This Time Is Different: Eight Centuries of Financial Folly Read online

  Governments frequently can and do make the financial repression tax even larger by maintaining interest rate caps while creating inflation. For example, this is precisely what India did in the early 1970s when it capped bank interest at 5 percent and engineered an increase in inflation of more than 20 percent. Sometimes even that action is not enough to satisfy governments’ voracious need for revenue savings, and they stop paying their debts entirely (a domestic default). The domestic default forces banks, in turn, to default on their own liabilities so that depositors lose some or all of their money. (In some cases, the government might actually have issued deposit insurance, but in the event of default it simply reneges on that promise, too.)

  Banking Crises and Bank Runs

  True banking crises, of the variety more typically experienced in emerging markets and advanced economies, are a different kind of creature. As we mentioned in the preamble, banks’ role in effecting maturity transformation—transforming short-term deposit funding into long-term loans—makes them uniquely vulnerable to bank runs.5 Banks typically borrow short in the form of savings and demand deposits (which can, in principle, be withdrawn at short notice). At the same time, they lend at longer maturities, in the form of direct loans to businesses, as well as other longer-dated and higher-risk securities. In normal times, banks hold liquid resources that are more than enough to handle any surges in deposit withdrawals. During a “run” on a bank, however, depositors lose confidence in the bank and withdraw en masse. As withdrawals mount, the bank is forced to liquidate assets under duress. Typically the prices received are “fire sale” prices, especially if the bank holds highly illiquid and idiosyncratic loans (such as those to local businesses about which it has far better information than other investors). The problem of having to liquidate at fire sale prices can extend to a far broader range of assets during a systemic banking crisis of the kind we focus on here. Different banks often hold broadly similar portfolios of assets, and if all banks try to sell at once, the market can dry up completely. Assets that are relatively liquid during normal times can suddenly become highly illiquid just when the bank most needs them.

  Thus, even if the bank would be completely solvent absent a run, its balance sheet may be destroyed by having to liquidate assets at fire sale prices. In such a case, the bank run is self-fulfilling. That is, it is another example of multiple equilibria, similar in spirit to when a country’s creditors collectively refuse to roll over short-term debt. In the case of a bank run, it is depositors who are effectively refusing to roll over debt.

  In practice, banking systems have many ways of handling runs. If the run is on a single bank, that bank may be able to borrow from a pool of other private banks that effectively provide deposit insurance to one another. However, if the run affects a broad enough range of institutions, private insurance pooling will not work. An example of such a run is the U.S. subprime financial crisis of 2007, because problematic mortgage assets were held widely in the banking sector. Exchange rate crises, as experienced by so many developing economies in the 1990s, are another example of a systemic financial crisis affecting almost all banks in a country. In crises represented by both of these examples, it is a real loss to the banking system that eventually sets off the shock. The shock may be manageable if confidence in the banking sector is maintained. However, if a run occurs, it can bankrupt the entire system, turning a damaging problem into a devastating one. Diamond and Dybvig argue that deposit insurance can prevent bank runs, but their model does not incorporate the fact that absent effective regulation, deposit insurance can induce banks to take excessive risk.6

  Bank runs, in general, are simply one important example of the fragility of highly leveraged borrowers, public and private, as discussed in the preamble to this book. The implosion of the U.S. financial system during 2007–2008 came about precisely because many financial firms outside the traditional and regulated banking sector financed their illiquid investments using short-term borrowing. In modern financial systems, it is not only banks that are subject to runs but also other types of financial institutions that have highly leveraged portfolios financed by short-term borrowing.

  Why Recessions Associated with Banking Crises Are So Costly

  Severe financial crises rarely occur in isolation. Rather than being the trigger of recession, they are more often an amplification mechanism: a reversal of fortunes in output growth leads to a string of defaults on bank loans, forcing a pullback in other bank lending, which leads to further output falls and repayment problems, and so on. Also, banking crises are often accompanied by other kinds of crises, including exchange rate crises, domestic and foreign debt crises, and inflation crises; we will explore the coincidence and timing of crises in more detail in chapter 16. Thus, one should be careful not to interpret this first pass at our long historical data set as definitive evidence of the causal effects of banking crises; there is a relatively new area in which much further work is yet to be done.

  That said, the theoretical and empirical literature on how financial crises can impact real activity is extremely broad and well developed. One of the most influential studies was reported in 1983 by Bernanke, who argued that when nearly half of all U.S. banks failed in the early 1930s, it took the financial system a long time to rebuild its lending capacity. According to Bernanke, the collapse of the financial system is a major reason that the Great Depression persisted, on and off, for a decade rather than ending in a year or two as a normal recession does. (Bernanke, of course, became Federal Reserve chairman in 2006 and had a chance to put his academic insights into practice during the Second Great Contraction, which began in 2007.)

  In later work with Mark Gertler, Bernanke presented a theoretical model detailing how the presence of imperfections in the financial market due to asymmetric information between lenders and borrowers can result in an amplification of monetary policy shocks.7 In the Bernanke-Gertler model, a decrease in wealth (due, say, to an adverse productivity shock) has an outsized effect on production as firms are forced to scale back their investment plans. Firms are forced to scale back on investment because, as their retained earnings fall, they must finance a larger share of their investment projects via more expensive external financing rather than by means of relatively cheap internal financing. Recessions cause a loss in collateral that is then amplified through the financial system.

  Kiyotaki and Moore trace out a similar dynamic in a richer intertemporal model.8 They show how a collapse in land prices (such as occurred in Japan beginning in the early 1990s) can undermine a firm’s collateral, leading to a pullback in investment that causes a further fall in land prices, and so on.

  In his 1983 article, Bernanke emphasized that the collapse of the credit channel in recessions is particularly acute for small and medium-sized borrowers who do not have name recognition and therefore have far less access than larger borrowers to bond and equity markets as an alternative to more relationship-oriented bank finance. Many subsequent papers have confirmed that small and medium-sized borrowers do suffer disproportionately during a recession, with a fair amount of evidence pointing to the bank lending channel as a central element.9 We will not dwell further on the vast theoretical literature on financial markets and real activity except to say that there is indeed significant theoretical and empirical support for the view that a collapse in a country’s banking system can have huge implications for its growth trajectory.10

  We now turn to the empirical evidence. Given the vulnerability of banking systems to runs, combined with the theoretical and empirical evidence that banking crises are major amplifiers of recessions, it is little wonder that countries experience greater difficulties in outgrowing financial crises than they do in escaping a long history of sovereign debt crises. In the latter it is possible to speak of “graduation,” with countries going for centuries without slipping back into default. But thus far, no major country has been able to graduate from banking crises.

  Banking Crises: An Equal-Opportunity
Menace

  As shown earlier, the frequency of default (or restructuring) on external debt is significantly lower in advanced economies than in emerging markets. For many high-income countries, that frequency has effectively been zero since 1800.11 Even countries with a long history of multiple defaults prior to 1800, such as France and Spain, present evidence of having graduated from serial default on external debt.

  The second column in tables 10.1 and 10.2 highlights the vast difference in the experience of sovereign default between emerging markets (notably in Africa and Latin America but even in several countries in Asia) and high-income Western Europe, North America, and Oceania. The third column of tables 10.1 and 10.2 presents the analogous calculation for each country for banking crises (i.e., the number of years in banking crises, according to the extended data set developed here, divided by the number of years since the country won independence, or since 1800 if it achieved independence earlier). One striking observation from the tables is that the average length of time a country spends in a state of sovereign default is far greater than the average amount of time spent in financial crisis. A country can circumvent its external creditors for an extended period. It is far more costly to leave a domestic banking crisis hanging, however, presumably due to the crippling effects on trade and investment.

  TABLE 10.1

  Debt and banking crises: Africa and Asia, year of independence to 2008

  TABLE 10.2

  Debt and banking crises: Europe, Latin America, North America, and Oceania, year of independence to 2008

  Tables 10.3 and 10.4 present a different perspective on the prevalence of banking crises. The second column tallies the number of banking crises (rather than the number of years in crisis) since a country’s independence or 1800; the third column narrows the window to the post–World War II period. Several features are worth noting. For the advanced economies over the full span, the picture that emerges is one of serial banking crises. The world’s financial centers—the United Kingdom, the United States, and France—stand out in this regard, with 12, 13, and 15 episodes of banking crisis since 1800, respectively. The frequency of banking crises dropped off markedly for the advanced economies and the larger emerging markets alike after World War II. However, all except Portugal experienced at least one postwar crisis prior to the recent episode. When the recent wave of crises is fully factored in, the apparent drop will likely be even less pronounced. Thus, although many now-advanced economies have graduated from a history of serial default on sovereign debt or very high inflation (above 20 percent), so far graduation from banking crises has proven elusive. As we will show later, the same applies to currency crashes. Indeed, tables 10.1–10.4 illustrate that despite dramatic differences in recent sovereign default performance, the incidence of banking crises is about the same for advanced economies as for emerging markets. It also should be noted that as financial markets have developed in the smaller, poorer economies, the frequency of banking crises has increased.12

  TABLE 10.3

  Frequency of banking crises: Africa and Asia, to 2008

  TABLE 10.4

  Frequency of banking crises: Europe, Latin America, North America, and Oceania, to 2008

  Tables 10.5 and 10.6 summarize, by region, the evidence on the number of banking crises and the share of years each region has spent in a banking crisis. Table 10.5 starts in 1800. (The table includes postindependence crises only, which explains why emerging markets have lower cumulative totals.) Table 10.6 gives the evidence for the period since 1945.

  Whether the calculations are done from 1800 (table 10.5) or from 1945 (table 10.6), on average there are no significant differences in either the incidence or the number of banking crises between advanced and emerging economies; indeed banking crises plague both sets of countries. In fact, prior to World War II, the advanced economies with their more developed financial systems were more prone to banking crises than were many of their smaller low-income counterparts.13 Of course, it can be plausibly argued that smaller countries used foreign creditors as their bankers, and therefore the string of defaults on external debts might have been domestic banking crises had they more developed financial sectors.

  TABLE 10.5

  Summary of the incidence and frequency of banking crises, 1800 (or independence) to 2008

  TABLE 10.6

  Summary of the incidence and frequency of banking crises, 1945 (or independence) to 2008

  Banking Crises, Capital Mobility,

  and Financial Liberalization

  Also consonant with the modern theory of crises is the striking correlation between freer capital mobility and the incidence of banking crises, as shown in figure 10.1. The figure is highly aggregated, but a breakdown to regional or country-level data reinforces the message of the figure. Periods of high international capital mobility have repeatedly produced international banking crises, not only famously, as they did in the 1990s, but historically. The figure plots a three-year moving average of the share of all countries experiencing a banking crisis on the right-hand scale. On the left-hand scale we have graphed the index of international capital mobility, using the same design principle as Obstfeld and Taylor, both updated and cast back in time, to cover our full sample period.14 Although the Obstfeld-Taylor index may have its limitations, we feel it nevertheless provides a concise summary of complicated forces by emphasizing de facto capital mobility based on actual flows.

  For the period after 1970, Kaminsky and Reinhart have presented formal evidence of the link between crises and financial liberalization.15 In eighteen of the twenty-six banking crises they studied, the financial sector had been liberalized within the preceding five years, usually less. In the 1980s and 1990s, most liberalization episodes were associated with financial crises of varying severity. In only a handful of countries (for instance, Canada) did liberalization of the financial sector proceed smoothly. Specifically, Kaminsky and Reinhart present evidence that the probability of a banking crisis conditional on financial liberalization having taken place is higher than the unconditional probability of a banking crisis. Using a fifty-three-country sample for the period 1980–1995, Demirgüç-Kunt and Detragiache also show, in the context of a multivariate logit model, that financial liberalization has an independent negative effect on the stability of the banking sector and that this result is robust across numerous specifications.16

  Figure 10.1. Capital mobility and the incidence of banking crises: All countries, 1800–2008.

  Sources: Kaminsky and Reinhart (1999), Bordo et al. (2001), Obstfeld and Taylor (2004), Caprio et al. (2005), and the authors’ calculations.

  Notes: This sample includes all countries (even those not in our core sample of sixty-six). The full listing of the dates of banking crises appears in appendixes A.3 and A.4. This figure shows that the recovery in equities is far swifter than that of the housing market. On the left-hand scale we updated our favorite index of capital mobility, admittedly arbitrary but a concise summary of complicated forces. The dashed line shows the index of capital mobility given by Obstfeld and Taylor (2004), backcast from 1800 to 1859 using the same design principle they used.

  The stylized evidence presented by Caprio and Klingebiel suggests that inadequate regulation and lack of supervision at the time of liberalization may play a key role in explaining why deregulation and banking crises are so closely entwined.17 Again, this is a theme across developed countries and emerging markets alike. In the 2000s the United States, for all its this-time-is-different hubris, proved no exception, for financial innovation is a variant of the liberalization process.

  Capital Flow Bonanzas, Credit Cycles, and Asset Prices

  In this section we examine some of the common features of banking crises across countries, regions, and time. The focus is on the regularities among cycles in international capital flows, credit, and asset prices (specifically, housing and equity prices).

  Capital Flow Bonanzas and Banking Crises

  One common feature of the run-u
p to banking crises is a sustained surge in capital inflows, which Reinhart and Reinhart term a “capital flow bonanza.” They delineate a criterion to define a capital flow bonanza (roughly involving several percent of GDP inflow on a multiyear basis), catalog (country-by-country) “bonanza” episodes for 1960–2006, and examine the links between bonanza spells and banking crises.18 They employ the crises as defined and dated in appendix A.3.19

  From the dates of banking crises and capital flow bonanzas, two country-specific probabilities can be calculated: the unconditional probability of a banking crisis and the probability of a banking crisis within a window of three years before and after a bonanza year or years—that is, the conditional probability of a crisis. If capital flow bonanzas make countries more crisis prone, the conditional probability of a crisis, P(Crisis Bonanza), should be greater than the unconditional probability, P(Crisis).

  Table 10.7 reproduces a subset of the results given by Reinhart and Reinhart that are relevant to banking crises.20 It presents aggregates of the country-specific conditional and unconditional probabilities for three groups (all countries, high-income countries, and middle- and low-income countries). The probability of a banking crisis conditional on a capital flow bonanza is higher than the unconditional probability. The bottom row of table 10.7 provides the share of countries for which P(Crisis Bonanza) = P(Crisis) as an additional indication of how commonplace it is across countries to see bonanzas associated with a more crisis-prone environment. The majority of countries (61 percent) register a higher propensity to experience a banking crisis around bonanza periods; this percentage would be higher if one were to include post-2007 data in the table. (Many countries that have experienced the most severe banking crises during the late 2000s also ran large sustained current account deficits in the run-up to the crisis. These include many developed countries, such as Iceland, Ireland, Spain, the United Kingdom, and the United States.)