austria, Belgium, Germany, Hungary, Japan, the Netherlands, Spain, the United Kingdom, and the United States. The figure shows two-year moving averages.

World Aggregates and Global Crises

To transition from the experience of individual countries to a world or regional aggregate, we take weighted averages across all countries or for a particular region. The weights, as discussed earlier, are given by the country's share in world output. Alternatively, one can calculate an average tally of crises across a particular country group using a simple unweighted average. We will illustrate both.

Historical Comparisons

Our aggregate crisis indexes are the time series shown for 1900–2008 in figures 16.2 and 16.3 for the world and for the advanced economies. The advanced economies aggregate comprises the eighteen high-income countries in our sample, while the emerging markets group aggregates forty-eight entries from Africa, Asia, Europe, and Latin America. The indexes shown are weighted by a country's share in world GDP, as we have done for debt and banking crises.⁹ The country indexes (without stock market crashes) are compiled from the time of each country's independence (if after 1800) onward; the index that includes the equity market crashes is calculated based on data availability.

Figure 16.2. Varieties of crises: World aggregate, 1900–2008.
Source: The authors’ calculations.
Notes: The figure presents a composite index of banking, currency, sovereign default, and inflation crises and stock market crashes (weighted by their share of world income).The banking, currency, default (domestic and external), and inflation composite (BCDI) index can take a value between zero and five (for any country in any given year) depending on the varieties of crises occurring in a particular year. For instance, in 1998 the index took on a value of 5 for Russia, which was experiencing a currency crash, a banking and inflation crisis, and a sovereign default on both domestic and foreign debt obligations. This index is then weighted by the country's share in world income. This index is calculated annually for the sixty-six countries in the sample for 1800–2008 (shown above for 1900 onward). In addition, we use the definition of a stock market crash given by Barro and Ursúa (2009) for the twenty-five countries in their sample (a subset of the sixty-six-country sample except for Switzerland) for the period 1864–2006; we update their definition of a crash through December 2008 to compile our banking, currency, default (domestic and external), and inflation composite (BCDI +) index. For the United States, for example, the index posts a reading of 2 (banking crisis and stock market crash) in 2008; for Australia and Mexico it also posts a reading of 2 (currency and stock market crash).

Figure 16.3. Varieties of crises: Advanced economies aggregate, 1900–2008.
Source: The authors’ calculations.
Notes: This figure presents a composite index of banking, currency, sovereign default, and inflation crises and stock market crashes, weighted by their share of world income. The banking, currency, default (domestic and external), and inflation composite (BCDI) index can take a value between zero and 5 (for any country in any given year) depending on the varieties of crises taking place in a particular year. For instance, in 1947 the index took on a value of 4 for Japan, which was experiencing a currency crash, an inflation crisis, and a sovereign default on both domestic and foreign debt obligations. This index is then weighted by the country's share in world income. This index is calculated annually for the eighteen advanced economies (includes Austria but not Switzerland) in the Reinhart-Rogoff sample for 1800–2008 (shown above for 1900 onward). In addition, we use the definition of a stock market crash given by Barro and Ursúa (2009) for eighteen advanced economies (includes Switzerland but not Austria) for the period 1864–2006; we update their definition of a crash through December 2008 to compile our banking, currency, default (domestic and external), and inflation composite (BCDI +) index. For the United States and the United Kingdom, for example, the index posts a reading of 2 (banking crisis and stock market crash) in 2008; for Australia and Norway it also posts a reading of 2 (currency and stock market crash). ERM is exchange rate mechanism of the euro system.

Although inflation and banking crises predated independence in many cases, a sovereign debt crisis (external or internal) is, by definition, not possible for a colony. In addition, numerous colonies did not always have their own currencies. When stock market crashes (shown separately) are added to the BCDI composite, we refer to it as the BCDI +.

Figures 16.2 and 16.3 chronicle the incidence, and to some degree the severity, of varied crisis experiences. A cursory inspection of these figures reveals the very different patterns of the pre-World War II and postwar experiences. This difference is most evident in figure 16.3, which plots the indexes for eighteen advanced economies. The prewar experience was characterized by frequent and severe crisis episodes ranging from the banking crisis-driven “global” panic of 1907 to the debt and inflation crises associated with World War II and its aftermath.¹⁰

The postwar periods offered some bouts of turbulence: the inflationary outbursts that accompanied the first oil shocks in the mid-1970s, the recessions associated with bringing down inflation in the early 1980s, the severe banking crises in the Nordic countries and Japan in the early 1990s, and the bursting of the dot-com bubble in the early 2000s. However, these episodes pale in comparison with their prewar counterparts and with the global contraction of 2008, which has been unparalleled (by a considerable margin) in the sixtyplus years since World War II (figure 16.3). Like its prewar predecessors, the 2008 episode has been both severe in magnitude and global in scope, as reflected by the large share of countries mired in crises. Stock market crashes have been nearly universal. Banking crises have emerged as asset price bubbles have burst and high degrees of leverage have become exposed. Currency crashes against the U.S. dollar in advanced economies took on the magnitudes and volatilities of crashes in emerging markets.

A growing body of academic literature, including contributions by McConnell and Perez-Quiros and Blanchard and Simon, had documented a post-mid-1980s decline in various aspects of macroeconomic volatility, presumably emanating from a global low-inflation environment. This had been termed “a Great Moderation” in the United States and elsewhere.¹¹ However, systemic crises and low levels of macroeconomic volatility do not travel hand in hand; the sharp increases in volatility that occurred during the Second Great Contraction, which began in 2007, are evident across asset markets, including real estate, stock prices, and exchange rates. They are also manifestly evident in the macroeconomic aggregates, such as those for output, trade, and employment. It remains to be seen how economists will assess the Great Moderation and its causes after the crisis recedes.

For many emerging markets, the Great Moderation was a fleeting event. After all, the debt crisis of the 1980s was as widespread and severe as the events of the 1930s (figure 16.3). These episodes, which affected Africa, Asia, and Latin America in varying degrees, often involved a combination of sovereign default, chronic inflation, and protracted banking crises. As the debt crisis of the 1980s settled, new eruptions emanated from the economies of Eastern Europe and the former Soviet Union in the early 1990s. The Mexican crisis of 1994–1995 and its repercussions in Latin America, the fierce Asian crisis that began in the summer of 1997, and the far-reaching Russian crisis of 1998 did not make for many quiet stretches in emerging markets. This string of crises culminated in Argentina's record default and implosion in 2001–2002.¹²

Until the crisis that began in the United States in the summer of 2007 and became global in scope a year later, emerging markets enjoyed a period of tranquility and even prosperity. During 2003–2007, world growth conditions were favorable, commodity prices were booming, and world interest rates were low, so credit was cheap. However, five years is too short a time span to contemplate extending the “Great Moderation” arguments to emerging markets; in effect, the events of the past two years have already rekindled volatility almost across the board.

Regional Observations

We next look at the regional profile of crises. In figures 16.2 and 16.3 we looked at averages weighted by country size. So that no single country will dominate the regional profiles, the remainder of this discussion focuses on unweighted simple averages for Africa, Asia, and Latin America. In figures 16.4–16.6 we show regional tallies for 1800–2008 for Asia and Latin America and for the post–World War II period for the more newly independent African states.

For Africa, the regional composite index of financial turbulence begins in earnest in the 1950s (figure 16.4), for only South Africa (1910) was a sovereign state prior to that period. However, we do have considerable coverage of prices and exchange rates for the years following World War I, so numerous preindependence crises (including some severe banking crises in South Africa) are dated and included for the colonial period. The index jumps from a low that is close to zero in the 1950s to a high in the 1990s. The thirteen African countries in our sample had, on average, two simultaneous crises during the worst years of the 1980s. In all cases, except that of Mauritius, which has neither defaulted on nor restructured its sovereign debts, the two crises could have been a pairing of any of our crisis varieties. The decline in the average number of crises in the 1990s reflected primarily a decline in the incidence of inflation crises and the eventual (if protracted) resolution of the decade-long debt crisis of the 1980s.

Figure 16.4. Varieties of crises: Africa, 1900–2008.
Source: The authors’ calculations based on sources listed in appendixes A.1-A.3.

Figure 16.5. Varieties of crises: Asia, 1800–2008.
Source: The authors’ calculations based on sources listed in appendixes A.1-A.3.

Figure 16.6. Varieties of crises: Latin America, 1800–2008.
Source: The authors’ calculations based on sources listed in appendixes A.1-A.3.
Notes: The hyperinflations in Argentina, Bolivia, Brazil, Nicaragua, and Peru sharply increase in the index (reflected in the spike shown for the late 1980s and early 1990s) because all these episodes register a maximum reading of 5.

The regional composite index of financial turbulence for Asia (figure 16.5) spans 1800–2008, for China, Japan, and Thailand were independent nations throughout this period. Having gained independence almost immediately following World War II, the remaining Asian countries in the sample then join in the regional average. The profile for Asia highlights a point we have made on more than one occasion: the economic claim of the superiority of the “tigers” or “miracle economies” in the three decades before the 1997–1998 crisis was naïve in terms of the local history. The region had experienced several protracted bouts of economic instability by the international standards of the day. The most severe crisis readings occurred during the period bracketed by the two world wars. In that period, China saw hyperinflation, several defaults, more than one banking crisis, and countless currencies and currency conversions. Japan had numerous bouts of banking, inflation, and exchange rate crises, culminating in its default on its external debt during World War II, the freezing of bank deposits, and its near-hyperinflation (approaching 600 percent) at the end of the war in 1945.

Perhaps Latin America would have done better in terms of economic stability had the printing press never crossed the Atlantic (figure 16.6). Before Latin America's long struggle with high, hyper-, and chronic inflation took a dark turn in the 1970s, the region's average turbulence index reading was very much in line with the world average. Despite periodic defaults, currency crashes, and banking crises, the average never really surpassed one crisis per year, in effect comparing moderately favorably with those of other regions for long stretches of time. The rise of inflation (which began before the famous debt crisis of the 1980s, the “lost decade”) would change the relative and absolute performance of the region until the second half of the 1990s. During Latin America's worst moments in the late 1980s—before the 1987 Brady plan (discussed earlier in box 5.3) restructured bad sovereign debts and while Argentina, Brazil, and Peru were mired in hyperinflation—as we can see from the index, the region experienced an average of almost three crises a year.¹³

Defining a Global Financial Crisis

Although the indexes of financial turbulence we have developed can be quite useful in assessing the severity of a global financial crisis, we need a broader-ranging algorithm to systematically delineate true crises so as to exclude, for example, a crisis that registers high on the global scale but affects only one large region. We propose the working definition of a global financial crisis found in box 16.1.

Global Financial Crises: Economic Effects

We next turn to two broad factors associated with global crises, both of which are present in the recent-vintage global contraction: first, the effects of the crisis on the level and the volatility of economic activity broadly defined and measured by world aggregates of equity prices, real GDP, and trade; and second, its relative synchronicity across countries, which is evident in asset markets as well as trends in trade, employment, and other economic sectoral statistics, such as housing. The emphasis of our discussion is on the last two global crises, the Great Depression of the 1930s and the Second Great Contraction, for which documentation is most complete. Obviously, looking at this broad range of macroeconomic data gives us a much more nuanced picture of a crisis.

Global Aggregates

The connection between stock prices and future economic activity is hardly new. The early literature on turning points in the business cycle, such as the classic by Burns and Mitchell, documented the leading-indicator properties of share prices.¹⁴ Synchronous (across-the-board) and large declines in equity prices (crashes) characterized the onset of the episode that became the Great Depression and somewhat more belatedly the recent global crisis. Figure 16.7 plots an index of global stock prices for 1929–1939 and for 2008–2009 (to the present). For the more recent episode, the index accounts for about 70 percent of world equity market capitalization and covers seven distinct regions and twenty-nine countries. Stock prices are deflated by world consumer prices. The data for 1928–1939 are constructed using median inflation rates for the sixty-six-country sample; for 2007–2009 they are taken from the end-of-period prices published in the World Economic Outlook.¹⁵ The years 1928 and 2007 marked the cycle peak in these indices.

The decline in equity markets during 2008 and beyond match the scale (and the cross-country reach) of the 1929 crashes. It is worth noting that during the crisis of the 1930s equity ownership worldwide was far more limited than it has become in the twenty-first century; the growth of pension funds and retirement plans and the ascent of an urban population have increased the links between household wealth and equity markets.

In much the same spirit as figure 16.7, figure 16.8 plots real per capita GDP (weighted by world population) for various country groupings for the two global crises.¹⁶ The aggregate for Europe corresponds to Maddison's twelve-country population-weighted aggregate;¹⁷ the index for Latin America is comprised of the region's eight largest countries. The year 1929 marked the peak in real per capita GDP for all three country groupings. The current data come from the World Economic Outlook. When all this information is taken together, it is difficult to reconcile the projected trajectory in real GDP, particularly for emerging markets, and the developments of 2008 through early 2009 in equity markets.

Figure 16.7. Global stock markets during global crises: The composite real stock price index (end of period).
Sources: Global Financial Data (GFD) (n.d.); Standard and Poor's; International Monetary Fund (various years), World Economic Outlook; and the authors (details provided in appendix A.1).
Notes: The world composite stock price index was taken from GFD for 1928–1939 and from S&P for 2007–2009. The S&P Global 1200 index covers seven distinct regions and twenty-nine countries and captures approximately 70 percent of the world market capitalization. Stock prices are deflated by world consumer prices. For 1928–1939 these have been constructed using median inflation rates for the sixty-six-country sample; for 2007–2009 these have been taken from the World Economic Outlook end-of-period prices. The years 1928 and 2007 marked the cycle peak in these indexes. The year of the crisis is indicated by t.

As for trade, we offer two illustrations of its evolution during the two global crises. The first of these (figure 16.9) is a reprint of an old classic titled “The Contracting Spiral of World Trade: Month by Month, January 1929-June 1933.” This inward spiral appeared in the World Economic Survey, 1932–1933, which in turn reprinted it from another contemporary source.¹⁸ The illustration documents the 67 percent decline in the value of trade as the Depression took hold. As has been extensively documented, including by contemporaneous sources, the collapse in international trade was only partially the byproduct of sharp declines in economic activity, ranging from about 10 percent for Western Europe to about 30 percent for Australia, Canada, New Zealand, and the United States.¹⁹ The other destructive factor was the worldwide increase in protectionist policies in the form of both trade barriers and competitive devaluations.

Figure 16.8. Real per capita GDP during global financial crises: Multicountry aggregates (PPP weighted).
Sources: Maddison (2004); International Monetary Fund (various years), World Economic Outlook; and the authors (details provided in appendix A.1).
Notes: The Europe aggregate corresponds to Maddison's twelve-country population-weighted aggregate; the Latin America index is comprised of the region's eight largest countries. The years 1929 and 2008 marked the peak in real per capita GDP for all three country groupings. The year of the crisis is indicated by t.

Figure 16.10 plots the value of world merchandise exports for 1928–2009. The estimate for 2009 uses the actual year-end level for 2008 as the average for 2009; this yields a 9 percent year-over-year decline in 2009, the largest one-year drop since 1938.²⁰ Other large post–World War II declines are in 1952, during the Korean War, and in 1982–1983, when recession hit the United States and a 1930s-scale debt crisis swept through the emerging world. Smaller declines occurred in 1958, the bottom of a recession in the United States; in 1998, during the Asian financial crisis; and in 2001, after September 11.

Figure 16.9. The contracting spiral of world trade month by month, January 1929-June 1933.
Source: Monatsberichte des Österreichischen Institutes für Konjunkturforschung 4 (1933): 63.

Cross-Country Synchronicity

The performance of the global aggregates provides evidence that a crisis has affected a sufficiently large share of the world's population and/or countries. However, because the information is condensed into a single world index, it does not fully convey the synchronous nature of global crises. To fill in this gap, we present evidence on the performance of various economic indicators during the most recent previous global crisis. Specifically, we present evidence on the changes in unemployment and indexes of housing activity, exports, and currency movements during 1929–1932.

Figure 16.10. World export growth, 1928–2009.
Sources: Global Financial Data (GFD) (n.d.); League of Nations (various years), World Economic Survey; International Monetary Fund (various years), World Economic Outlook; and the authors (see notes).
Notes: No world aggregate is available during World War II. The estimate for 2009 uses the actual year-end level for 2008 as the average for 2009; this yields a 9 percent year-over-year decline in 2009, the largest postwar drop. Other large post–World War II declines were in 1952, during the Korean War, and in 1982–1983, when recession hit the United States and a 1930s-scale debt crisis swept through the emerging world. Smaller declines occurred in 1958, the bottom of a recession in the United States; in 1998, during the Asian financial crisis; and in 2001, after September 11.

The massive collapse in trade at the height of the Great Depression was already made plain by the two figures displaying world aggregates. Figure 16.11 adds information on the widespread nature of the collapse, which affected countries in all regions, low-, middle-, and high-income alike. In other words, the world aggregates are truly representative of the individual country experience and are not driven by developments in a handful of large countries that are heavily weighted in the world aggregates. Apart from wars that have involved a significant share of the world either directly or indirectly (including the Napoleonic Wars), such across-the-board synchronicity is not to be found in the data.

Cross-country synchronicity is not limited to variables for which one would expect close cross-country co-movement, such as international trade or exchange rates. The construction industry, which lies at the epicenter of the recent boom-bust cycle in the United States and elsewhere, is usually best characterized as being part of the “nontraded sector.” Yet the decline in housing-related construction activity during 1929–1932 was almost as synchronous as that seen in trade, as illustrated in table 16.1.

Figure 16.11. The collapse of exports, 1929–1932.
Sources: The individual country sources are provided in appendix A.1; the authors’ calculations were also used.

With both traded and nontraded sectors shrinking markedly and consistently across countries, the deterioration in unemployment reported in table 16.2 should come as no surprise. Unemployment increases almost without exception (no comparable 1929 data are available for Japan and Germany) by an average of 17 percentage points. As in the discussion of the aftermath of the postwar crises in the preceding chapter, the figures reflect differences in the definition of unemployment and in the methods of compiling the statistics; hence cross-country comparisons, particularly of the levels, are tentative.

TABLE 16.1

Sources: League of Nations, World Economic Survey (various issues), Carter et al. (2006). Note: Note the differences in the definition of the indicator from country to country.
^aThrough February 2009.

TABLE 16.2

Sources: League of Nations (various issues), World Economic Survey; Carter et al. (2006). Note: The figures reflect differences in the definition of unemployment and in the methods of compiling the statistics, so cross-country comparisons, particularly of the levels, are tentative.
^aAnnual averages.

Some Reflections on Global Crises

Here we pause to underscore why global financial crises can be so much more dangerous than local or regional ones. Fundamentally, when a crisis is truly global, exports no longer form a cushion for growth. In a global financial crisis, one typically finds that output, trade, equity prices, and other indicators behave qualitatively (if not quantitatively) much the same way for the world aggregates as they do in individual countries. A sudden stop in financing typically not only hits one country or region but to some extent impacts a large part of the world's public and private sectors.

Conceptually, it is not difficult to see that for a country to be “pulled” out of a postcrisis slump is far more difficult when the rest of the world is similarly affected than when exports offer a stimulus. Empirically, this is not a proposition that can be readily tested. We have hundreds of crises in our sample, but very few global ones, and, as noted in box 16.1, some of the earlier global crises were associated with wars, which complicates comparisons even further.

More definitively, it can be inferred from the evidence of so many episodes that recessions associated with crises (of any variety) are more severe in terms of duration and amplitude than the usual business cycle benchmarks of the post–World War II period in both advanced economies and emerging markets. Crises that are part of a global phenomenon may be worse still in the amplitude and volatility (if not duration) of the downturn. Until the most recent crisis, there had been no postwar global financial crisis; thus, by necessity the comparison benchmarks are prewar episodes. As to severity, the Second Great Contraction has already established several postwar records. The business cycle has evidently not been tamed.

The Sequencing of Crises: A Prototype

Just as financial crises have common macroeconomic antecedents in terms of asset prices, economic activity, external indicators, and so on, common patterns also appear in the sequencing (temporal order) in which crises unfold. Obviously not all crises escalate to the extreme outcome of a sovereign default. Yet advanced economies have not been exempt from their share of currency crashes, bouts of inflation, severe banking crises, and, in an earlier era, even sovereign default.

Investigating what came first, banking or currency crises, was a central theme of Kaminsky and Reinhart's “twin crises” work; they also concluded that financial liberalization often preceded banking crises; indeed, it helped predict them.²¹ Demirgüç-Kunt and Detragiache, who employed a different approach and a larger sample, arrived at the same conclusion.²² Reinhart examined the link between currency crashes and external default.²³ Our work here has investigated the connections between domestic and external debt crises, inflation crises and default (domestic or external), and banking crises and external default.²⁴ Figure 16.12 maps out a “prototypical” sequence of events yielded by this literature.

As Diaz-Alejandro narrates in his classic paper about the Chilean experience of the late 1970s and early 1980s, “Goodbye Financial Repression, Hello Financial Crash,” financial liberalization simultaneously facilitates banks’ access to external credit and more risky lending practices at home.²⁵ After a while, following a boom in lending and asset prices, weaknesses in bank balance sheets become manifest and problems in the banking sector begin.²⁶ Often these problems are more advanced in the shakier institutions (such as finance companies) than in the major banks.

The next stage in the crisis unfolds when the central bank begins to provide support for these institutions by extending credit to them. If the exchange rate is heavily managed (it does not need to be explicitly pegged), a policy inconsistency arises between supporting the exchange rate and acting as lender of last resort to troubled institutions. The numerous experiences in these studies suggest that (more often than not) the exchange rate objective is subjugated to the role of the central bank as lender of last resort. Even if central bank lending to the troubled financial industry is limited in scope, the central bank may be more reluctant to engage in an “interest rate defense” policy to defend the currency than would be the case if the financial sector were sound. This brings the sequence illustrated in figure 16.12 to the box labeled “Currency crash.” The depreciation or devaluation of the currency, as the case may be, complicates the situation in (at least) three ways: (1) it exacerbates the problem of the banks that have borrowed in a foreign currency, worsening currency mismatches;²⁷ (2) it usually worsens inflation (the extent to which the currency crisis translates into higher inflation is highly uneven across countries, for countries with a history of very high and chronic inflation usually have a much higher and faster pass-through from exchange rates to prices);²⁸ and (3) it increases the odds of external and domestic default if the government has foreign currency-denominated debt.

Figure 16.12. The sequencing of crises: A prototype.
Sources: Based on empirical evidence from Diaz-Alejandro (1985), Kindleberger (1989), Demirgü j-Kunt and Detragiache (1998), Kaminsky and Reinhart (1999), Reinhart (2002), and Reinhart and Rogoff (2004, 2008c), among others.

At this stage, the banking crisis either peaks following the currency crash (if there is no sovereign credit crisis) or keeps getting worse as the crisis mounts and the economy marches toward a sovereign default (the next box in figure 16.12).²⁹ In our analysis of domestic and external credit events we have not detected a well-established sequence between these credit events. Domestic defaults have occurred before, during, and after external defaults, in no obvious pattern. As regards inflation, the evidence presented in chapter 9 all points in the direction of a marked deterioration in inflation performance after a default, especially a twin default (involving both domestic and foreign debt). The coverage of our analysis summarized here does not extend to the eventual crisis resolution stage.

We should note that currency crashes tend to be more serious affairs when governments have been explicitly or even implicitly fixing (or nearly fixing) the exchange rate. Even an implicit guarantee of exchange rate stability can lull banks, corporations, and citizens into taking on heavy foreign currency liabilities, thinking there is a low risk of a sudden currency devaluation that will sharply increase the burden of carrying such loans. In a sense, the collapse of a currency is a collapse of a government guarantee on which the private sector might have relied, and therefore it constitutes a default on an important promise. Of course, large swings in exchange rates can also be traumatic for a country with a clear and explicit regime of floating exchange rates, especially if there are substantial levels of foreign exchange debts and if imported intermediate goods play an important role in production. Still, the trauma is typically less, because it does not involve a loss of credibility for the government or the central bank. The persistent and recurring nature of financial crises in various guises through the centuries makes us skeptical about providing easy answers as to how to best avoid them. In our final chapter we sketch out some of the issues regarding the prospects for and measurement of graduation from these destabilizing boom-bust cycles.

Summary

This chapter has greatly extended our perspective of crises by illustrating quantitative measures of the global nature of a crisis, ranging from our composite index of global financial turbulence to comparisons of the aftermath of crises between the Great Depression of the past century and the recent Second Great Contraction. We have seen that by all measures, the trauma resulting from this contraction, the first global financial crisis of the twenty-first century, has been extraordinarily severe. That its macroeconomic outcome has been only the most severe global recession since World War II—and not even worse—must be regarded as fortunate.