Chapter 13 The U.S. Subprime Crisis

1. As indicated in note 7 to the preamble, we use the term “Second Great Contraction” after Friedman and Schwartz's (1963) depiction of the 1930s as “The Great Contraction.” See also Felton and Reinhart (2008, 2009), who use the term “First Global Financial Crisis of the 21st Century.”

2. See chapter 10 for further discussion.

3. We have explored this issue further in chapter 10.

4. Although China's heavy-handed capital controls shielded it from contagious currency crashes during Asia's turmoil, they did not protect it from a systemic and costly banking crisis emanating primarily from large-scale lending to inefficient and bankrupt state-owned enterprises.

5. Figure 13.1 does not fully capture the extent of the present upsurge in financial crises, for Ireland and Iceland (both of which are experiencing banking crises at the time of this writing) are not part of our core sixty-six-country sample.

6. The Case-Shiller index is described by Robert Shiller (2005) and in recent years has been published monthly in conjunction with Standard and Poor's (as described at their Web site, www.standardandpoors.com). The Case-Shiller index focuses on resales of the same houses and therefore is arguably a more accurate gauge of price movements than indexes that look at all sales. Of course, there are many biases even in the Case-Shiller index (e.g., it is restricted to major metropolitan areas). Nevertheless, it is widely regarded as the most accurate gauge of changes in housing prices in the United States.

7. The Case-Shiller index appears to paint a quite plausible history of housing prices, but as a caveat we note that construction of the series required a significant number of assumptions to interpolate data missing for some intervals, particularly prior to World War II.

8. The current account balance is basically a broader measure of the trade balance—imports minus exports—extended to include investment returns. Note that the current account represents the sum of both government and private borrowing flows from abroad; it is not the same thing as the government deficit. It is perfectly possible for the government to be running a fiscal deficit and yet for the current account to be in surplus, provided the private savings compensate.

9. Greenspan (2007).

10. Economist Magazine, “The O'Neill Doctrine,” lead editorial, April 25, 2002.

11. Bernanke (2005).

12. See Philippon (2007).

13. Securitization of mortgages involves the bunching and repackaging of mortgage pools to transform highly idiosyncratic individual loans into more standardized products. Thus, to the extent that the U.S. current account was being driven by superior U.S. financial innovation, there was also nothing to worry about. Or so top U.S. financial regulators maintained.

14. See Obstfeld and Rogoff (2001, 2005, 2007).

15. Obstfeld and Rogoff (2001).

16. Roubini and Setser (2004).

17. Krugman (2007). Wile E. Coyote is the hapless character from Chuck Jones's Road Runner cartoons. His schemes invariably fail, and, as he runs off a cliff, there is a moment or two before the recognition sets in that nothing is below him.

18. See Obstfeld and Rogoff (2009) for a more detailed discussion of the literature; see also Wolf (2008).

19. Dooley et al. (2004a, 2004b).

20. Cooper (2005).

21. Hausmann and Sturzenegger (2007).

22. Curcuru et al. (2008) argue that the “dark matter” hypothesis is at odds with the data.

23. See Bernanke and Gertler (2001).

24. Bordo and Jeanne (2002), Bank for International Settlements (2005).

25. See Rolnick (2004).

26. We first noted the remarkable similarities between the 2007 U.S. subprime crisis and other deep financial crises in Reinhart and Rogoff (2008b), first circulated in December 2007. By the time of this writing, of course, the facts overwhelmingly support this reading of events. Our sources have included Caprio and Klingebiel (1996 and 2003), Kaminsky and Reinhart (1999), and Caprio et al. (2005).

27. Later we look at some alternative metrics for measuring the depth of these financial crises, arguing that the traditional measure—fiscal costs of the bank cleanup—is far too narrow.

28. See, for example, Kaminsky, Lizondo, and Reinhart (1998) and Kaminsky and Reinhart (1999).

29. For the United States, as earlier in this chapter, house prices are measured by the Case-Shiller index. The remaining house price data were made available by the Bank for International Settlements and are described by Gregory D. Sutton (2002). Of course, there are many limitations to the international housing price data; they typically do not have the long history that allows for a richer comparison across business cycles. Nevertheless, they probably reasonably capture our main variable of interest, peak-to-trough falls in the price of housing, even if they perhaps exaggerate the duration of the fall, because they are relatively slow to reflect changes in underlying market prices.

30. For the United States, the index is the S&P 500.

31. According to Reinhart and Reinhart (2009), during 2005–2007 the U.S. episode qualified as a “capital flow bonanza” (i.e., a period of abnormally large capital inflows, which is a different way of saying above-average borrowing from abroad).

32. In principle, the rise in real public debt is determined by taking the rise in nominal public debt and adjusting for the rise that represents inflation in all prices.

33. See the conclusions of Reinhart and Reinhart (2008), who explain these changes in interest rates and exchange rates as anomalies for the United States—because the United States is too big to fail.

Chapter 14 The Aftermath of Financial Crises

1. Also included in the comparisons are two prewar episodes in developed countries for which we have housing price and other relevant data.

2. To be clear, peak-to-trough calculations are made on an individual series-by-series basis. The trough and peak dates are those nearest the crisis date and refer to the local (rather than global) maximum or minimum, following much the same approach pioneered by Burns and Mitchell (1946) in their classic study of U.S. business cycles. So for example, in the case of Japan's equity prices, the trough is the local bottom in 1995, even though the subsequent recovery in the equity market left prices well below their prior peak before the crisis (and that the subsequent troughs would see prices at lower levels still).

3. In chapter 10, we looked at financial crises in sixty-six countries over two hundred years, emphasizing the broad parallels between emerging markets and developing countries, including, for example, the nearly universal run-up in government debt.

4. The historical average, which is shaded in black in the diagram, does not include the ongoing crises.

5. Notably, widespread “underemployment” in many emerging markets and the vast informal sector are not fully captured in the official unemployment statistics.

6. Again, see Calvo (1998) and Dornbusch et al. (1995).

7. See International Monetary Fund (various years), World Economic Outlook, April 2002, chapter 3.

8. Other noteworthy comparisons and parallels to the Great Depression are presented in Eichengreen and O'Rourke (2009).

Chapter 15 The International
Dimensions of the Subprime Crisis

1. The IMF, of course, is effectively the global lender of last resort for emerging markets, which typically face severe strains in floating new debt during a crisis. Given the quadrupling of IMF resources agreed to at the April 2, 2009, London meeting of the Group of 20 heads of state (including those of the largest rich countries and the major emerging markets), world market panic about the risks of sovereign default have notably abated. The IMF guarantees apply only to government debt, however, and risk spreads on the corporate debt of emerging markets remain elevated as of mid-2009, with rates of corporate default continuing to rise. It remains to be seen to what extent, if any, these debt problems will spill over to governments through bailouts, as they often have in the past.

2. Kaminsky, Reinhart, and Végh (2003); quote on p. 55, emphasis ours.

3. Bordo and Murshid (2001), Neal and Weidenmier (2003). Neal and Weidenmier emphasize that periods of apparent contagion can be more readily interpreted as responses to common shocks, an issue we return to in the context of the recent crisis. But perhaps the bottom line as regards a historical perspective on financial contagion is best summarized by Bordo and Murshid, who conclude that there is little evidence to suggest that cross-country linkages are tighter in the aftermath of a financial crisis for the recent period as opposed to 1880–1913, the earlier heyday of globalization in financial markets that they study.

4. Table 15.1 does not include the bunching of other “types” of crises, such as the wave of sovereign defaults during 1825 or the currency crashes or debasements of the Napoleonic Wars. Again, the indexes developed in chapter 16 will allow us to capture this kind of bundling of crises across both countries and types of crises.

5. See Neal and Weidenmier (2003) and Reinhart and Rogoff (2008a).

6. Owing to the opaqueness of balance sheets in many financial institutions in these countries, the full extent of exposure is, as yet, unknown.

7. See Reinhart and Reinhart (2009) for a full listing of episodes of capital inflow bonanzas.

Chapter 16 Composite Measures of Financial Turmoil

1. Kaminsky and Reinhart (1999).

2. The tally would come to six varieties of crises if we included currency debasement. We do not follow this route for two reasons: first, there are far fewer sources of data across countries (about a dozen or so) on the metallic content of their currencies; second, the printing press displaced debasement and decoupled currencies in circulation from a metallic base with the rise of fiat money. Because the period we analyzed for the turbulence composite was after 1800 (when our dating of banking crises begins in earnest), the exclusion of debasement crises is not as troublesome as for 1300–1799, when debasement was rampant.

3. This goes back to the dichotomous measures of crises that we (and most studies) employ. Of course, it is possible to consider additional gradations of crises to capture some measure of severity.

4. As noted, one could easily refine this measure to include three categories, say, high inflation (above 20 percent but less than 40), very high inflation (above 40 percent but less than 1,000), and hyperinflation (1,000 percent or higher).

5. Namely, crash episodes associated with international financial crises and turbulence (mostly in advanced economies).

6. Our list of economic crises does not include a growth collapse crisis as defined by Barro and Ursúa (2008, 2009), which is an episode in which per capita GDP falls cumulatively by 10 percent or more. An important share of the crisis episodes we identify are candidates for this definition as well. We examine this issue later. Nor does our composite index of financial turbulence necessarily include all “sudden stop” episodes as defined by Guillermo Calvo and coauthors in several contributions (see references). The reader will recall that a sudden stop is an episode in which there is an abrupt reversal in international capital flows, often associated with loss of capital market access. It is noteworthy that most systemic banking crises past and present (the 2007 U.S. subprime crisis is an exception) have been associated with sudden stops. The same could be said of sovereign external defaults.

7. Barro and Ursúa (2009). They identify 195 stock market crashes for twenty-five countries (eighteen advanced economies and seven emerging markets) over 1869–2006.

8. Samuelson (1966).

9. The reader will recall from earlier chapters that our sixty-six-country sample accounts for about 90 percent of world GDP.

10. It is important to note that Austria, Germany, Italy, and Japan remained in default for varying durations after the end of the war.

11. See McConnell and Perez-Quiros (2000) and Blanchard and Simon (2001).

12. As in nearly all previous historical crises in Argentina, the 2001–2002 episode was followed by a crisis in its small neighbor, Uruguay.

13. The hyperinflation episodes are the most notorious, obviously, but the share of countries in the region with an annual inflation rate above 20 percent, thereby meeting our threshold for a crisis, hit a peak of nearly 90 percent in 1990!

14. Burns and Mitchell (1946). For more recent treatments of the early warning properties of equity markets in the context of crises, see Kaminsky et al. (1998), Kaminsky and Reinhart (1999), and Barro and Ursúa (2009).

15. International Monetary Fund (various years), World Economic Outlook.

16. Eichengreen and O'Rourke (2009) add trade to highlight the similarities while noting the difference in monetary policy response (specifically, central bank discount rates).

17. Maddison (2004).

18. League of Nations (various years), World Economic Survey.

19. See, for example, League of Nations (1944).

20. Although we have reliable trade data for most countries during World War II, there are sufficient missing entries to make the calculation of the world aggregate not comparable to other years during 1940–1947.

21. Kaminsky and Reinhart (1999).

22. Demirgüc-Kunt and Detragiache (1998).

23. Reinhart (2002).

24. Reinhart and Rogoff (2004) also examined the relationship between currency crashes and inflation as well as the timing of currency crashes and capital control (specifically, dual or multiple exchange rates).

25. Diaz-Alejandro (1985).

26. In contrast to other studies of banking crises, Kaminsky and Reinhart (1999) provide two dates for each banking crisis episode—the beginning of a banking crisis and the later peak.

27. See Goldstein and Turner (2004).

28. See Reinhart, Rogoff, and Savastano (2003a).

29. The second and third effects of the depreciation or devaluation of the currency listed earlier are less of an issue for advanced economies.