Banking Crises and Economic Activity: Observations from Past Crises in Developed Countries*

Authors


  • *

    The views expressed in this paper are those of the author and are not necessarily those of the Reserve Bank of Australia.

Ellis Connolly, Economic Analysis Department, Reserve Bank of Australia, GPO Box 3947, Sydney, NSW 2001, Australia. Email: connollye@rba.gov.au

Abstract

This paper examines twelve banking crises that occurred in developed countries between the 1970s and 1990s and highlights the macroeconomic adjustment paths experienced during these episodes. The banking crises were generally preceded by financial deregulation, which sparked rapid real credit growth and asset price inflation. During the booms, the lending standards of banks deteriorated and prudential supervision was inadequate. The most serious crises coincided with deep and prolonged slowdowns in economic activity, starting with falling credit growth and asset prices, followed by a sharp fall in investment and a slowing in consumption, partly offset by a rise in net exports. Governments responded with a range of policies to restore the health of the banking sector, along with monetary and fiscal stimulus. The paper concludes by comparing these past crises with the recent banking crises in the United States and parts of Europe, where the deterioration of lending standards appears to have been an important factor once again.

1. Introduction

The recent global financial turmoil, in which around ten developed countries have experienced banking crises, has rekindled interest in past banking crises and their effects on economic activity.2Caprio and Klingebiel (2003) identified twenty banking crises that affected sixteen developed countries since the early 1970s. Although these past crises were not as synchronous as in the current episode, around half were clustered in the early 1990s when many developed countries experienced recessions (Figure 1). Caprio and Honohan (2008) highlight that over history, banking crises have tended to occur in waves such as these. Twelve of the crises were clearly associated with economic slowdowns and are the focus of this paper (Table 1 contains brief descriptions). The evolution of key macroeconomic variables for each of the crisis countries are examined to highlight the common elements, as well as isolating the country-specific factors and government policies that appeared to affect the depth of the economic downturns.

Figure 1.

 Banking Crises. Sources: Caprio and Klingebiel (2003); Caprio et al. (2005); Reinhart and Rogoff (2008b); Furceri and Mourougane (2009)

Table 1.   Banking Crises in Developed Economies
 Recapitalisation cost (% GDP)
  1. Sources: Caprio and Klingebiel (1996, 2003) and Caprio et al. (2005). Caprio and Klingebiel used judgment in timing the episodes of bank insolvency and the statistics are consensus numbers. For more detailed descriptions of the systemic crises, see Haugh et al. (2009).

Systemic
 Finland1991–943 banks (31% system deposits) nationalised11
 Japan1992–02Banks’ non-performing loans (NPLs) rose to 10–25% GDP24
 Norway1987–933 banks (85% system assets) nationalised 8
 Spain1977–8552 banks (20% system deposits) had solvency problems 6
 Sweden1991–945 banks (70% system assets) experienced difficulties 4
Non-systemic (with downturn)
 Australia1989–922 state banks recapitalised, NPLs rose to 6% assets 2
 Denmark1987–92Loan losses of 9% of loans; 40 of 60 problem banks merged 
 Greece1991–95Public funds injected into specialised lending institutions 
 Italy1990–9558 banks (11% of lending) merged with other institutions 
 NZ1987–90State-owned bank (25% assets) had solvency problems 1
 UK1974–76Secondary banking crisis following collapse of property market 
 USA1988–911400 Savings and Loan Associations and 1300 banks failed 3
Non-systemic (isolated/no downturn)
 Canada1983–8515 deposit taking institutions failed, including 2 banks 
 France1994–95Credit Lyonnais experienced serious solvency problems due to fraud 
 Germany1977–79“Giroinstitutions” faced problems 
 Iceland1985–86One of the three state-owned banks became insolvent 
 Iceland1993Government recapitalised large state-owned bank after loan losses 
 UK1984Failure of Johnson Matthey Bankers due to high-risk loans 
 UK1991Failure of Bank of Credit and Commerce International (BCCI) due to fraud 
 UK1995Failure of Barings Bank due to fraud 

The crises have varied significantly in terms of their magnitude and length, and there is inevitably some imprecision involved in defining and dating them.3 Five of the crises identified by Caprio and Klingebiel (2003) are considered to have been systemic, as a large share of each country’s banking system became illiquid or insolvent, requiring costly government intervention. In these countries, the recapitalisation cost to the government ranged from 4 to 24 per cent of GDP, although Reinhart and Rogoff (2008b) argue that the fiscal burden of such banking crises extended well beyond the recapitalisation cost due to the effects of the associated economic downturns on government revenue and expenses.4 Seven countries experienced non-systemic crises associated with economic downturns, which all involved bank insolvencies that affected a smaller share of the banking system and required less government intervention. Among the more serious of these was the Savings & Loan Crisis in the United States in 1988–1991. At the other end of the spectrum, the remaining eight crises were typically isolated to just one institution and did not appear to significantly disrupt economic activity.

2. Conditions Prior to the Banking Crises

The twelve crises listed at the top of Table 1 share some common causes. Each followed economic booms that generated positive output gaps, with rapid asset price inflation and real credit growth (Table 2). The rapid expansion of bank balance sheets often followed financial liberalisation, which exposed banks and regulators to risks for which they were unprepared (Honohan and Laeven, 2005). It is notable that economic conditions prior to the crises were quite similar in the countries which experienced systemic or less severe crises.

Table 2.   Conditions Prior to Banking Crises with Economic Slowdowns
Banking crisesPeak in GDP growth rate (%)Peak in output gap (% GDP)Peak in real credit growth rate (%)Peak in interest rate§ (%)Current account (% GDP in pre-crisis year)
RealNominal
  1. Notes: Output gap estimates from OECD except for Spain pre-1978, which is a deviation from HP-filtered trend.

  2. Domestic credit; data adjusted for noticeable breaks: Finland (December 1988), Spain (March 1972) and Greece (March 1987).

  3. §Short-term money market or Treasury bill rates; real rates are nominal rates less CPI inflation.

  4. Sources: ABS, Bank of Italy, Thomson Reuters, IMF, OECD, Norges Bank, RBA, RBNZ, author’s calculations.

Systemic
 Finland (1991–94)6716916−5
 Japan (1992–02)7511582
 Norway (1987–93)7410915−6
 Spain (1977–85)9313−315−4
 Sweden (1991–94)4315614−2
Non-systemic
 Australia (1989–92)61161018−4
 Denmark (1987–92)6425611−5
 Greece (1991–95)528624−4
 Italy (1990–95)5211613−1
 NZ (1987–90)8315725−6
 UK (1974–76)10420112−1
 USA (1988–91)421158−3

The wave of banking crises that occurred in the 1970s and 1980s tended to follow the deregulation of financial markets. Prior to deregulation, banks had been losing market share in some countries to more lightly regulated non-bank financial intermediates due to financial innovation.5 The deregulation of banks then sparked the rapid expansion of credit, as the private sector responded to the greater availability of finance by borrowing to purchase assets. This contributed to share market booms in all the countries examined, along with property price booms, particularly in Japan and the Nordic countries. Rapid credit growth placed stress on the ability of banks to properly appraise credit risk, and more prudent banks felt pressured to relax their lending standards to maintain their market share (Honkapohja, 2008; Caprio and Honohan, 2008).

Failures of banking supervision in the deregulated environment also played a role in the development of imbalances. The quality of supervision may also have been an important factor in explaining why some countries experienced systemic crises while others did not. In the early 1990s, Denmark was the only Nordic country that avoided a systemic crisis. Honkapohja (2008) argues that this was partly due to stricter prudential supervision, which meant that Danish banks were in a better position to raise private capital when asset prices fell, rather than requiring recapitalisation by the government.

In many countries, the deregulation of the banking system also coincided with capital account liberalisation, giving banks greater access to foreign capital to fund the growth in credit. As a result, most of the countries were running significant current account deficits over the year prior to when the crisis commenced. Central banks were tightening interest rates in response to the overheating economies, with real interest rates peaking at high levels (except in the 1970s crises). However, some of the countries liberalised their capital accounts while maintaining fixed exchange rates, rendering monetary policy ineffective. For instance, in Finland and Sweden, where the currencies were pegged to the Deutsche mark, the central banks had to sharply raise interest rates from 1990 to 1992 to maintain the interest differential with Germany (where monetary policy was being tightened), encouraging further capital inflow (Honkapohja, 2008).6

3. Economic Slowdowns Associated with Banking Crises

In the countries that suffered systemic crises, asset prices and credit growth tended to peak at around the same time, a couple of years prior to the start of the crisis. Equity prices then fell precipitously, with four of the five countries experiencing real share price falls of greater than 50 per cent (Table 3; Norway was the exception, only experiencing a temporary fall in share prices in 1987). Several countries also experienced prolonged falls in property prices, particularly Japan and the Nordic countries.

Table 3.   Economic Slowdowns associated with Banking Crises. Peak to Trough Change in Each Variable Around the Time of the Banking Crisis
Banking crisesReal share prices (% change)Real credit growth (%pts change)GDP growth (%pts change)Output gap (%pts change)Years from peak to trough in output gap
  1. Sources: ABS, Bank of Italy, Thomson Reuters, IMF, OECD, Norges Bank, RBA, RBNZ, author’s calculations.

Systemic
 Finland (1991–94)−71−30−14−19
 Japan (1992–02)−57−14−8−5
 Norway (1987–93)−35−15−9−11
 Spain (1977–85)−90−18−9−78
 Sweden (1991–94)−52−32−6−10
Non-systemic
 Australia (1989–92)−51−20−8−7
 Denmark (1987–92)−37−34−10−5
 Greece (1991–95)−70−14−10−54
 Italy (1990–95)−52−5−7−6
 NZ (1987–90)−59−15−12−5
 UK (1974–76)−75−36−13−8
 USA (1988–91)−21−13−5−4

When asset prices begin to fall after a credit boom, there can be a breakdown of trust between banks and their creditors over the scale of losses on the less credit-worthy lending that took place, reducing the funding available to banks. The banks also face an adverse selection problem, with potential borrowers during a downturn more likely to be in financial distress. In response, banks seek to reduce their leverage by reducing credit supply and they also charge higher risk premiums to borrowers. The end result can be a damaging feedback loop from the financial sector to the real economy, which is exacerbated if financial institutions fail (see Greenlaw et al., 2008, Reinhart and Rogoff, 2008b; Furceri and Mourougane, 2009; Haugh et al., 2009). Consistent with this, in the systemic crises, financial institutions sharply slowed lending, with real credit growth rates falling by around fifteen to thirty percentage points over the following years.

The economic downturns associated with systemic crises tended to be both deep and protracted, followed by a muted recovery.7 The output gap tended to peak around one year after the peak in asset prices and credit growth before falling by around ten percentage points, taking around three to four years to reach the trough. While Japan appeared to experience a relatively mild fall in the output gap (Table 3), this was just the first slowdown associated with the banking crisis (Figure 2). Non-performing loans continued to grow as a share of GDP through the mid-1990s and there was a second sharp slowdown in the growth of output and credit in the late 1990s.

Figure 2.

 Japan. Sources: IMF; Thomson Reuters; Author’s Calculations

The process was broadly similar in the countries that experienced non-systemic crises. In most countries, there were large falls in asset prices, real credit growth and the output gap. The main difference between these crises and the systemic crises was the depth of the slowdowns, with the non-systemic crises involving around a five percentage point fall in the output gap. The length of the slowdown in credit growth was also shorter, with credit growth tending to trough within one to three years in the non-systemic crises (compared with around five years in the systemic crises). Serious non-systemic crises occurred in Australia in the early 1990s following a commercial property price bubble (Figure 3) and the UK in the mid-1970s following the first OPEC oil shock.

Figure 3.

 Australia. Sources: ABS, RBA; Author’s Calculations

It is not straightforward to identify whether the banking crises caused the downturns, particularly as some of the countries were also experiencing country-specific shocks around the same time. For instance, one reason why Finland experienced such a deep downturn during its banking crisis was the collapse of the Soviet Union in 1991, which resulted in a very sharp decline in exports to Russia (Figure 4). Finnish exports were uncompetitive in western markets at the pegged exchange rate, leading to a harsh adjustment process for the exports sector (Honkapohja, 2008). By contrast, Norway experienced a negative external shock much earlier than Finland when oil prices fell significantly in 1986. Honkapohja (2008) argues that the timing of this shock was fortuitous, preventing a longer-lasting boom as occurred in Finland and therefore reducing the size of the subsequent economic downturn.

Figure 4.

 Finland. Sources: IMF; Thomson; Reuters; Author’s Calculations

3.1 Adjustment of Domestic Spending

To highlight how the composition of domestic spending adjusted around banking crises, we examine the change in the share of investment in GDP and household consumption growth between the peak and the trough in the output gap (Table 4). Most of the adjustment to domestic spending appeared to occur through large falls in the investment share in GDP.8 This was particularly the case in the systemic crises and in Australia in the early 1990s (Figure 5). This would partly reflect firms responding to reduced demand, along with “financial accelerator effects”, where the deterioration in credit market conditions lowers firms’ access to credit, forcing them to scale back their investment plans (see Bernanke et al., 1996; Krugman, 2008), along with reduced demand from the corporate sector. The slowing in consumption growth was of a broadly similar magnitude to the slowing in GDP growth during most of the crises, with consumption subdued as a result of negative wealth effects from the falls in asset prices as well as reduced access to finance.

Table 4.   Expenditure and the Current Account. Change in Each Variable from the Peak to the Trough in the Output Gap
Banking crisesHousehold consumption growth (%pts change)Investment share (%pts GDP change)Trade balance (%pts GDP change)Current account (%pts GDP change)
  1. Sources: IMF, OECD, author’s calculations.

Systemic
 Finland (1991–94)−10−1385
 Japan (1992–02)−1−412
 Norway (1987–93)−9−51311
 Spain (1977–85)−6−522
 Sweden (1991–94)−7−51−4
Non-systemic
 Australia (1989–92)−4−732
 Denmark (1987–92)−10−245
 Greece (1991–95)−7−203
 Italy (1990–95)−7−332
 NZ (1987–90)−9−434
 UK (1974–76)−61−1−2
 USA (1988–91)−2−211
Figure 5.

 Australia. Sources: IMF; OECD; Author's Calculations

3.2 External Adjustment and the Exchange Rate

Consistent with falling domestic demand, the crisis countries tended to experience a significant rise in their trade balance due to falling imports (Table 4). This resulted in a narrowing of the current account deficit, consistent with a reduction in foreign capital inflows relative to the peak of the boom. The other potential method for external adjustment was through a depreciation of the exchange rate, which was an important shock absorber (Broda, 2004). However, most of the countries did not have floating exchange rates at the time their banking crisis started, and the authorities did not allow the exchange rate to depreciate (Table 5). Two exceptions were Finland and Sweden, which experienced banking and currency crises simultaneously. The authorities were forced to abandon their fixed exchange rate regimes as a result of speculative attacks during the Exchange Rate Mechanism (ERM) crisis of 1992, leading to significant depreciations (Figure 6). Overall, Finland’s real exchange rate depreciated by around 30 per cent, while Sweden’s depreciated by around 20 per cent, contributing to the narrowing of their current account deficits over 1993 by stimulating net exports (Honkapohja, 2008; Haugh et al., 2009).9 Looking at the countries with floating currencies, only Australia and the United States experienced depreciations, both of which were mild. The depreciation of the Australian dollar was probably also related to the fall in the terms of trade in the early 1990s.

Table 5.   Exchange Rates, Monetary Policy and Inflation. Change in Each Variable from the Peak to the Trough in the Output Gap
Banking crisesExchange rate regime at start of crisisReal exchange rate†‡ (%pts change)Real interest rate (%pts change)Inflation rate lagged one year relative to output gap (%pts change)
  1. Notes: IMF Real Effective Exchange Rates based on relative consumer prices; data unavailable for Spain and the United Kingdom during those crises.

  2. Quasi-fixed pegs under the Exchange Rate Mechanism of the European Monetary System.

  3. Sources: IMF, OECD, Forssbæck and Oxelheim (2006), RBA, author’s calculations.

Systemic
 Finland (1991–94)Adjustable peg−260−5
 Japan (1992–02)Floating27−3−1
 Norway (1987–93)Fixed peg4−3−6
 Spain (1977–85)Adjustable peg 5−5
 Sweden (1991–94)Fixed peg−31−5
Non-systemic
 Australia (1989–92)Floating−9−6−3
 Denmark (1987–92)ERM peg73−1
 Greece (1991–95)Managed float19−1−10
 Italy (1990–95)ERM peg−9−2−3
 NZ (1987–90)Floating225−10
 UK (1974–76)Floating −11−2
 USA (1988–91)Floating−5−2−2
Figure 6.

 Twin Banking and Currency Crises. Sources: IMF; OECD; Author's Calculations

4. Government Policy Responses

The length of the crises appears to depend in part on the policies implemented by governments in addressing the crises. In the early stages, central banks would typically provide liquidity support, and where the solvency of some institutions was in doubt, governments would take an ad hoc approach, encouraging mergers and takeovers, or injecting equity. When such measures failed to rebuild confidence, governments tended to adopt system-wide measures such as issuing a blanket guarantee of bank liabilities, announcing recapitalisation programmes and shifting troubled assets into a government-owned asset management company (Claessens et al., 2001; Honkapohja, 2008; Furceri and Mourougane, 2009). If governments adopted such an approach relatively quickly, as in the Nordic countries, the economy tended to recover earlier than when governments persisted with a case-by-case approach, as occurred in Japan from 1992 to 1997. When the Japanese banking crisis escalated with the Asian financial crisis, the government then launched a more comprehensive bailout package, but the recovery was very slow, with the Japanese banking sector not returning to profitability until 2003 (Haugh et al., 2009).

Governments also used fiscal and monetary policies to address the economic downturns associated with the banking crises. In most countries, monetary policy was loosened, as indicated by a fall in the real interest rate (Table 5). The exceptions tended to be countries with fixed exchange rates, many of which maintained higher interest rates to support their peg. Stimulatory monetary policy during the crises did not appear to result in a build up of inflationary pressures. The inflation rate fell in each country, consistent with the slowing in domestic demand, with the trough in the inflation rate tending to lag the trough in the output gap by around a year. For those countries that experienced significant depreciations, the fall in the output gap appeared to be large enough to more than offset the rise in tradables inflation. Governments tended to run large budget deficits during the crises to recapitalise the banking system and provide fiscal stimulus, and also as a result of the operation of the automatic stabilisers. Unsurprisingly, public debt increased significantly as a result of banking crises (Reinhart and Rogoff, 2008b; Haugh et al., 2009).

5. Comparisons with the Recent Banking Crises

There are some clear parallels between past banking crises and the recent crises in the United States and several European countries. Financial innovation, in the form of the rapid growth and diversification of asset-backed securities markets, exposed banks to risks. Particularly in the United States, lending standards weakened significantly. This culminated in the sub-prime lending boom, where more marginal borrowers were offered mortgages at low initial interest rates that were scheduled to reset some years later at higher rates than the borrowers could afford. These loans were then bundled into asset-backed securities and purchased by foreign banks in a range of countries, particularly in Europe, without adequate due diligence being performed. Consistent with past crises, the downturn has been severe and the recovery is generally expected to be slow. The governmental response has also followed a similar pattern to previous crises, starting with an ad hoc approach before shifting to a more comprehensive approach following the collapse in confidence in late 2008.

The major difference between the crises and the past episodes has been the highly synchronised global downturn. While around ten developed countries have experienced banking crises as a result of the deterioration of lending standards, virtually all countries have experienced an economic downturn. This is partly due to the increased integration of global capital markets, along with the crises having a greater effect on the confidence and behaviour of households as a result of higher indebtedness and larger falls in wealth (see Kent and Lawson, 2007; Besley et al., 2008). As the crises have led to a global downturn, there is less scope for the exchange rates of the most affected countries to depreciate to stimulate growth in the trade sector. At the same time, the affected countries have not experienced currency crises (with the exception of Iceland) as occurred in the Nordic countries in the early 1990s.

Footnotes

  • 2

    For instance, see Reinhart and Rogoff (2008a). Furceri and Mourougane (2009) list ten developed countries where financial institutions have recently failed.

  • 3

    Boyd et al. (2008) suggest that the dating of banking crises often lags the onset of the crises, as they are dated as having commenced with the first government response, rather than with the shocks that triggered the crises.

  • 4

    On the other hand, Honkapohja (2008) points out that the recapitalisation cost of the crises in Sweden, Norway and Finland were at least partly recouped by the governments when the assets were sold back to the private sector.

  • 5

    This was particularly the case for the Savings and Loan Associations in the United States during the 1970s: Haugh et al. (2009).

  • 6

    Reinhart and Rogoff (2008b) find that periods of high international capital mobility are often associated with international banking crises.

  • 7

    Haugh et al. (2009) also draw this conclusion by comparing the downturns associated with the five systemic crises and the Savings and Loan Crisis with other downturns experienced by these countries.

  • 8

    Investment disaggregated by dwellings, private and public, was not available for many of the countries in this paper. Haugh et al. (2009) examined the five systemic crises and the Savings and Loan Crisis and concluded that business and housing investment fell considerably more during banking crises than during other economic downturns.

  • 9

    Furceri and Mourougane (2009) found it hard to identify a common pattern between currency depreciations and banking crises.

Ancillary