BIS Papers No 1 139. By Andrew G Haldane, Glenn Hoggarth and Victoria Saporta, 1)
Bank of England
1) Bank of England, Threadneedle Street, London, EC2R 8AH. We are grateful to Alastair Clark, Paul Tucker, Simon Buckle, Patricia Jackson and Ian Michael for comments. However, views expressed are not necessarily those of the Bank of England.
2) Other examples would include Eastern Europe in the 1920s and Israel in the 1980s.
1. Introduction
Financial stability is concerned with an economy’s saving-investment nexus. Deviations from the
optimal saving-investment plan generate a welfare cost. These deviations may arise from
inefficiencies in the functioning of the financial system or from instabilities in this system in the face of shocks. These welfare frictions are behaviourally distinct, though they are closely interlinked. There may also at times be a trade-off between the two. For example, an increase in competitiveness may accentuate the financial system’s vulnerability to shocks, while conversely guarantees to the safety of the system as a whole may reduce its efficiency. An extreme version of the latter was witnessed in the financial systems of eastern Europe and the former Soviet Union during the command economy period. Any potential trade-off between financial stability and efficiency may be reduced, however, by having in place an adequate financial infrastructure for intermediating flows of funds or settling payments, and for regulating the financial system.
These frictions in the financial system are a potential public policy concern. They may justify public policy oversight and/or intervention and have, as a result, long been reflected in the mandate of central banks. For example, the Federal Reserve System was set up in 1914 “to furnish an elastic currency” - act as lender of last resort - against a backdrop of 14 separate episodes of banking panic between 1793 and 1914. Through extended periods over the last two centuries, financial stability has clearly been the primary concern of central banks around the world. The Bank of England is no exception.
The welfare costs of financial instability are often closely associated with monetary instabilities.
Monetary instability may give rise to both inefficiencies and instabilities in the financial system. The Great Depression is a classic example of extreme financial instability that was, in part at least, induced by monetary instability. At the end of 1933, the number of banks in the United States was half the number that existed in 1929 (Bernanke (1983)), during which time money income had fallen by 53% and real income by 36% from their 1929 peak (Wood (1999)).
The link between monetary stability and financial inefficiencies is harder to pinpoint. But recent work by English (1999) offers some interesting insights. English’s analysis starts from the observation that financial sectors increase markedly in size in economies undergoing high inflation or hyperinflation. He quotes the example of Germany in the 1920s, where the number of bank employees doubled between 1920 and 1923, at the peak of the hyperinflation, before returning to their earlier levels in 1924 as inflation subsided.2 The reason for this is that agents switch out of (non-interest bearing) money balances to make greater use of banking services as inflation rises. But this is a socially wasteful switch, because financial services resources could otherwise be put to more productive uses. At high rates of inflation, the financial sector is above its socially efficient level. English estimates that a 10 percentage point rise in inflation has a welfare cost equal to around 1¼% of GDP working through this financial sector channel. This is a non-trivial cost. It is an example of monetary instabilities generating well-defined welfare inefficiencies from a financial system perspective.
There have been few (if any) studies evaluating precisely the direct welfare costs associated with
financial instability and financial inefficiency. There has been recent work, however, quantifying some of the costs of financial instability and inefficiency in terms of foregone output. One strand of the literature has quantified the costs of recent banking and currency crises. Crises are, almost by definition, episodic and extreme instances of financial instability. As evidence on the output costs of financial instability more broadly, they are therefore limited and partial. To draw an analogy with monetary stability, they are equivalent to quantifying the output costs of hyperinflation - extreme monetary events. Nonetheless, the studies are illuminating and there have been enough recent crisis episodes to reach some fairly robust conclusions. The World Bank (Caprio and Klingebiel (1996,1999)) for example, document 69 instances of “systemic” crises since the late 1970s.
What broad conclusions can we draw from these studies? First, it is clear that banking crises are
associated with periods of low output relative to various measures of pre-crisis trend levels. Using various measures of output loss, Hoggarth, Reis and Saporta (2001) estimate average output losses of 15-20% of annual GDP for a sample of 43 banking crises. Second, banking crises are not confined to developing economies. Twelve out of the 54 global banking crises documented by the IMF in an interesting study (World Economic Outlook, 1998) occurred in industrial countries. Moreover, banking crises in high-income countries have tended to last longer and, on some estimates, have been associated with greater cumulative output losses than crises in middle and low-income countries.
According to the same IMF study, the average crisis length in the sample industrial countries is
4.1 years compared to 2.8 years in the sample emerging economies. Using a different sample of
crises, Hoggarth et al (2001) confirm the IMF finding and report cumulative output losses of 24% and 14% of annual GDP, on average, in high and medium/low income countries respectively. Third, “twin crisis” episodes (when banking instability and sharp pressures on a country’s exchange rate occur at the same time) are associated with considerably higher output losses than “single” banking crises -between three and five times as large according to Hoggarth et al’s (2001) estimates. Similarly, estimates of the fiscal costs of banking crisis resolution are much larger when there is a twin crisis (Aziz, Camarazza and Salgado (2000), Hoggarth et al (2001)), especially when the exchange rate was previously fixed. This is consistent with the notion that the macroeconomic consequences of banking sector fragility are amplified when banking sector problems are intertwined with or, as some of the recent literature on the causes of the Asian crisis suggests, cause exchange rate vulnerability.3
A second strand of literature has looked at the effects of financial inefficiencies, and in particular the effects of financial development, on growth. Though empirical work on this issue began 30 years ago (Goldsmith (1969), McKinnon (1973)), the most compelling evidence is recent work looking across a broader cross section of countries (see Levine and King (1993) and Levine (1997)). This finds a statistically significant, behaviourally important and seemingly causal link between various measures of financial development and growth, even after controlling for other factors. These financial development measures usually include the proportion of credit allocated to the private versus the public sector and the size of financial intermediaries in relation to the economy as a whole. To give an idea of magnitudes, Levine (1997) offers the example of Bolivia. Had Bolivia’s financial depth been equal to the mean for all developing countries in 1960, this would have boosted its growth rate by 0.6% per year thereafter. This is a huge gain when accumulated over time.
There is also strong evidence of faster-growing countries being associated with larger non-bank
financial sectors and larger stock market capitalisation, though it is more difficult to tell causal stories about these relations. There is likewise no clear-cut evidence on the relationship between financia structure - or example, bank versus capital market-based financing - and economic growth. That awaits further research. But in general this cross-sectional evidence is strongly supportive of financial development enhancing growth and productivity, and in non-trivial magnitudes.
There is relatively little, if any, literature that weaves together these two strands: the welfare costs of financial instability on the one hand; the welfare benefits of financial efficiency on the other. There is, however, some work exploring the trade-off between the two in the context of financial liberalisation.
For example, Demirguc-Kunt and Detragiache (1998a) consider these relationships across a panel of countries from the 1980s onwards. They reach three intriguing conclusions. First, there is evidence that financial liberalisation - here identified to be the removal of interest rate controls - materially increases the estimated probability of banking crisis, by a factor of around four. Second, there is evidence of liberalisation reducing financial sector inefficiencies, for example, by lowering the return on equity in the banking sector and reducing its concentration ratio. But third, the effect of liberalisation on 3 Examples include the work on the so-called “third generation” models of crisis. The most well cited example, perhaps, is the theoretical work of Chang and Velasco (1999). This shows that foreign exchange illiquidity alone can result in bank runs which would then lead to the collapse of the currency regime (see also Section 4 below). In the empirical literature, Kaminsky and Reinhart (1999) have found that banking crises are a leading indicator of currency crises which is consistent with (but not conclusive evidence of) the former causing the latter.
140 BIS Papers No 1
the probability of banking crisis is mitigated if the institutional infrastructure is robust - for example, if contracts are enforceable, there is high quality supervision, an absence of corruption etc. Taking these three findings together, there appears to be some evidence of an important trade-off between financial efficiency and financial stability.4 This trade-off can be improved, however, by improvements in the infrastructure of the financial system, including its prudential oversight by the authorities. This type of analytical framework, linking together financial efficiency and stability with the system’s infrastructure, warrants further theoretical analysis.
In the following sections we discuss all three pieces of the financial stability jigsaw. In Section 2 we discuss some of the techniques the Bank of England uses in the course of its surveillance of financial stability risks. A key issue here is aggregation: how to measure aggregate system-wide financial stability risks from individual institutions’ data. Section 3 discusses the efficiency versus macrostability trade-off in the context of the revised Basel accord. Section 4 discusses the implications of liquidity crises for macroprudential analysis and policy, thereby touching on both the aggregation and stability/efficiency issues. Section 5 briefly concludes by outlining some areas of future research at the Bank of England.
글
댓글 없음:
댓글 쓰기