Macroprudential regulation--the missing policy pillar.

Davis, E. Philip ; Karim, Dilruba

The recent Sub-Prime crisis has prompted a close focus on the causes of financial instability as well as the issue of whether it can be prevented. There is a growing realisation that the Sub-Prime crisis, although having some important unique features, also had a number of generic aspects in common with earlier financial crises, of which a large number have been seen in recent decades. Accordingly, the crisis has prompted a debate about macroprudential policy, which focuses on the financial system as a whole, treating aggregate risk as endogenous with regard to collective behaviour of institutions. Our survey shows that a great deal of progress has been made in 'macroprudential surveillance' and related research on causes and predictors of crises. Much less progress has been made in 'macroprudential regulation', the design and implementation of policies to prevent or mitigate threatened crises.

Keywords: Financial crises; macroprudential policy; financial regulation

JEL Classifications: G28; E44

I. Introduction

The current financial crisis has led to an increased focus on financial instability, but also a realisation that similar events have been extremely common in recent decades. (1) We define financial instability, or systemic risk, as entailing heightened risk of a financial crisis--"a major collapse of the financial system, involving an inability to provide payments services or to allocate credit". Two main sources of such systemic risk can be discerned. First is the tendency for financial and nonfinancial firms and households to overexpose themselves to risk in the upturn of the credit cycle, and then become risk averse in the downturn. This may relate to credit, liquidity or market risk. Second is the tendency of financial firms to ignore spillover effects of their behaviour on the rest of the financial system (Bank of England, 2009).

Macroprudential policy aims to counteract threats to financial instability. Consistent with the above, such macroprudential policy can be defined as that which focuses on the financial system as a whole, and also treats aggregate risk as endogenous with regard to collective behaviour of institutions. It aims to limit system wide distress so as to offer a stable provision of financial services to the economy and thus avoid output costs associated with financial instability (Borio, 2009). Although ultimately leading to efficient allocations of productive capital, in common with microprudential policy, macroprudential policy is likely to impose short-term costs on the financial sector and hence reduce availability of credit to some households and companies. These costs need to be weighed against benefits of greater financial stability. (2)

This article seeks to assess the need for a macroprudential policy pillar and the degree to which progress has been made in instituting one, viewed in the light of lessons learnt in the field of financial stability over the past decade. During this period there have been numerous crises, coinciding with considerable research and developments in policy and culminating in the lessons of the current crisis. A major impetus to these have been a growing realisation of the costs of crises, which can be over 20 per cent of GDP (Hoggarth and Sapporta, 2001). A further stimulus is the realisation that such crises have related to shared exposure to macroeconomic risks and not solely the failure of a single large institution. Many central banks have obtained a mandate to pursue financial stability (Das et al., 2004). A key policy development has been Macroprudential Surveillance, although as highlighted, there remain some unanswered questions as to its use in practice in full-blown Macroprudential Regulation.

2. Why we need macroprudential regulation--developments in the understanding of the process of financial instability

A generic approach

A primary issue, and background for the rest of our article, is a growing realisation of the generic nature of financial crises. In other words, they are not disparate random events but share key common features. As highlighted in the table below, there are both exogenous and endogenous aspects.

The process often starts with a primary shock to the economy and financial system that is favourable to growth and investment. But this leads to a process of propagation, whereby there is a build-up of vulnerability in the economy and financial system, associated with overextension of balance sheets and build-up of financial imbalances. Price based measures of asset values rise and price based measures of risk fall, while risk appetites increase. Balance sheets grow (including off balance sheet exposures), short-term funding increases and leverage rises. Hence there is increased aggregate exposure to credit risk (leverage), liquidity risk (short-term funding) and market risk (size and volatility of balance sheets). (3) Firms may also become more interconnected, increasing network risk, unless regulation penalises such links.

The aggregate risks exacerbate the boom, leading to a crisis when a secondary (adverse) shock hits a vulnerable financial system. This causes asset prices and market liquidity to fall, and defaults to increase, leading to a sharp rise in risk aversion and a credit contraction. In turn, there is further propagation in a crisis period (systemic risk), as for example failure of interconnected firms may lead to major spillovers. This typically prompts policy reactions if the crisis is sufficiently severe, and considerable adverse economic consequences (the 'costs of instability').

Types of crisis

Whereas traditional banking crises remain a major manifestation of financial instability, especially in emerging market economies, awareness has grown in the past two decades that there are alternative forms of crisis that may equally lead to systemic consequences. These are related to the ongoing securitisation of financial systems. One is extreme market price volatility after a shift in expectations (see Davis, 2002). Whereas violent price movements may in themselves not have systemic implications, (4) these may emerge when such movements threaten institutions that have taken leveraged positions on the current levels of asset prices. Currency crises, a subset of these, may sharply affect banking systems, giving rise to "twin crises" (Kaminsky and Reinhart, 1999) as in Asia in 1997 and Argentina in 2001.

There may instead be protracted (5) collapses of debt or derivatives market liquidity and issuance. These are analogous to bank runs in that asymmetric information and uncertainty generate liquidity collapses via panic sales of the asset concerned (see Davis, 1994). The risks are acute not only for those holding positions in the market but also for those relying on the market for debt finance or liquidity--which increasingly include banks. The Russia/LTCM crisis of 1998 and, particularly, the US Sub-Prime crisis which began in 2007, showed that these market-liquidity crises are recurrent features of modern financial systems (Davis, 2009a) given the structure of the modern financial system they can shift rapidly across borders. The Sub-Prime crisis has shown that interbank market liquidity can be highly vulnerable as well as that of securitised debt markets. Periodic collapse has been a feature of international interbank markets (Bernard and Bisignano, 2000) but had hitherto been less common in domestic interbank markets.

Lessons from traditional theory

The traditional theories of financial instability are set out in Davis (1995, 1999, 2002). Rather than repeating them here, we highlight some aspects shown to be of particular relevance to the macroprudential approach. The 'financial fragility' view (6) (Kindleberger, 1978; Minsky, 1977) has been repeatedly vindicated by recurrence of credit and asset price booms, followed by crises, as in Asia 1997 and in the US Sub-Prime episode. The overall structure of the 'generic crisis' set out in the table links closely to this approach.

Secondly, the relevance of the 'uncertainty' approach (7) has been shown by repeated problems with institutions' exposures to financial innovations, whose properties are not yet tested over a full cycle and are thus subject to uncertainty. The recent bank exposures to Asset Backed Securities (ABS) entailing credit risk transfer is a case in point. Equally, the 'disaster myopia' theory (8) is illustrated by the apparent short memories of instability once a period of calm has been observed. Disaster myopia highlights incentives leading to underpricing of risk and these may in turn reflect expectations of safety net provision and regulatory arbitrage, both of which have been seen in the Sub-Prime crisis.

Recent theoretical research findings

Selected recent research findings are highlighted here, again for the light they cast on the macroprudential channels of transmission of financial instability.

Allen (2005) (9) models links between asset price bubbles and financial fragility. Central to bubble creation are principal-agent conflicts since bank managers' upside-risk payoffs increase in risk whereas limited financial liability ensures restricted downside-risk losses. (10) With intermediation, investors place borrowed funds (11) into risky assets such as commercial property and transfer default risk onto lenders. This motivates borrowers to bid-up asset prices above their fundamental (12) values creating a bubble. Moreover, bubbles are propagated by investors' expectations of higher future credit availability and credit volatility, since this allows higher asset returns via risk transfer. Hence macroprudential focus should be on incentives and bubble detection, inter alia.

Theoretical and empirical models of contagion--noted above as central to the macroprudential approach--are extensively reviewed by De Bandt and Hartmann (2000). Freixas and Parigi (1998) and Freixas et al. (2000) focus on direct bank linkages and suggest borrowing arrangements between banks cause a domino effect if one bank is unable to meet its obligations, sometimes due to depositors withdrawing funds from a single bank in fear of deposit withdrawals from others. (13) Resulting externalities which explain 'rational herding' are modelled by Chen (1999).

Empirical work has shown contagion to be a valid concept. Autocorrelation between bank failures indicates concentrations of failures i.e. contagion (De Bandt and Hartmann, 2002). Abnormal bank stock price behaviour alongside 'bad news' of banks' performance (tested by event studies) also shows contagion, as does depositors' behaviour in response to bad news. Calomiris and Mason (2000) identified abnormally high withdrawals during the Great Depression. Jayanti and Whyte (1996) find significant increases in UK and Canadian banks' Certificate of Deposit rates after the Continental Illinois failure (1984), indicating international contagion.

As regards fundamental-driven contagion, most banking crisis prediction models (such as Demirguc Kunt and Detragiache, 2005) employ purely domestic variables, albeit often capturing cross-border impacts (such as terms of trade and the exchange rate). Santor (2003) finds crises are more likely following the occurrence of crises in countries in the same income group--a result he attributes to information rather than fundamental driven contagion. On the other hand Barrell et al. (2010) found that weighted (14) occurrence of a crisis in an OECD country affects crisis probabilities elsewhere in that group. This may be attributable to a variety of contagion channels.

Aspachs et al. (2007) suggest two components should define financial fragility, usable in macroprudential analysis, namely, reduced bank profitability and increased default probability. The advantage is that such indicators can be applied at a firm or aggregate level (Goodhart et al., 2006). The combination is used because neither component alone implies fragility; lower profitability could arise through recession and excessive risk taking could raise defaults without instability. Heterogeneous agents with a distribution of risk appetites are used to link fragility with welfare changes; Aspachs et al. (2007) show an exogenous banking system shock increases aggregate defaults, decreases profits and reduces agents' welfare. These theoretical predictions are calibrated against UK data.

Shin (2008) suggests that by holding assets which are claims against other borrowers and by holding claims against each other, financial institutions generate complex webs of risk exposures; relative asset and liability values, credit availability and asset prices become interdependent. Resulting externalities mean shocks to financial systems are amplified, causing spillovers onto many balance sheets. (15) The author models a system of interlocking balance sheets to solve for asset prices which depend on the creditworthiness of other institutions in the system. He shows this fixed point problem has a well defined solution; each claim can be uniquely priced in terms of parameters describing the underlying financial system (current prices of underlying assets, debt levels and structure, and the profile of balance sheet inter-linkages), which are hence areas requiring focus.

Recent research has also focussed on regulators' incentives. Kocherlakota and Shim (2006) suggest the regulator's response to instability is conditioned on the ex-ante probability of asset price collapse; if this is high, remedial action is optimal, otherwise the regulator shows forbearance towards instability. Principal-agent and political motives explain why regulators show forbearance (instead of 'prompt corrective action') towards instability. (16) Degennaro and Thompson (1996) suggest regulators who act as utility-maximising agents view forbearance as an attractive gambling strategy where taxpayers bear the costs for losing the gamble as has occurred already to some extent in the sub-prime context.

The sub-prime crisis has led to further advances in the understanding of financial instability relevant for macroprudential purposes. For example, Adrian and Shin (2008) suggest that contagion during the current crisis differed, in quite specific ways, from that in traditional liquidity crisis models. The traditional view is that credit risk leads to contagion, either via direct exposures or uncertainty over opaque balance sheets. In the current world, Adrian and Shin argue that contagion occurs via changes in market prices, according to the way that risks are measured and the mark-to-market practices of financial institutions. Financial institutions are seen to manage balance sheets actively in response to price changes and measured risk. Moreover, this appears to have led to a positive relation between changes in leverage of commercial banks and balance sheet size, as they have taken on behaviour patterns hitherto more typical of investment banks.

A helpful complementary paradigm of funding liquidity that encompasses some of the events of the 2007 and 2008 crisis is provided by Freixas et al. (2000). According to this model, liquidity may dry up for a solvent bank in the interbank market if there is imperfect information, or if there is market tension which reduces the lending banks' excess liquidity and reduces its scope to diversify. The interbank market as a whole may face liquidity problems if each bank refuses to lend to others because it cannot be confident of its own ability to borrow, a form of liquidity crisis akin to the Diamond-Dybvig (1983) model.

Brunnermeier (2009) talks of four mechanisms by which small shocks are amplified, leading to a loss of liquidity. These are first, borrowers' balance sheet effects comprising a loss spiral (as an initial loss on a leveraged balance sheet leads to a decline in net worth, sales and price movements, further reducing net worth) and a margin spiral (as increased margins lead to deleveraging and sales, leading to lower prices, further increasing margins). Second is a lending channel effect (notably precautionary hoarding of liquidity). Third are runs on institutions and markets (including the interbank, ABCP and investment bank repo markets). Fourth are network effects, for example, when Goldman Sachs expressed concerns about exposures to Bear Stearns via swap netting arrangement, hedge funds avoided Bear Stearns as a prime broker thereby helping to bring about its demise.

Finally, off-balance sheet risks are a new feature of the current crisis, a manifestation of the increasing deregulation and innovation within the OECD financial markets as noted recently by Acharya and Richardson (2009). They argue that the post-2000 explosion of asset backed security issuance was driven by banks' desire to avoid holding costly capital against their assets. (17) This regulatory arbitrage is what the authors cite as the main cause of the Sub-Prime episode.

Farhi and Tirole (2009) suggest that the maturity mismatch within special investment vehicles (SIVs) and conduits (between long-term mortgage backed assets and the short term commercial paper used to finance them) was a structural feature of the business models of most banks who displayed strategic complementarities with their peers. When authorities use monetary (or fiscal) policy to bail out failing banks, society incurs a fixed cost which is only justified if sufficient banks need bailing out. Therefore each individual bank correlates its risk exposure with other banks such that OBS risks can become systemically high. This highlights the need for macroprudential surveillance to recognise risks arising from innovations in banking.

3. Progress with the macroprudential pillar--the development of macroprudential oversight

Definition

Owing to costs of crisis, it has been realised that there is an immense premium on timely warnings regarding systemic risks as an input to policy decisions as well as to strategies and market behaviour of financial institutions. There is also a realisation following the theory and experience outlined above that forms of aggregate risk which pervade in periods of vulnerability are easily missed by regulators focussed on individual institutions, since risks depend on institutions' collective behaviour (Bank of England, 2009). (18) Equally, network risk depends on the scale of interconnection between firms that requires an overall view of such links. Both can give rise to credit and liquidity risks. Accordingly, in the past decade 'macroprudential surveillance'--defined as monitoring of conjunctural and structural trends in financial markets so as to give warning of the approach of financial instability--has become a core activity for many central banks. We summarise progress here before highlighting below that such surveillance is necessary but not sufficient to provide an alternative policy pillar.

Methods of surveillance

Typically, central banks institute regular Financial Stability Reviews to assess the outlook for financial stability. Already by end-2005, 50 central banks had done so (Cihak, 2006), often prompted by IMF/World Bank Financial Sector Assessment Programmes (FSAPs).

Data needs (Davis, 1999) include macroeconomic and financial data for assessing conjunctural conditions, non-financial sector debt, leverage and asset prices for considering vulnerability of borrowers, and in the light of these, bank balance sheets and income and expenditure for considering robustness of banks. Risk measures derived from financial prices complement leverage and income indicators. Stress tests and forecasts of indicators and derived stability indicators such as defaults and bankruptcies, including risks to the central projection, are needed to tell a full story.

As discussed in Davis (1999), key lesson learnt in surveillance practice for these data include the importance of economy in the number of variables to tell a coherent story, and derivation of data needs from theory and experience. Furthermore, there is a need to use qualitative aspects; surveillance cannot be purely numerical and unlike inflation is not easily summarised in a single index. Changes in regulation and competition, and innovation, are among key qualitative inputs. Equally, there is a need to develop benchmarks and norms for the economy and financial system to assess deviations, while remaining aware that these can change (e.g. during financial liberalisation). Cross-border as well as domestic influences need to be taken into account, not least given the internationalisation of banking. Also new players such as hedge funds need to be incorporated when they become relevant.

There is a need for observation of overall patterns in the light of past occurrences of financial instability, both at home and abroad, developments in theory and the generic view set out in table 1. Viewed in the light of that table, data can show either shocks (e.g. triggering boom or crisis) or propagation mechanisms (showing whether a boom is underway with heightened vulnerability). But since shocks are random, the vulnerabilities are the main focus. Then there is a need for a judgmental approach in drawing conclusions, using again a conceptual framework derived from theory such as that of table 1 (how vulnerable is the system--what shocks could take place?). Fell and Schinasi (2005) suggest the use of an implicit corridor of financial stability, akin to an exchange rate target, with judgements made as to whether the system is inside the corridor, approaching the edge, just outside or systematically outside, which will imply different policy recommendations.

As an example of a procedure, the ECB undertake a 7-point vulnerabilities exercise, first identifying vulnerabilities and imbalances, then translating them into potential risk scenarios, identifying triggers (shocks) for the scenarios, assessing the likelihood of scenarios arising, estimating the costs for the financial system, assessing robustness to such shocks and then ranking the risk. They note the need to include endogenous sources of risk (within institutions, markets and infrastructure) as well as exogenous risks from the macroeconomy. Fell and Schinasi (2005) give a check list of criteria for sound analysis including: Is the process systematic? Are the risks identified plausible? Are the risks identified systemically relevant? Can linkages and transmission (or contagion) channels be identified? Have risks and linkages been cross-checked? Has the identification of risks been time consistent?

Tools for macroprudential analysis

'Distance to Default' (DtD) (19) measures credit risk by expressing a firm's net worth as a proportion of asset price volatility; (20) the higher this ratio, the lower the likelihood of default. Looking at all banks, one can take a view of collective risks to the banking sector. (21) Any asset with a liquid secondary market (22) can be used because assuming market efficiency, prices should incorporate markets' forward looking expectations of firm default (Chan-Lau, 2006). However, when applied to banking distress, DtD ignores the likelihood of regulators intervening well before default. Hence Chan-Lau and Sy (2007) suggest banking DtD measures should reflect regulatory capital requirements, in line with the Basel Committee.

For a given confidence interval and time span, Value at Risk (VAR), a market risk measure, indicates the maximum expected portfolio loss under normal market conditions (Benninga and Weiner, 1998). Again these may be aggregated for macroprudential purposes. Basel sets a 99 per cent confidence interval and a 10-day horizon, based on at least twelve months' historic data. (23) Banks must hold at least three times this VAR amount in capital. Criticisms arise because means, variances and correlations of asset returns are based on assumed probability distributions. Also, quantifying actual portfolio positions requires detailed knowledge of all asset risks by banks.

Stress tests quantify portfolio movements for unlikely but feasible events (see BIS, 2000 and 2001 for more detail). Scenario tests simultaneously vary several risks in one direction, emulating historic events or hypothetical scenarios. Sensitivity tests shock individual risks symmetrically. (24) Limitations arise because probabilities of shocks materialising are not computed. Also, risk parameters are at times subjectively chosen by managers and impose high computational costs on institutions.

Bubble (25) detection searches for bubble premia, excess volatility and cointegration between dividends and prices (Brooks and Katsaris, 2003). In older models, bubble components exploded over time, whereas recent, more realistic models allow prices to return to fundamentals via crashes (Raymond Feingold, 2001). This makes detection by cointegration harder due to bias and kurtosis (Evans, 1991), hence correction for these will improve bubble detection (Raymond Feingold, 2001).

Early Warning Systems (EWS) generate out-of-sample probabilities of crisis using historic data. Demirguc-Kunt and Detragiache (1997) developed a parametric EWS for banking crises using a multinomial logit model with macroeconomic, financial and structural variables as inputs. Logistic models are appropriate for explaining binary banking crisis observations in panel data. Davis and Karim (2008a) improved prediction by introducing more countries, crises (26) and dynamics in the macro variables; over 90 per cent of in-sample crises were correctly identified. More recently, Barrell, Davis, Karim and Liadze (2009) have shown that heterogeneous country samples are inadequate for EWS design since, for the OECD at least, crisis determinants are entirely different from other countries; in the OECD, banking liquidity, leverage and real house price growth supersede the traditional macroeconomic factors significant in global models (27) as the main predictors of banking crises.

Kaminsky and Reinhart (1999) developed a nonparametric signal extraction EWS which tracks individual time series for abnormal behaviour that has previously been associated with crises. If a variable subsequently behaves abnormally, a crisis probability can be computed. Davis and Karim (2008a) improve signal extraction for banking crisis prediction by creating composites of indicators weighted by their signalling quality.

Difficulties in identifying systemic crises compromise the EWS dependent variable. Also, predictive accuracy varies according to the cut-off threshold subjectively chosen by the policymaker. Nevertheless, logistic and signal extraction techniques provide a computationally easy way to predict banking crises using global and country-specific data respectively and are thus useful tools to complement macroprudential surveillance.

A recent development is the binary recursive tree technique, which can be used to answer the question "which non-linear variable interactions make an economy more vulnerable to crisis than others?" It can be argued that liquidity, credit and market risks are all potentially non-linear (e.g. once a threshold level of credit risk is surpassed, a decline in GDP may have a heightened impact on the probability of a crisis). The estimator identifies the single most important discriminator between crisis and non-crisis episodes across the entire sample, thereby creating two nodes. These nodes are further split into sub-nodes based on the behaviour of splitter variables' non-linear interactions with previous splitter variables. This generates nodal crisis probabilities and the associated splitter threshold values. This is an innovative approach used mainly in medical research to date. The technique has been applied to systemic banking crises by Duttagupta and Cashin (2008) and Davis and Karim (2008b). The key indicators found to be most useful in these studies include real interest rates, GDP growth, inflation and credit variables.

The limits of macroprudential oversight--an evaluation for the subprime crisis

Davis and Karim (2008b) show that the US sub-prime crisis was only partly foreseen by the policy community. Looking at reports from the BIS, IMF, ECB and Bank of England they found that, although all had important insights in their headline sections, none of the reports highlighted the conduits and SIVs that were a key feature of the crisis. Equally, none foresaw the collapse of the interbank market or the overall magnitude of the effects from the sub-prime crisis alone. None considered possible links from financial instability to the real economy on the scale seen in 2008-9.

Davis and Karim argued that the BIS had the most forward looking analysis of events and possible policy responses, reflecting its longer-term concern over the build-up of debt, risks in structured products and rising asset prices. Even they failed to see some of the consequences of the crisis, notably the seizing up of interbank markets.

They also saw limits in the use of macroprudential tools. Among global early warning systems calibrated for the US and UK, the logit performed best but was still only marginally able to help predict the crisis (although the BRT model had a higher average crisis prediction score). These results to some extent show that the subprime crisis had specific features that were not typical of the average banking crisis in both advanced and emerging economies. However, their contention was that, rather than rejecting such models, the results show they should be better adapted for the specific features of advanced countries, that may also include aspects of securities market instability.

As noted, progress on this in Barrell, Davis, Karim and Liadze (2009) estimating logit models shows indeed that there are different determinants of OECD crises from EMEs, namely bank capital, bank liquidity and house price growth. This model was able to predict the crisis for a number of countries, albeit not the US itself. Equally, they maintain that a generic features checklist would also be useful complement for such analysis. Nevertheless, it is important to acknowledge the ongoing limitations of knowledge and ability to predict crises using macroprudential surveillance tools. Hence qualitative analysis and judgement remain essential to such surveillance.

4. Policy issues--making macroprudential regulation operational

The broad issue

The initial two policy objectives of macroprudential regulation are early identification of potential vulnerabilities and, through their public reporting, to encourage financial institutions to do stress testing. And these can be achieved by the surveillance highlighted above. The more difficult policy decision is what to do if there are macroprudential warnings, given the third objective is to promote preventative and remedial policies to prevent financial instability. Surveillance should also help to resolve instabilities (crisis resolution), when preventative and remedial measures fail, but this is not the desirable scenario.

Moral suasion and intensified supervision are obvious preventative measures but may not be sufficient. Or equally, the question arises whether monetary policy can deal with asset price bubbles--many central bankers seem to oppose the idea that monetary policy should aim to deflate nascent bubbles, arguing that the interest rate is best devoted to control of general price inflation. Bank of England (2009), for example, argues that both the real economy and inflation expectations could have been destabilised by a policy tightening in response to the credit/asset price bubble of 2003-7, while the required tightening is itself hard to calibrate given shifting risk premia. Rather, both theory and experience suggest there is a need for variation or adjustment in prudential parameters, which we explore in this section.

Bear in mind that the whole rationale of macroprudential regulation is that relying on individual bank supervision at a micro level is necessary but not sufficient for financial stability. Clearly higher capital adequacy makes banks individually less vulnerable to failure, but equally, the past decade has also shown the adverse incentive effects of capital adequacy. These include maximising risk in the buckets, setting up avenues of disintermediation via subsidiaries, and also banks skimping liquidity protection. More recently, opportunities for regulatory arbitrage have created incentives for managers to avoid holding risk-weighted capital off-balance sheet by selling 364-day commercial paper. The procyclicality of the financial system, which is already apparent with Basel 1, seems set to worsen with Basel 2 (Goodhart, 2005). For example Basel 2 prompts banks to seek to sell assets in a recession due to rating downgrades and higher capital charges. This not only can become cumulative but also worsens borrower solvency further and may spread credit rationing to other markets (Warwick Commission, 2009). Even lacking such regulation, in trying to make themselves safer, for example selling an asset when the price of risk rises, banks may collectively act in a way that generates systemic risk.

It is widely argued that the most desirable means of preventing financial crises is to implement standards which automatically act to prevent 'financial fragility' from arising. One part of this will be tightened microprudential standards on which the Basel Committee is already working, (28) but another is macroprudential surveillance, as discussed below. There remains the further issue of how to link macroprudential and microprudential regulation more effectively (Barrell and Davis, 2008). One aspect is the appropriate design of a countercyclical regulatory framework (the time series dimension). It also includes increasing risk weights for risks that are common across institutions rather than idiosyncratic, or which can spillover from core institutions (the cross section dimension). It is vital that regulation of both types remains effective in the sense of ensuring that risky activities do not migrate either internationally or to less-regulated financial institutions such as hedge funds (Goodhart, 2008). The structure of regulation more broadly may need to be reconsidered.

Countercyclical regulation

As noted, there is an ongoing debate about whether Basel 2 has been made the financial system more or less procyclical. Through-the-cycle ratings should tend to dampen the procyclical forces previously at work (Bank of England, 2009). The requirement for stress testing for a downturn under Pillar 2 should act also against procyclicality by making banks consider the range of risks that can occur during a cycle or even a long period of time. Tougher liquidity standards being introduced since the crisis will help to reduce procyclicality, especially if they reduce the ability for banks to expand balance sheets rapidly using wholesale funding rather than solely raising the volume of liquid assets. However, the increased use of marking-to-market and banks' own assessment of risk may lead banks hit by falling credit quality to raise capital or contract balance sheets in downturns (since raising capital is difficult in such conditions) and thereby exacerbate the underlying weakness (Goodhart, 2005).

There is limited international experience with operation of regulatory standards that act against credit and asset-price cycles, for example by increasing minimum capital ratios to dampen lending growth, ensuring banks have buffers to draw on in the downturn. Dynamic provisioning, as in Spain, is a rule requiring banks to build up general loss reserves in good times consistent with the long-run average default performance of the relevant loans, to cope with losses in bad times (Fernandez de Lis and Herrero, 2009). In other countries, including the UK, by contrast, banks did not build up extra provisions in the upturn; the UK nonperforming loans/total loan ratio fell from 2.5 per cent in 2003 to 1 per cent in 2006, while the ratio of provisions to non-performing loans fell from 70 per cent to 54.6 per cent. Potential conflicts with tax rules and with accounting standards would need to be addressed if the Spanish approach is to be implemented elsewhere. Equally, it should be noted that Spain did not avoid a credit boom, rather, its banks were better able to withstand the eventual downturn. Furthermore, the dynamic provisioning regime only applies to the 'banking book' and not the 'trading book' exposures which have also been a cause for concern.

A number of other national authorities are considering explicit countercyclical regulation (Bank of England, 2008, 2009). These could include an overall leverage ratio of capital to unadjusted (rather than risk-weighted) assets. This limits the scope under Basel 2 arrangements for banks to assess their own risk by providing a one-size-fits-all ceiling. It may also be helpful in making regulation more transparent, although it is essential that the ceiling applies to all relevant assets and does not encourage banks to use off-balance sheet structures to escape the ceiling.

Time varying capital requirements related to lending growth are also under consideration, alongside the purchase of catastrophe capital insurance (Kashyap et al., 2008). Goodhart (2005) suggests relating the capital requirement on bank lending to the rate of change of asset prices, while another alternative is to limit individual bank asset growth to a rate consistent with an inflation target (Warwick Commission, 2009). Bank of England (2009) recommends a capital surcharge for all banks linked to estimates of aggregate credit risk derived from macroprudential surveillance and related stress tests, or on sectoral lending exposures in a similar manner. They note that similar changes could be applied to underpriced liquidity risk (e.g. of rollover or market liquidity).

Bank balance sheet based countercyclical regulation could be complemented by regulation of remuneration with the same effect. For example, supervisors could encourage risk adjustment remuneration to managers and traders that is based on long-term or realised return and not short-term book profit (FSA, 2009).

Time-series macroprudential regulation could be based either on rules or discretion. In general, such countercyclical policies based on discretion may risk being ineffective as the authorities can easily share the same excessive optimism as the private sector about future prospects and risks (disaster myopia). In addition, authorities may face political pressures if they try to contain the expansionary phase of a credit cycle, particularly from the financial industry. Regulatory capture is a risk in this context also. Decision-making would be less predictable, leading to banks holding higher precautionary buffers, or at least to react less to policy anticipations.

A rules-based approach to procyclicality, although blunter, would have the benefit of being transparent. One possible approach is to use credit growth and asset prices as inputs to such a rule (Borio and Drehmann, 2009); the Spanish approach to dynamic provisioning itself is a rules-based approach based on balance sheet growth in excess of long-term averages. It would be important that such an approach does not limit the important role of banks' risk management in ensuring the stability of the financial system. Furthermore, the standard should ideally not only reduce risk appetite at the peak of the boom but also ameliorate the credit crunch in the downturn.

Bank of England (2009) argues discretion is necessary to allow adaptability in the case of structural change or uncertainty, as well as permitting judgement. They contend that constraints could be built in to discretion via, for example, setting out details of the process and analysis underlying decision-making, as well as parliamentary scrutiny and a fixed timetable.

Borio (2009) notes also that capital adequacy or even banking regulation generally are only one aspect of the procyclicality problem. Fair value accounting, for example, increased procyclicality in the sub-prime crisis. Monetary and fiscal policy as well as the nature of the safety net may also give rise to incentives. For example, changes in interest relief on real estate loans may provoke instability as in Sweden prior to the 1991 crisis. Equally, generous deposit insurance was a key background for the US Savings and Loans crisis. Hence there is a need for a holistic approach to the problem.

Cross-sectional regulation

As mentioned, a second aspect of macroprudential regulation is to allow for cross-sectional risks. A key aspect of this is that not all institutions give rise to similar systemic risks. Some may be of particular importance for being large, highly connected with other institutions and/or with a high propensity for fire sales of assets that would affect other firms' balance sheets. Such core institutions are most likely to be judged too big to fail. Developing indicators, stress tests and models to measure such systemic importance is a key ongoing task, a good example of which is the Austrian model of interbank links (Elsinger et al., 2006). However, data for constructing such models are absent in many countries.

Regulation may then need to impose closer regulation and possibly higher capital requirements on systemically important institutions (i.e. those giving rise to high system-wide losses should they fail). Such capital requirements would also give an incentive to reduce balance sheet size and interconnections, as well as to develop so-called living wills which facilitate orderly wind down. It could be complemented by a shift from 'process driven regulation' which rewards large banks that are best able to construct credit risk models and systems, to 'results driven regulation' that would be based on better risk management and would be more size neutral. Penalising large firms (and other policies such as transactions taxes) would reduce the risk of regulatory capture and some contend would limit the damage the sector could inflict on the wider economy (Warwick Commission, 2009). We note that this is an empirical question which is not resolved at present. While fiscal losses may indeed link to the size of the financial sector, wider economic damage can arise from cross-border lending unrelated to the size of the domestic sector, as in the 1997 Asian crisis and many other Emerging Market crises.

The fact that many interconnected institutions are cross-border in nature gives an additional difficulty to their regulation, especially as the Lehman's failure showed that such firms are 'global in life but national in death'. As suggested by IMF (2009), this requires agreement by home and host supervisors that such institutions act as subsidiaries in local markets where they are systemic in size.

Quite apart from tightly regulating large and interconnected institutions, the cross-section approach may entail higher capital charges on behaviour that is typically common across banks relative to that which is idiosyncratic, even if the risks from the individual institutions are the same. A focus is needed, in other words, on areas where banks are susceptible to herding behaviour. So for example there could be tougher controls on lending to finance real estate, a major component of collective losses as asset prices fall. (29) These could include limits to loan-to-value (LTV) ratios for property loans or restrictions on income gearing.

Also affecting cross-section risk is the degree of bank reliance on wholesale funds, which have been seen to dry up simultaneously for all institutions. This could be discouraged by raising capital requirements for banks that have high wholesale funding ratios, or against the mismatch in maturity of assets and liabilities, adjusted for liquidity (e.g. a discount on assets acceptable as collateral by the central bank), (Brunnermeier et al., 2009). Further benefits in reducing cross-section risk could be gained by having large liability limits related to the size of the exposure to the debtor (as opposed to the creditor in large exposure), structural policies for central counterparties in liquid markets and separate capitalisation of different geographical parts of a cross-border firm. More detailed policies which enforce capital requirements linked to degree of maturity mismatch could be envisaged (Warwick Commission, 2009). These would encourage institutional investors rather than banks to hold liquidity risks--and the opposite for credit risks--that would make the system more resilient because it allocates risks to those better able to hedge them.

As argued in Barrell and Davis (2008), macroprudential problems are actually threefold--bad lending leading to losses giving rise to liquidity problems, excessive reliance on wholesale markets and complex instruments leading to confusion. Given the uncertainty problem with lack of experience of behaviour in downturns, there is a need for further regulation of complex instruments or an increase in capital held against them. For if they fail when tested, as have the majority of asset backed security innovations of the past decade, they can cause huge damage. Possibly this could be in the form of higher capital charges until a recession has taken place. Or more radically, there could be a need to justify innovations before they can be introduced, as is the case for drug testing.

Changes in the structure of regulation

The boundary problem in regulation arises for all of the above issues. The basic point, as set out in Goodhart (2008) is to ensure that regulation which is effective does not merely lead to substitution flows towards the unregulated sector. He suggests a resolution could be to ensure that regulation only is effective occasionally, i.e. during the height of a credit/asset price driven boom. There is also a need for vigilance by regulators to ensure banks do not set up subsidiaries (e.g. SIVs) outside the regulatory net for which they are sufficiently reputationally connected to not allow them to fail. Bank of England (2009) suggests that the boundary problem could be countered by tight controls on bank exposures to non-banks or coverage of 'all' institutions. On the other hand they note that international leakages would arise for a macroprudential regime, for example due to lending by foreign branches or direct cross-border lending into the country that is not subject to the macroprudential capital charges. Coordination among international regulators may accordingly be needed, as well as a move from branch to subsidiary-based operation by foreign banks (see also FSA, 2009, and Warwick Commission, 2009).

Barrell and Davis (2008) argue that one key root cause of the crisis has been the decision in the 1980s and 1990s to move away from structural regulation (e.g. limits on competition, prices and scope of activities for financial institutions) to an almost sole concern with the efficiency of the financial system. Both should be of concern to regulators, and the costs and benefits of both should have been balanced. Efficient financial markets increase welfare, reduce risk premia and raise sustainable output. However, they also have a tendency to produce products that become widely adopted before they are stress tested in a recession, as discussed above, and also unsustainable levels of competition. Further prudential tightening and even some direct controls on bank activities may be needed.

The Obama administration in the United States announced on 21 January 2010 a proposal to limit certain activities by commercial banks deemed as risky and inappropriately covered by deposit insurance and access to the discount window. These are proprietary trading of securities, ownership of hedge funds and of internal private equity funds. A further aim is to reduce the size of institutions directly, possibly by limiting market share of non-deposit funding. Some commentators have seen parallels with the Glass-Steagall regulation in the 1930s which separated commercial from investment banking and was only abolished in 1999. In the UK City Minister Lord Myners said the US proposals were "very much in accordance with the direction we have been setting", while the UK Conservative Party has declared an interest in introducing similar legislation, should it win this year's election. Whereas the Obama proposals may not pass Congress in their current form, it remains appropriate to evaluate them in the context of this article as a radical form of structural regulation that would seek to reduce cross-sectional risks from large and risky institutions as set out above. Accordingly, a key aim of the legislation is that commercial banks should be forced to downsize so they do not become "too big to fail" and subject to moral hazard at taxpayers' expense, as well as reducing the risk on their balance sheets, which should reduce the probability of failure. There could well be benefits from such an approach.

On the other hand, it is not clear that the specific elements targeted by the legislation were actually the principal causes of the Sub-Prime crisis. These were more directly linked to excessive investment by banks in overpriced structured products (Collateralised Debt Obligations) in off balance sheet subsidiaries, together with inadequate capital and liquidity to cover risks being incurred, and a lack of focus on macroprudential risks by market participants and regulators. Hence, it is essential that such rules should be supplemented by tighter microprudential and macroprudential regulation as set out above. Equally, potential costs such as lesser liquidity of securities markets if proprietary trading and activity of hedge funds are restricted, as well as weaker corporate governance if private equity is restrained, should be taken into account. Further issues are whether pure investment banks would be allowed to operate outside the regulation if they renounce banking status, and whether authorities outside the US and UK would fail to adopt similar measures. These would weaken the global benefits of the proposal and lead to major regulatory arbitrage.

As recommended by Weale (2008), it would be of major benefit if new prudential regulations could be self-enforcing. For example, if there are limits on LTV, banks have an incentive to exceed them to gain custom. However, if the national register of land titles (known as the Land Registry in the UK) refuses to register mortgages with LTVs over the limit, then there would be incentive to comply. Or, alternatively, as in Germany, a limit on LTVs for mortgages eligible to be securitised would be self-reinforcing (the limit in Germany is 80 per cent). Again, a tax on credit, if such were considered helpful for macroprudential purposes, would be self-enforcing if creditors have to show it has been collected before they can enforce debts.

Geographically, it can be argued that European-wide financial markets need a single regulatory structure that ensures the solvency as well as the liquidity of all financial institutions in the European Economic Area. Macroprudential regulation raises important home/host issues since, if there is most leverage for the home supervisor, the scope to take host country macroprudential issues into account will be limited. Arguably the macroprudential approach calls for host country supervision to have precedence, which could be enforced by the right of host authorities to insist on subsidiarisation of branches (Warwick Commission, 2009).

In this context, a key issue is whether new institutions are needed, as proposed in the EU and US, to detect macroprudential risks? Some doubt is justified, in our view, since they may lack both credibility and leverage over the supervisors and central banks who will take the relevant decisions. Financial stability reports from such institutions may thus be disregarded even more readily than those produced by existing institutions.

More appropriate may be close attention to the development of macroprudential regulation between the central bank and regulator (if they are separate) or within the central bank. It can be argued that regulators outside the central bank are often less focussed on macroprudential issues than the central bank themselves. One reason may be greater focus on depositor and investor protection, which leads to a concentration on individual institutions. Another is that the central bank is more likely to have to deal with the initial consequences of systemic risk via the lender of last resort facility. Hence a rebalancing of responsibilities may be appropriate to ensure sufficient macroprudential focus.

In the UK for example, under the Banking Reform Act, the Bank of England has a legal objective "to contribute to protecting and enhancing the stability of the financial systems of the United Kingdom" (see Davis, 2009b). The Act formalises the Bank's role in supervision of payments systems. A Financial Stability Committee is being set up within the Bank to deal with this objective, reporting to the Governor (HM Treasury, 2008). An important innovation under the Act is that the Bank is to be able to request data from banks through the FSA, whereas at present the FSA can only collect data it needs itself. Furthermore, these arrangements provide the basis for the Bank to have a much more detailed understanding of developments in the banking system.

Conclusion

The past decade has seen an unprecedented focus on financial instability, its nature and its early detection. Whereas this is clear progress compared to the blind manner in which many countries encountered financial instability in the late 1980s, there is still a long way to go. This was illustrated by the failure of Financial Stability Reviews (e.g. of the UK, ECB and the IMF) to detect the liquidity aspect of the Sub-Prime crisis, even though they did give warnings of the credit risk issues arising. (See Banque de France (2008) for an assessment of liquidity issues in the wake of the sub-prime crisis.) Also, the crisis has shown a rich array of adverse incentive effects on risk taking, often arising from regulation. It is suggested that the role of incentives is an emerging issue which warrants close focus (Chai and Johnston, 2000, gave an early approach to it). And even if vulnerability had been detected, experience has also shown that the means by which authorities can address macroprudential concerns are limited. Hence the need for a wider range of macroprudential tools and particularly those that will tend to limit ex ante the scope of systemic risk, in time-series and cross-sectional terms.

REFERENCES

Acharya, V. and Richardson, M. (2009), 'Causes of the financial crisis', Critical Review, 21, pp. 195-210.

Adrian, T. and Shin, H.S. (2008), 'Liquidity and financial contagion', in Banque de France Financial Stability Review: Special Issue on Liquidity, No. 11, pp. 1-7.

Allen, F. (2005), 'Modelling financial instability', National Institute Economic Review, 192, April.

Allen, F. and Gale, D. (2000), 'Bubbles and crises', Economic Journal, 110, pp. 236-55.

Aspachs, O., Goodhart, C.A.E., Tsomocos, D.P. and Zicchino, L. (2007), 'Towards a measure of financial fragility', Annals of Finance, 3, pp. 37-74.

Bank of England (2008), Financial Stability Report, October 2008, London, Bank of England.

--(2009), 'The role of macroprudential policy, a discussion paper', November, London, Bank of England.

Banque de France (2008), Financial Stability Review, Special Issue on Liquidity, Paris, Banque de France.

Barrell, R. and Davis, E.P. (2008), 'The evolution of the financial market crisis in 2008', National Institute Economic Review, 206, pp. 5-14.

Barrell, R., Davis, E.P., Fic, T., Holland, D., Kirby, S. and Liadze, I. (2009), 'Optimal regulation of bank capital and liquidity: how to calibrate new international standards', FSA Occasional Paper No 38.

Barrell, R., Davis, E.P., Karim, D. and Liadze, I. (2009), 'Bank regulation, property prices and early warning systems for banking crises in OECD countries', National Institute Discussion Paper No. 330, and Journal of Banking and Finance (forthcoming).

--(2010), 'Contagious effects of banking crises in OECD countries', NIESR, mimeo.

Benninga, S. and Weiner, Z. (1998), 'Value at risk (VaR)', Mathematica in Education and Research, 7, 4, pp. 39-45.

Bernard, H. and Bisignano, J. (2000), 'Information, liquidity and risk in the international interbank market: implicit guarantees and private credit market failure', BIS Working Paper No. 86.

BIS (2000), Stress Testing by Large Financial Institutions: Current Practice and Aggregation Issues, Committee on the Global Financial System, Switzerland, Bank for International Settlements.

--(2001), A Survey of Stress Tests and Current Practice at Major Financial Institutions, Report by a Task Force Established by the Committee on the Global Financial System of the Central Banks of the G10 Countries. BIS Publications, Switzerland, Bank for International Settlements.

Borio, C. (2009), 'The macroprudential approach to regulation and supervision, What? How Why?', presentation to the Banque de France and Toulouse School of Economics Conference on the Future of Financial Regulation, 28 January.

Borio, C. and Drehmann, M. (2009), 'Assessing the risk of banking crises--revisited', BIS Quarterly Review, March, pp. 29-46.

Brooks, C. and Katsaris, A. (2003), 'Rational speculative bubbles: an empirical investigation of the London Stock Exchange', Bulletin of Economic Research, 55, pp. 319-46.

Brunnermeier, M. (2009), 'Deciphering the 2007/8 Liquidity and Credit crunch', Journal of Economic Perspectives, 23, pp. 77-100.

Brunnermeier, M., Crockett, A., Goodhart, C., Persaud, A.D. and Shin, H. (2009), 'The fundamental principles of financial regulation', Geneva Reports on the World Economy II, International Center for Monetary and Banking Studies and Centre for Economic Policy Research

Calomiris, C.W. and Mason, J.R. (2000), 'Causes of U.S. bank distress during the depression', NBER Working Papers 7919, National Bureau of Economic Research, Inc.

Chai, J. and Johnston, R.B. (2000), 'An incentive approach to identifying financial system vulnerabilities', IMF Working paper No WP/00/211.

Chan-Lau, J.A. (2006), 'Market-based estimation of default probabilities and its application to financial surveillance', International Monetary Fund Working Paper 06/104, Washington, DC, IMF.

Chan-Lau, J.A. and Sy, A.N.R.(2007), 'Distance-to-default in banking: a bridge too far?', Journal of Banking Regulation, 9, I, pp. 14-24, November.

Chen, Y. (1999), 'Banking panics: the role of the first-come, first-served rule and information externalities', Journal of Political Economy, 107, pp. 946-68.

Cihak, M. (2006), 'How do central banks write about financial stability?', IMF Working Paper WP/06/163.

Crosbie, P.J. and Bohn, J.R. (2002), Modelling Credit Risk, California, KMV LLC.

Das, U., Quintyn, M., Chenard, K. (2004), 'Does regulatory governance matter for financial system stability?', IMF Working Paper WP/04/89.

Davis, E.P. (1991), 'Bank credit risk', Working Paper no. 8, London, Bank of England.

--(1994), 'Market liquidity risk', in Fair, D. and Raymond, R. (eds), The Competitiveness of Financial Institutions and Centres in Europe, Kluwer Academic Publishers.

--(1995), Debt, Financial Fragility and Systemic Risk, Oxford, Oxford University Press.

--(1999), 'Financial data needs for macroprudential surveillance: what are the key indicators of risk to domestic financial stability?' Lecture Series No 2, Centre for Central Banking Studies, Bank of England.

--(2002), 'A typology of financial instability', Financial Stability Report, 2, Vienna, Oesterreichische Nationalbank.

--(2009a), 'The lender of last resort and liquidity provision how much of a departure is the subprime crisis?', paper presented at the Conference entitled 'Regulatory Response to the Financial Crisis', London School of Economics, 19th January, and Brunel Economics and Finance Working Paper No. 09-09.

--(2009b), 'Financial stability in the United Kingdom, banking on prudence', OECD Economics Department Working Paper No. 716.

Davis, E.P. and Karim, D. (2008a), 'Comparing early warning systems for banking crises', Journal of Financial Stability, 4, pp. 89-120.

--(2008b), 'Could early warning systems have helped to predict the sub-prime crisis?', National Institute Economic Review, 206, pp. 35-47.

--(2008c), 'Research in and policy for financial stability--what have we learnt?', Economics and Finance Working Paper, No. 08-14, Brunel University.

De Bandt, O. and Hartmann, P. (2000), 'Systemic risk: a survey', ECB Working Paper No. 35, European Central Bank.

Degennaro, R.P. and Thomson, J.B. (1996), 'Capital forbearance and thrifts: examining the costs of regulatory gambling, Journal of Financial Services Research, 10, pp. 199-211.

Demirguc-Kunt, A. and Detragiache, E. (2005), 'Cross-country empirical studies of systemic bank distress: a survey', IMF Working Papers 05/96, International Monetary Fund.

--(1997), 'The determinants of banking crises--evidence from developing and developed countries', IMF Working Papers 97/106, International Monetary Fund.

Diamond, D.W. and Dybvig, P.H. (1983), 'Bank runs, deposit insurance, and liquidity', Journal of Political Economy, 9, 3, pp. 401-19.

Duttagupta, R. and Cashin, P. (2008), 'The anatomy of banking crises', IMF Working paper 08/93.

ECB (2009), EU Banking Sector Stability August 2009, Frankfurt, European Central Bank.

Elsinger, H., Lehar, A. and Summer, M. (2006), 'Risk assessment for banking systems', Management Science, 52, pp. 1301-14.

Evans, G.(1991), 'Pitfalls in testing for explosive bubbles in asset prices', American Economic Review, 81, pp. 922-30.

Farhi, E. and Tirole, J. (2009), 'Collective moral hazard, maturity mismatch and systemic bailouts', NBER Working Paper, No. 15138.

Fell, J. and Schinasi, G. (2005), 'Assessing financial stability--exploring the boundaries of analysis', National Institute Economic Review, 192, pp. 70-85.

Fernandez de Lis, S. and Herrero, A. (2009), 'The Spanish approach: dynamic provisioning and other tools', in Mayes, D., Pringle, R. and Taylor, M. (eds), Towards a New Framework for Financial Stability, Central Banking Publications.

Freixas, X. and Parigi, B. (1998), 'Contagion and efficiency in gross and net interbank payment systems', Journal of Financial Intermediation, 7, pp. 3-31.

Freixas, X., Parigi, B. and Rochet, J.-C. (2000), 'Systemic risk, interbank relations and liquidity provision by the central bank', Journal of Money, Credit and Banking, 32, pp. 611-38.

FSA (2009), The Turner Review. A Regulatory Response to the Global Financial Crisis, March, London.

Goodhart, C. (2005), 'Financial regulation, credit risk and financial stability', National Institute Economic Review, 192, pp. 70-79.

--(2008), 'The boundary problem in financial regulation', National Institute Economic Review, no. 206, January, pp. 48-55.

Goodhart, C., Sunirand, P. and Tsomocos, D. (2006), 'A model to analyse financial fragility', Economic Theory, Springer, 27, pp. 107-42.

HM Treasury (2008), Financial Stability and Depositor Protection: Strengthening the Framework, January, London, HM Treasury.

Hoggarth, G. and Sapporta, V. (2001), 'Costs of banking system instability, some empirical evidence', Financial Stability Review, Summer, Bank of England.

IMF (2009), Global Financial Stability Report October 2009--Navigating the Financial Challenges Ahead, Washington DC, International Monetary Fund.

Jayanti, S. and Whyte, A. (1996), 'Global contagion effects of the continental Illinois failure', Journal of International Financial Markets, Institutions and Money, 6, pp. 87-99.

Kaminsky, L.G. and Reinhart, C.M. (1999), 'The twin crises; the causes of banking and balance-of-payments problems', American Economic Review, 89, pp. 473-500.

Kane, E. (1992), 'Corporate capital and government guarantees', Journal of Financial Services Research, 5, pp. 357-68.

Kashyap, A., Rajan, B. and Stein, J. (2008), 'Rethinking capital regulation', paper presented at the Kansas Fed Symposium at Jackson Hole, Wyoming, August.

Kindleberger, C.P. (I 978), Manias, Panics and Crashes, A History of Financial Crises, New York, Basic Books.

Knight, F.H. (1921), Risk, Uncertainty and Profit, Boston, No. 16 in a series of rare texts in economics, republished by the LSE.

Kocherlakota, N. and Shim, I. (2006), 'Forbearance and prompt corrective action', BIS Working Paper No. 177, Bank of International Settlements.

Mayes, D., Pringle, R. and Taylor, M. (eds) (2009), Towards a New Framework for Financial Stability, London, Central Banking Publications.

Minsky, H.P. (1977), 'A theory of systemic fragility' in Altman, E.I. and Sametz, A.W. (eds), Financial Crises, New York, Wiley.

Raymond Feingold, H. (2001)', Preventing crisis by testing for bubbles: a comparative study of the financial fragility of Europe relative to the USA and Japan', EFMA 2001 Lugano Meetings, European Financial Management Association.

Santor, E. (2003), 'Banking crises and contagion: empirical evidence', Working Papers 03-1, Bank of Canada.

Shin, H.S. (2008), 'Risk and liquidity in a system context', Journal of Financial Intermediation, 17, pp. 315-29.

Warwick Commission (2009), The Warwick Commission on International Financial Reform, in Praise of Unlevel Playing Fields, Warwick University.

Weale, M. (2008), 'Commentary: the banking crisis and the economy', National Institute Economic Review, 204, April, p.7.

NOTES

(1) Financial crises, and notably banking crises, were infrequent during the Bretton Woods period of fixed exchange rates and exchange controls from the Second World War till the early 1970s. However, this was an historic anomaly, as witness the high frequency of crises in the interwar and pre-First World War periods.

(2) See Barrell et al. (2009) for an estimate of costs and benefits of tighter capital and liquidity regulation in the UK.

(3) Especially when such growth is focussed on securities and/or there is mark-to-market accounting.

(4) They may, however, lead to resource misallocation and an increased cost of capital, with deflationary macroeconomic implications.

(5) It is not denied that all sharp asset price changes will tend to affect market liquidity to a greater or lesser degree.

(6) This suggests that financial crises follow a credit cycle with an initial positive shock (displacement) provoking rising debt, mispricing of risk by lenders and an asset bubble, which is punctured by a negative shock, leading to a banking crisis.

(7) As opposed to risk (in the sense of Knight, 1921), seen as a key feature of financial instability, in that unlike the cycle, one cannot apply probability analysis to rare and uncertain events such as financial crises and policy regime shifts and hence price them correctly. Innovations are by definition subject to such uncertainty as probabilities are not yet known over the full cycle.

(8) This suggests that competitive, incentive-based and psychological mechanisms in the presence of uncertainty lead financial institutions and regulators to underestimate the risk of financial instability, accepting concentrated risks at low capital ratios.

(9) The underlying model which also incorporates contagion via overlapping claims in inter-bank markets, is detailed in Allen and Gale (2000).

(10) At worst the manager loses her job and possibly reputation. Allen (2005) suggests that if default is not penalised and if reputation risk is low, the manager improves her expected return by gambling depositors' funds.

(11) Repayable at the borrowing rate.

(12) A fundamental asset price is simply its payoff discounted by the investor's opportunity cost of her own funds.

(13) This is motivated by the 'first come, first served' process of depositor reimbursement.

(14) Weights are 2005 GDP weights.

(15) Presence of externalities means shocks are amplified symmetrically; a positive shock generates positive externalities and bank balance sheets benefit from asset price booms.

(16) Kane (I 992) defines forbearance towards low banking capital as a situation where "Deposit-institution regulators engage in capital forbearance when they lower standards for minimum net worth at de-capitalised institutions", (page 359). Prompt Corrective Action necessitates immediate termination of critically undercapitalised banks via asset liquidation (Kocherlakota and Shim, 2006).

(17) One way banks did this was by removing assets off the balance sheet by holding asset backed securities in SIVs and conduits, for which banks then guaranteed the asset backed commercial paper financing. The other was holding other banks' AAA ABS tranches on balance sheet.

(18) Consider a phase of vulnerability with growing leverage and mismatch in funding. Looking at individual firms may miss the system-wide credit and liquidity risks since they depend on aggregate lending conditions, other banks' reliance on the same funding source or diversifying in the same manner. There can also be funding chains whereby banks lend at nearly-matched maturities but the system as a whole has a major maturity mismatch.

(19) For a practical guide onto DtD, see Crosbie and Bohn (2002).

(20) Standard deviation of the annual percentage change in asset value.

(21) See for example ECB (2009), pages 27-28.

(22) Credit default swaps, corporate or sovereign bonds.

(23) Hence this market risk measure is backward looking.

(24) E.g. parallel yield curve shifts.

(25) See part 2, third section.

(26) 105 countries are covered by data spanning 1979-2003 which yields 72 or 102 systemic banking crises depending on the crisis definition used.

(27) Such as credit growth, M21FX reserves ratio, the fiscal balance, GDP growth, real interest rates and inflation.

(28) Besides higher capital per se, proposals include improved quality of capital, a leverage ratio to complement risk-adjusted capital adequacy, as well as better quality capital, an international liquidity standard and proposals to improve risk capture of capital across trading as well as banking books.

(29) Even for an individual institution, commercial property lending is risky at an individual level and losses are most highly correlated with aggregate losses by banks (Davis, 1991).

E. Philip Davis * and Dilruba Karim **

* National Institute of Economic and Social Research and Brunel University; e-mail: e_philip_davis@msn.com. ** Brunel University; e-mail: dilly.karim@gmail.com. Parts of this paper are derived from Davis and Karim (2008c) which also appeared in Mayes et al. (2009). An earlier version was keynote address at the 6th Euroframe Conference on Economic Policy Issues in the European Union, 12 June 2009, entitled 'Causes and consequences of the current financial crisis, what lessons for EU countries?' The authors thank participants in the Euroframe Conference and at a seminar at HM Treasury in October 2009 for helpful comments.

Table 1. Generic features of financial instability

Phase of crisis Nature Example of features

Primary Diverse Deregulation, monetary or
(favourable) fiscal easing, productive
shock innovation, change in market
 sentiment.

Propagation-- Common--main New entry to financial
build-up of subject of markets, debt accumulation,
vulnerability macroprudential asset price booms, innovation
 surveillance. in financial markets,
 underpricing of risk, risk
 concentration and lower
 capital adequacy for banks,
 unsustainable macro policy.

Secondary Diverse Monetary, fiscal or regulatory
(adverse) shock tightening, asymmetric trade
 shock.

Propagation-- Common Failure of institution or
crisis market leading to failure of
 others via direct links or
 uncertainty in presence of
 asymmetric information--or
 generalised failure due to
 common shock.

Policy action Common--main Deposit insurance, bank
 subject of recapitalisation and
 crisis guarantees, lender of last
 resolution. resort, general monetary and
 fiscal easing.

Economic Common--scope Credit rationing and wider
consequences depends on uncertainty leading to fall in
 severity and GDP, notably via investment.
 policy action.