118 NATIONAL INSTITUTE ECONOMIC REVIEW No. 192 APRIL 2005 FINANCIAL REGULATION, CREDIT RISK AND FINANCIAL STABILITY C.A.E. Goodhart* In contrast to recent successful developments in macro monetary policies, the modelling, measurement and management of systemic financial stability has remained problematical. Indeed, the focus of most effort has been on improving individual, rather than systemic, bank risk management; the Basel II objective has been to bring regulatory bank capital into line with the (sophisticated) banks’ assessment of their own economic capital. Even at the individual bank level there are concerns over (i) appropriate diversification allowances, (ii) differing objectives of banks and regulators, (iii) the need for a buffer over regulatory minima, and (iv) the distinction between expected and unexpected losses (EL and UL). At the systemic level the quite complex and prescriptive content of Basel II raises dangers of ‘endogenous risk’ and procyclicality. Simulations suggest that this latter could be a serious problem. Keywords: Financial regulation; systemic stability; credit risk; capital adequacy requirements; Basel II; operational risk; internal risk based models; procyclicality; endogenous risk. JEL classification: E4, E44, E5, E55, G2, G21, G28 1. Introduction I have been privileged to have been able to participate, both as an academic and a central bank official, in a massive improvement in the theory and practice of macro monetary policy over the past fifteen, or so, years. A key starting point was the allocation of inflation targetry, together with operational independence, to the Reserve Bank of New Zealand at the end of the 1980s. This established that the overriding function of a central bank was to achieve a single, primary target, price stability, by manipulating its single instrument, short-term interest rates. Since the effect of interest rates on inflation was lagged, this involved setting interest rates now on the basis of forecasts of future output growth and future inflation, a methodology rather loosely modelled by the ubiquitous Taylor reaction functions. Meanwhile, the mechanics of holding short-term interest rates close to their policydetermined level were increasingly being achieved through the adoption of a narrow corridor between a remunerated central bank deposit rate and a higher lending rate, to which commercial banks had automatic access. Again this process was initiated in New Zealand, but has since then been adopted by both the ECB and, this last July, by the Bank of England. Central bank practitioners have been fortunate in having had the benefit of analysis and advice from a number of leading monetary economists, notably Lars Svensson and Michael Woodford, in these reforms, though more often ex post, after the reforms had already been initiated, than ex ante. This has had the consequence that the differences between theory (as represented for example, by Woodford’s 2003 book, Interest and Prices), and practice, as represented by what central bankers (now themselves often professional economists, such as Ben Bernanke, Otmar Issing and Mervyn King), see themselves doing, has become vanishingly small. There is consensus now, where some thirty years ago there was confusion and lack of communication. I believe that these structural and theoretical improvements have led to better policies, and that such better policies have played a role in the greater stability of our economies over the past decade, with a reduction both in the level and volatility of inflation and in the volatility of output growth. This same enthusiastic paean of praise cannot, however, be applied to the conduct of either theory or practice in the area of central bank’s second core purpose, to wit the maintenance of financial stability (FS). The achievement of FS is, however, much more difficult and complex than the accomplishment of price stability, in the guise of inflation targets (IT). Unlike price stability, financial stability cannot be readily measured, modelled, or forecast (see also Fell and Schinasi, 2005, this volume). There is no * Financial Markets Group, London School of Economics and Political Science. My thanks are due to Jon Danielsson, Andy Mullineux, Miguel Segoviano, Ashley Taylor and Geoffrey Wood for help and advice in preparing this article, but responsibility remains with me. GOODHART FINANCIAL REGULATION, CREDIT RISK AND FINANCIAL STABILITY 119 straightforward instrument that a central bank can use to counter deviations from a desired equilibrium, and such mechanisms as can be deployed, such as Capital Adequacy Regulations, have to be agreed at an international level, largely because of the ease of disintermediation, i.e. in the guise of locational shifts of business, within a system of free capital movements, instantaneous electronic transfers, and a global financial system. has shown, it makes the implicit assumption that there is just one single systemic risk factor (other risks being idiosyncratic), to which all borrowers from any given are, to a greater or lesser extent, exposed. For banks which operate within a single country that may be approximately true, though even in this case there are differing sectoral, industrial, and in a sizeable country, geographical, risks, so the benefits of diversification are not being fully rewarded. There is no consensus either between academics and practitioners, or indeed within either camp, on how the financial stability objective might best be pursued. In this context, for example, the recent massive labours of the Basel Committee on Banking Supervision did not start from any economic theory of how best to establish Capital Adequacy Requirements (CARs) in formulating Basel II. This is not surprising since no such consensus theory exists. The smaller, and less developed, the country, the more likely is risk going to be concentrated. The exposure of banks in Iceland to the continued success of the fishing industry, in Hong Kong to the local property market, in Mauritius to the textile industry, will all have been large. In this context, the worldwide development of the credit default swap (CDS) market, which allows a separation of the specialist origination of loans from the need to continue to bear those loans’ credit risk to maturity may do more to reduce risk concentrations and resultant financial crises than all the recent reforms to the CARs. It is such credit default swaps that in my view will do most to allow banks, wherever sited, effectively to achieve a desired level of diversification. But, like all derivatives, they allow those who use them either to assume, or to lay off, risk; and such deals may, or may not, be correctly priced. So, like other powerful instruments they can be used for good or ill; what is perhaps most needed by regulators is greater transparency. What does exist instead are models and measurements for individual1 bank risk, often developed by the commercial banks themselves as managerial control tools; VaR and KMV models are probably now the best known of these. To some large extent the public sector officials at Basel tried to piggy-back official CARs on the basis of the (best available) commercial bank models. Indeed a proud boast of the authors of Basel II is that this reform has brought regulatory capital much closer into line with the economic capital that the more sophisticated banks would have adopted on their own, and as a desirable by-product helped to educate the less sophisticated banks about optimal risk management. 2. Some problems with Basel II Even taken on its own terms, as an approach to make the individual bank manage risk better, there are a number of problems. I shall mention four here briefly. These concern: (i) portfolio theory, (ii) differing requirements, i.e. differing objective functions between regulators and managers, (iii) the need for a buffer over required minima, and (iv) the distinction between expected loss (EL) and unexpected loss (UL). Let us start with portfolio theory. Almost the first lesson in finance is that the risk of a portfolio is determined by the covariances between its constituent elements, not their individual variances. After all, if you can find any two assets with perfect negative covariance, you can combine them into a riskless asset, irrespective of their individual variances. In dealing with covariances and correlations between asset returns, Basel II is certainly far superior to Basel I, but still leaves much to be desired. Not only does it ignore non-linear dependence, but also, as Gordy (2003) But to return to my main topic, outside the EU, in the US for example, Basel II is seen as most appropriate to large sophisticated global banks. While it is true that there is something of a common cycle amongst developed countries, it is far from general; note the differing time paths of the USA, Japan and the EU. Moreover the correlation between fluctuations in GDP in the developed (Northern) countries and the developing (Southern) world, and also perhaps now between Western and Eastern EU states, is even lower. Thus lending to borrowers in emerging economies is likely to become less attractive to large international banks, because such borrowers will probably be unrated, and hence carry a larger riskweighting, without any offset for their effect in diversifying and lowering the exposure to common developed-country cycles. On all this, see the papers by Segoviano and Lowe (2002), Altman et al. (2002) and Griffith-Jones et al. (2002). Let me turn next to the differing requirements, and objective (loss) functions, of regulators and bankers. Bankers are concerned naturally with the fate of their own individual institution, not of the welfare of the system as a whole. Moreover, there are institutional arrangements, 120 NATIONAL INSTITUTE ECONOMIC REVIEW No. 192 APRIL 2005 such as limited liability and generous bankruptcy arrangements, and conditions such as when capital has already been eroded, that may lead bankers willingly to assume more risk than regulators would want. It can, therefore, be problematical for regulators just to piggyback on techniques developed by bankers for their own, perfectly proper, purposes. Let me take two examples. The first, which has been splendidly dissected by my colleague, Jon Danielsson, (2002), relates to the VaR, value at risk technique. This was developed, entirely sensibly, by commercial bankers to give them a metric of their market risk under normal conditions. In most applications, excluding those using long historical data sets, it is used on the conditional assumption of a normal distribution of asset returns (log normal prices). But asset market returns have fat tails; that is, the probability of really large jumps in asset prices (remember October 1987) is far greater than in a normal distribution. Thus while a VaR metric is a perfectly respectable technique for bankers, it is not for regulators who need to focus on adverse tail events, for which the appropriate measurement technologies are quite different. One reason why we have official regulations at all is that there may be externalities, so the social costs may differ from the private costs. The externality that regulators fear above all in banking is that of contagion, that the collapse of one institution may have a domino effect on others, and possibly on financial markets and systems, such as the payment system. As already noted, a particular problem about credit risk is that this does tend to be systemic, so the failure of one bank will tend to be correlated with fragility in others; and so that initial failure will, for a variety of reasons, via direct interconnections and also through reputational effects, drive other banks to the brink. Hence the social cost of failure may well be greater than the private. It is far from clear whether any such externalities attend operational risks and, in those few cases where they may do so, whether more capital is a suitable remedy. Such risks, of credit card fraud, computer failure, trader fraud, mis-pricing, etc., certainly exist, and banks are indeed right to apply their own internal capital against those of such risks (not all) that can be reasonably quantified, e.g. credit card fraud. But it is difficult, at least for me, to see why a Nick Leeson at Barings or the reputational failure of a bank involved in an Enron, or Parmalat, scandal has any obvious potential downside effects on other commercial banks, or even why internal capital represents a sensible prophylactic in such cases.2 If the costs of operational risks are fully internalised, what then is the case for having a socially imposed minimum requirement? Just because operational capital is applied by good international banks does not of itself provide any justification for it to be a public requirement. There are a few cases where operational risk does have externalities. A computer failure, and/or the absence of a secure back-up after such a failure (perhaps due to terrorism or natural causes), can prevent a bank from making payments, and thereby adversely affect the liquidity of other banks. Indeed so, but insofar as capital has any relevance in such an example (in contrast to lending by the central bank which does), the capital of a bank needs to rise the greater the threat of such operational failures in other banks. Moreover, if the central bank does provide the necessary liquidity, as in the case of 9/11 and the Bank of New York computer failure at an earlier date, then there would be no need for capital in such cases. A critic might, indeed, argue that the claim that one aim of Basel II was to make regulatory capital requirements accord closer to economic capital implied, ipso facto, that those involved had not stopped to ask themselves on what underlying principles public regulation could, and should, be justified. Moreover, it is just not possible to make regulatory capital equate to voluntary desired economic capital. Regulation implies by definition that the designated capital levels are required. If they are thereby required to be maintained, there must be some form of sanction, if only reputational or levied in terms of additional visitations from the supervisors, in those cases where the minimum requirement is breached. In the context of the Basel approach, which involves discussions between a small group of self-appointed regulators, central banks and specialised supervisors, not a treaty between countries, it has not, however, been possible to establish a common approach towards sanctions. Indeed one of the main weaknesses of the Basel approach is that the Committee focuses on best practice without any discussion on how to respond to shortfalls from such best practice. In that the Basel approach differs sharply from the American Federal Deposit Insurance Corporation Improvement Act (FDICIA) of 1991, which is much more broad brush on capital requirements, but detailed and prescriptive on a ladder of sanctions as shortfalls increase. Be that as it may, there will in each country be some, though country-specific, expectation of sanction for breaching the CARs. Consequently an (optimising) bank will want to maintain some buffer of free (or, though the word is inappropriate, excess) margin of spare capital over GOODHART FINANCIAL REGULATION, CREDIT RISK AND FINANCIAL STABILITY 121 the regulatory minima. While in some cases I have heard claims that Basel II will have the, supposedly desirable, effect of equating regulatory and economic capital, in other cases I have heard the claim that, since most banks already hold capital quite well in excess of likely Basel II requirements, its introduction will have no effect whatsoever. Neither of those positions is correct. A recent empirical study by the FSA (Alfon et al., 2004) found that an increase in the regulatory requirement of, say 10 per cent, was matched in the longer term by a change of about 7 per cent in the desired level of capital, i.e. part, about one third, of the regulatory change is absorbed in a change in the desired level of the buffer, but the greater part, around two thirds or more, feeds through to the desired level of capital.3 Let me turn next to the fourth problem that I want to highlight, which is the distinction between expected and unexpected loss, EL and UL. As we are well aware, a key feature of Basel II is the focus on the probability of default, PD. Let me show a diagram of simulated distributions over time of PD for two different types of credit extension, perhaps credit card lending as contrasted with lending to foreign sovereigns (see chart 1). The distribution for credit card lending is A, that for sovereign lending is B. For which type of loan should a bank hold more capital? If you just look at average PD the answer is obviously A, credit card lending. But note that, with credit card lending, the expectation of default losses can, as assumed here, be estimated quite closely in advance. If default probability can thus be estimated in advance, the appropriate default premia can, and should, be included in the interest rate charged. If the relative ex ante interest rate differential does take account of the mean expected default probability, then capital should be required not against mean expected PD, but against the variance and skew of the expected distribution of PD. Indeed, on this assumed simulation, banks should be Chart 1. % occurrence A 0 B 5 10 15 Distribution of expected loss 20 required to hold much more capital against sovereign lending than against credit card lending, which should require hardly any capital backing. This issue did arise, although it seems rather late in the day, at Basel in the guise of the discussion whether capital should be applied against expected loss (EL) or just against unexpected loss (UL), and in which particular cases. But, once again, I did not note any generic discussion on the economic principles determining the need for capital, and in particular the relative roles of default risk premia in interest rates on the one hand as compared with capital on the other. There is no question but that regulators and supervisors are closely concerned about such credit risk premia. There is, for example, often a fear in such quarters that such premia are excessively volatile, going down too far in boom, and calmer, times, and rising so far during periods of crisis as to interrupt completely flows of new loans to those sectors and countries perceived as now suddenly risky. A diagram of spreads, over US Treasuries, on bonds of emerging countries illustrates this point (see chart 2). One reason regulators and supervisors are hesitant about going all the way towards the UL concept, that is to focus on the expected distribution, not the mean, of PD, is that they fear that, whereas expected PD could in principle be met by an appropriate interest differential, in practice it will not be. There have been numerous occasions when regulators have complained that various factors, e.g. excessive competition, state support for certain banks, especially of public sector intermediaries, such as the Post Office Savings Bank in Japan, have held down interest margins below the level consistent with a healthy banking system. Given that regulators have not felt themselves able to interfere, or to intervene directly, in the interest rate setting decisions of banks, this has meant that the regulators have felt the need to maintain capital requirements in most cases against EL, rather than UL. When there is a higher capital requirement per unit of lending, there should be some pressure on banks to raise interest rate margins on average to maintain the return on equity. But a likely effect of Basel II will be to lower capital requirements for the largest international banks, which may be a reason why they supported its introduction, and they are likely to take the lead in setting interest rates. Moreover, the possibly greater procyclicality of CARs under Basel II, a subject to which we shall turn shortly, may also have the effect of enhancing the procyclicality of credit default spreads, which is already a 122 NATIONAL INSTITUTE ECONOMIC REVIEW No. 192 APRIL 2005 Chart 2. Emerging market sovereign bond yield spreads Basis points 1,000 900 800 700 600 500 400 300 200 100 0 Jan. 02 July 02 Jan. 03 July 03 Jan. 04 July 04 Source: JP Morgan Chase & Co. matter of some concern. More generally, the financial strength of banks, and banking systems, is a somewhat complex combination of both capital ratios and interest rate spreads and profit margins. Focussing solely on one of these two legs is likely to lead to a somewhat unbalanced position. 3. Should we abandon risk-related capital adequacy ratios? Indeed for these and other related reasons, I have come at long last to the reluctant conclusion that the concept of relating officially set and required capital ratios (CARs) to the relative riskiness of a bank’s portfolio of assets has been, in practice, a wrong turn. The idea that officially set and required CARs should be risk-related appears intuitively natural, even obvious. Surely a bank with a portfolio full of risky loans to property speculators is more likely to go bust than a bank holding government Treasury bills, and should therefore be required to have a larger capital buffer. Most commentators have strongly supported the idea of such risk-related CARs, and indeed I did so myself until quite recently. Let me start by recalling the several drawbacks of riskrelated CARs. First, in a sophisticated and fast-changing market, it will be well-nigh impossible for officials, however able and devoted, to get the assessment of relative risks correct, and, if accurate now, innovation will soon make them outdated and erroneous. Even after all the huge effort that has been put into the Basel II process, there are still likely to be several major deficiencies in relative risk assessment. For example, as already noted, the basic structure of the credit risk approach adopted effectively rests on the premise of a single systematic risk factor (Gordy, 2003; Repullo and Suarez, 2004), which is in most cases presumably the domestic economy. If there are – as is surely the case – numerous other systemic risk factors, then the Basel II approach probably gives insufficient weight to the risk mitigation inherent in diversification across countries and across industries. Be that as it may, even with much improved risk assessment, officials will, indeed can, never get it exactly right. There will always remain gaps, lacunae and errors that can be exploited for regulatory arbitrage and by ‘gaming’, and these will only increase over time. Second, the attempt to achieve ever more accurate risk measurement inevitably increases the complexity of the whole exercise. A comparison of Basel II and Basel I makes the point. Moreover, any attempt to fill some of the remaining gaps in Basel II would just make this whole problem worse. Third, despite the reliance on banks’ internal risk-based assessments, in the Foundation and Advanced IRB versions, Basel II appears, at least to this observer, to provide a rather pervasive model for how a commercial bank’s own risk analysis should be undertaken. Many argue that this will bring major benefits. They claim that the application of Basel II will encourage many, perhaps most, banks to bring their own internal risk-assessment models up to speed. Indeed, one of the main benefits of Basel II is seen to lie in its impetus to improving the education of many commercial bankers on the basis of proper risk assessment. No doubt there is truth in this, but is there not also a danger that as prescriptive an approach as Basel II could lead to an excessive focus on one single methodology, in a context where uncertainty and innovation suggest the advantage of encouraging multiple competing models of risk assessment? And could that focus also enhance ‘herd behaviour’, whereby all banks tend to respond similarly and simultaneously to common stimuli? This can increase the riskiness of the system as a whole, even if each individual bank appears to be behaving exactly according to the new, ‘improved’, rules. This interactive effect has been termed ‘endogenous risk’ by my colleagues Danielsson, Shin and Zigrand (2004). Blum and Hellwig (1995) have also emphasised that the macro effect of a (regulatory) measure cannot necessarily, or simply, be ascertained just by considering the individual micro effects, also see Summer (2003). GOODHART FINANCIAL REGULATION, CREDIT RISK AND FINANCIAL STABILITY 123 The important result to observe is the much greater variance of the simulated outcomes for the IRB than for the standardised or ICRM approaches. During periods of strong growth, high profits and low non-performing loans (USA in the mid-1990s and Norway in 1997), the IRB has a lower CAR than the standardised approach in all our developed countries; whereas in recessions (e.g. USA in ICRM 9.60 8.94 8.93 9.13 9.46 9.46 9.46 9.56 9.56 9.99 9.69 9.29 8.90 8.51 8.25 8.29 8.31 8.40 8.41 8.53 8.31 8.11 8.96 0.34 8.59 7.19 7.62 8.02 9.99 9.82 8.66 10.80 11.68 11.43 8.06 6.47 5.40 5.56 5.65 5.94 6.51 7.81 8.13 8.25 8.18 6.60 8.02 3.39 8.07 6.80 7.031 7.26 8.74 8.55 6.99 6.49 7.60 7.54 6.47 4.67 3.78 4.09 4.32 4.84 5.83 6.70 7.16 7.24 6.78 6.26 6.51 1.95 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Average Variance Chart 3. CARs for the USA 14 12 10 8 6 4 2 Standardised IRB F 2002 2000 1998 1996 1994 1992 1990 0 1988 Anyhow, we have simulated the time paths of CARs under each of our three approaches, standardised, IRB Foundation (IRB F) and ICRM, for our various countries, and the results are set out in tables 1 to 3 and charts 3 to 5. IRB F 1986 The results of this exercise for the three countries examined are stark. We compared the implied capital requirements for our ‘typical’ bank under three regulatory regimes; first the standardised approach in Basel II (which is close to that applied in Basel I); second, the Foundation Internal Ratings Based (IRB) approach (i.e. assuming a constant Loss Given Default, since we have no good time series in any country for average LGD); and third, an Improved Credit Risk Method (ICRM). This latter uses a Merton approach to model credit quality changes and an indirect approach to model correlations amongst the individual credits in the overall portfolio. The construction of an ICRM is, however, quite complex, and interested readers should consult our companion paper, Goodhart and Segoviano (2004). Standardised 1984 My colleagues and I have done some work on this subject (Goodhart, Hofmann and Segoviano, 2004). Very briefly we reconstructed a typical bank portfolio as follows. We assumed that each portfolio consisted of 1000 loans, each one with equal exposure. From each specific country data sources, we obtained the through time proportion of assets (bonds for the USA or corporate loans for Mexico and Norway) that were classified under each of the reported ratings for a given country. With this information we constructed the benchmark portfolio that we used to compute capital requirements at each point in time. Period 1982 4. Is procyclicality a serious worry? Table 1. CARs for the USA Percentage That consideration leads on to the fourth problem of relating capital requirements to a common measure of relative risk, which is, of course, that it is likely to engender procyclicality. This is now widely understood, and so much has been said on the general point that it is unnecessary to add more. The issue has moved from that of general theory to a question of specific empirical quantification. Is it likely to be a serious problem in practice, or not? ICRM 1990/1, Mexico mid 1995/6 and Norway in 1994/5), the CAR is markedly higher for the IRB than in the other two approaches. In Mexico, an emerging market economy, the average quality of loan is lower throughout than in developed countries, so the IRB gives a higher CAR in all years but, as in developed countries, the variance of the CAR (up in recessions as in 1995/6, and lower during the better years) is greater for the IRB than in the other two approaches. 124 NATIONAL INSTITUTE ECONOMIC REVIEW No. 192 APRIL 2005 Table 2. CARs for Norway Period 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Average Variance Table 3. CARs for Mexico Standardised IRB F ICRM Period 9.99 10.27 10.47 10.37 10.27 10.94 11.32 10.67 10.27 10.27 10.27 10.27 10.36 10.46 10.44 0.11 8.31 9.28 9.78 9.93 9.52 13.24 14.07 12.14 8.86 9.00 9.219 9.49 9.65 9.76 10.16 2.94 7.58 8.13 8.68 9.03 9.19 9.82 11.08 9.72 7.32 7.42 7.53 7.93 8.33 8.34 8.58 1.19 March 1995 June 1995 Sept. 1995 Dec. 1995 March 1996 June 1996 Sept. 1996 Dec. 1996 March 1997 June 1997 Sept. 1997 Dec. 1997 March 1998 June 1998 Sept. 1998 Dec. 1998 March 1999 June 1999 Sept. 1999 Dec. 1999 Average Variance Chart 4. CARs for Norway 16 Standardised IRB F ICRM 8.77 9.22 9.30 9.49 9.25 9.49 9.56 10.30 9.43 9.27 9.40 8.93 8.81 8.85 9.06 9.04 9.05 8.98 9.14 8.97 9.22 0.12 13.86 16.65 17.10 18.15 17.07 18.45 19.42 24.23 19.09 17.50 18.25 15.19 14.40 14.43 15.55 15.46 15.52 15.30 15.98 15.35 16.85 5.64 10.46 12.29 12.71 12.82 12.59 13.25 14.89 17.65 15.15 13.90 14.34 14.80 13.67 12.26 11.62 11.80 12.00 12.25 12.73 12.10 13.16 2.59 14 Chart 5. CARs for Mexico 10 30 8 6 25 4 20 The implication of this is that procyclicality may well still be a serious problem with Basel II, even after the smoothing of the risk curves that were introduced between CP2 and CP3 to mitigate this problem. Basel II will, however, be a regime change, and one of the purposes of this is to make bankers more conscious of risk assessment and risk management. It has already succeeded in this. One hope is that it will induce bankers to be more Standardised IRB F Sep-99 Mar-99 Sep-98 Mar-98 Sep-97 0 Mar-97 It follows that the percentage change in the required CAR under the IRB as a country moves from boom to recession (up) and back to boom again (down) is likely to be much more extreme under the IRB than under the other two approaches. This is shown in table 4. 5 Sep-96 ICRM 10 Mar-96 2001 1999 1997 IRB F 15 Sep-95 Standardised 1995 1993 1991 1989 0 Mar-95 2 Percentage Percentage 12 ICRM prudent during booms despite declines in CARs. An implication of a move from the standardised to an IRB approach is that the individual bank making this transition will be encouraged to shift its portfolio to higher-quality, higher rated credits, because it then benefits from a lower CAR. This is good of itself, but the higher the quality of the credit, the steeper is the risk curve (relating required risk ratio to the quality of the loan); so the procyclicality is likely to be enhanced, even if average quality improves. GOODHART FINANCIAL REGULATION, CREDIT RISK AND FINANCIAL STABILITY 125 Table 4. Maximum percentage change in CARs Upwards 2 consecutive periods 1 Period Date A. IRB USA NORWAY MEXICO 0.25 0.39 0.25 1989 1994 Dec 96 B. ICRM USA NORWAY MEXICO 0.21 0.13 0.18 C. Standardised USA NORWAY MEXICO 0.04 0.07 0.08 Downwards Date 2 consecutive periods Dates 1 Period 0.33 0.45 0.30 1989/0 1994/5 Sep/Dec 96 –0.29 –0.27 –0.21 1992 1997 Mar 97 –0.49 –0.41 –0.30 1992/3 1996/7 Mar/Jun 97 1998 1995 Dec-96 0.33 0.20 0.30 1998/9 1994/5 Sep/Dec 96 –0.28 –0.25 –0.14 1993 1997 Mar-97 –0.47 –0.37 –0.22 1993/4 1996/8 Mar/Jun 97 Jun-05 Jun-05 Dec-96 0.06 0.10 0.08 1985/6 1994/5 Sep/Dec 96 –0.07 –0.06 –0.08 1983 1997 Mar-97 –0.09 –0.10 –0.10 1994/5 1996/7 Mar/Jun 97 When a regime change is introduced, no one in truth can predict its ramifications, certainly not me. Nevertheless these simulations suggest that procyclicality could remain a serious concern. It is even possible that with the advent of a serious downturn, if one was to occur, the impact of abiding by the IRB would be too severe for the authorities in some countries to countenance. Perhaps, like the Stability and Growth Pact, it would only be observed in the breach when it began to bite hard. Possibly an even greater worry might be that the adoption of Basel II, while not being so adverse as to force reconsideration, might yet exacerbate future economic fluctuations. Certainly there remains a tension between relating CARs more closely to underlying risks in individual banks, and in trying for macroeconomic purposes to encourage contracyclical variations in bank lending in aggregate. How to square this circle is the subject of the following section. 5. A second instrument? However desirable on other grounds the recent changes to the accounting and regulatory regimes, with the shift to fair (market) values under the International Accounting Standards (IAS) and the adoption of Basel II, they will do nothing to check such bank lending/asset price cyclical volatility. Indeed, if some of the more pessimistic prognostications about procyclicality turn out to be justified, such volatility may be considerably exacerbated. In the meantime the sole instrument that central banks currently wield, their command over short-term interest Dates rates, is predicated to the maintenance of stability in the consumer price index, and rightly so. Despite proposals to shade interest rate decisions to offset asset price volatility (e.g. Cecchetti et al., 2000), the difficulties of doing so (see Greenspan, 2002, Bernanke and Gertler, 1999) are considerable. Rather than distort the use of the interest rate instrument to try to achieve a second objective, what is needed is a second instrument. The purpose of this second instrument would be to maintain systemic financial stability. This latter objective remains a core purpose of central banks, whether or not they also supervise the individual banks, and is complementary to their primary role in achieving price stability, but – at present at least – this is a field where central banks have few, if any, stabilising instruments at hand, apart from emergency liquidity assistance to help mop up after something goes wrong in the financial system. The need is to introduce an instrument that will have countercyclical characteristics, which could serve to check bank lending during asset price hikes, and vice versa. The BIS have been advocating such general measures for some time (Borio and White, 2003). In the Conclusions to their 74th Annual Report (2004), the BIS wrote (p. 143), “Fortunately, the global economy now appears to be on an upward path, and there is less call for macroeconomic stimulus. We should use this opportunity to reflect on the processes that allowed our armoury of macroeconomic instruments to become so depleted. An obvious point, but not without objections, is that this situation should be addressed directly through 126 NATIONAL INSTITUTE ECONOMIC REVIEW No. 192 APRIL 2005 more aggressive tightening in good times. In addition, policies to strengthen the financial system, and to encourage more prudent lending behaviour in upturns, might help to mitigate the damage in downturns and reduce the need to resort to aggressive policy easing in the future.” The next question is how do you do this? In an earlier paper (Goodhart and Hofmann, 2001), I had suggested that, analogously to the method of measuring the output gap, we could use deviations of asset prices from a smoothed (Hodrick-Prescott filter) trend to assess the gap between the current asset price and its ‘fundamental’ value. But that got roundly criticised on the grounds of inconsistency with efficient market hypotheses. Borio and Lowe (2002), also see Borio, Furfine and Lowe (2001) and Borio (2005, this volume), suggest that the rate of growth of bank lending itself is the key determinant of future asset price movements, but the lags are long and the relationship subject to structural changes in financial intermediation, etc. My own view now is that a better (perhaps best) approach would be to relate the capital requirement on bank lending to the rate of change of asset prices in the relevant sector. Thus the capital adequacy requirement (CAR) on mortgage lending could be related to the rise in housing prices (relative to HICP inflation), and lending to construction and property companies to the rise in property prices. For manufacturing and services more broadly, the CAR could be related to the rise in equity prices, up when equity prices were appreciating, and vice versa. Similarly, required solvency ratios for life insurance companies would be adjusted counter-cyclically in response to shifts in equities, bond and property prices. The purpose of the exercise would be both to build up reserves and to restrain bank lending during asset price booms, so as to release them during asset price depressions. In this respect it has much in common with the current Spanish pre-provisioning policy proposals. The flip-side, however, is that it relaxes prudential requirements most during bad recessions, just when individual banks and other financial intermediaries are individually at their most fragile. But when the concern of the central bank is for the aggregate, systemic state of the system, surely this is the right course. There are numerous practical problems, notably the possibility of disintermediation abroad in a world without exchange controls where a global financial system exists. My belief is that such problems, though real and serious, can be overcome, and I have started to do some work on this subject (see Goodhart and Hofmann, 2004). 6. Conclusions One of the reasons why I came to enjoy economics as an undergraduate was that it was such an immature subject. Unlike much of my prior studies at school, there was no necessarily correct answer to the various questions being asked. As can be seen here, this certainly remains the case in the field of financial regulation. The current adoption of Basel II, and the new accounting standard, IAS39, do represent improve-ments on what had gone before, but clearly Basel II has deficiencies and problems. It is, perhaps, best seen as a stage in a continuing process of refining and reforming our regulatory system. At present neither theory nor practice seems firmly anchored. It will be for the next generation to make progress. NOTES 1 2 3 It is possible to measure and to model the risk of a single asset quite accurately. It is somewhat more difficult, but possible, to model the risk of portfolios of assets, e.g. because of non-linear dependence, time-varying correlations. It becomes even harder to model the risk of a banking institution incorporating asset portfolios and operational functions. It is hardest of all to measure the risk of a system of banks, though in work done with P. Sunirand and D. Tsomocos, I have recently been trying to do just that, see Goodhart, Sunirand and Tsomocos, (2003, 2004 and 2005). In effect, the greater the degree of aggregation, the more difficult financial risk assessment becomes. I am grateful to Jon Danielsson for this latter thought. Instead, as Instefjord, et al. (1998) have argued, a better approach is to apply incentives/penalties to managers who unearth and prevent such frauds, also see Goodhart (2001). The short-term effect is less, about 50%. Also, the adjustment appears to be asymmetric, in that banks experiencing a decrease in their requirement only adjust their actual capital by about 20% of that. REFERENCES Alfon, I., Agrimon, I. and Bascuñana-Ambrós, P. (2004), ‘What determines how much capital is held by UK banks and building societies?’, FSA Occasional Paper Series, 22, July. Altman, E., Elizondo, A. and Segoviano, M. (2002), Medicion Integral del Riesgo de Credito, Mexico, Editorial Limusa, December. Bank for International Settlements (BIS) (2004), Annual Report, No. 74 (Basel, BIS). Bernanke, B. and Gertler, M. (1999), ‘Monetary policy and asset price volatility’, in New Challenges for Monetary Policy, Federal Reserve Bank of Kansas City, proceedings of the Jackson Hole Conference, August 26–28. GOODHART FINANCIAL REGULATION, CREDIT RISK AND FINANCIAL STABILITY 127 Blum, J. and Hellwig, M. (1995), ‘The macroeconomic implications of capital adequacy requirements for banks’, European Economic Review, 39, 3–4, pp. 739–49. Borio, C., Furfine, C. and Lowe, P. (2001), ‘Procyclicality of the financial system and financial stability: issues and policy options’, in Marrying the Macro and Micro-Prudential Dimensions of Financial Stability, BIS Papers, 1, March, pp. 1–57. Borio, C. and Lowe, P. (2002), ‘Asset prices, financial and monetary stability: exploring the nexus’, paper presented at the BIS conference on ‘Changes in Risk through Time: Measurement and Policy Options’, BIS working papers, 114, July. Borio, C. and White, W.R. (2003), ‘Whither monetary and financial stability? The implications of evolving policy regimes’, in Monetary Policy and Uncertainty: Adapting to a Changing Economy, Jackson Hole Symposium, Federal Reserve Bank of Kansas City, August. Cecchetti, S., Genberg, H., Lipsky, J. and Wadhwani, S. (2000), Asset Prices and Central Bank Policy, Geneva Report on the World Economy 2, CEPR and ICMB. Danielsson, J. (2002), ‘The emperor has no clothes: limits to risk modelling’, Journal of Banking and Finance, 26(7), pp. 1273–96. Danielsson, J., Shin, H.S. and Zigrand, J.-P. (2004), ‘The impact of risk regulation on price dynamics’, Journal of Banking and Finance, 28(5), pp. 1069–87. Goodhart, C.A.E. (2001), ‘Operational risk’, Financial Markets Group, LSE, Special Paper no. 131, September. Goodhart, C.A.E. and Hofmann, B. (2001), ‘Asset prices, financial conditions and the transmission of monetary policy’, paper presented at the Federal Reserve Bank of San Francisco and Stanford Institute for Economic Policy Research conference on ‘Asset Prices, Exchange Rates, and Monetary Policy’, Stanford University, 2–3 March. —(2004), ‘A second instrument?’, Paper presented at Prof. N. Thygesen festschrift, Copenhagen, December 9 (forthcoming in Proceedings). Goodhart, C.A.E., Hofmann, B. and Segoviano, M. (2004), ‘Bank regulation and macroeconomic fluctuations’, Oxford Review of Economic Policy, 20(4), pp. 591–615. Goodhart, C.A.E. and Segoviano, M. (2004), ‘Basel and procyclicality: a comparison of the standardised and IRB approaches to an improved credit risk model’, Financial Markets Group, London School of Economics, Discussion paper no. 524, October. Goodhart, C.A.E., Sunirand, P. and Tsomocos, D.P. (2003), ‘A model to analyse financial fragility’, Oxford Financial Research Centre Working Paper no. 2003fe13. —(2004), ‘A model to analyse financial fragility: applications’, Journal of Financial Stability, 1(1), pp. 1–30. —(2005), ‘A risk assessment model for banks’, Annals of Finance, 1(2), pp. 197–224. Gordy, M.B. (2003), ‘A risk-factor model foundation for ratingsbased bank capital rules’, Journal of Financial Intermediation, 12 (3), July, pp. 199–232. Greenspan, A. (2002), ‘Economic volatility’, Remarks at the Symposium on Rethinking Stabilization Policy of the Federal Reserve Bank of Kansas City at Jackson Hole, Wyoming, August 30. Griffith-Jones, S., Segoviano, M. and Spratt, S. (2002), ‘Basel II and developing countries: diversification and portfolio effects’, December, LSE Financial Markets Group, http://fmg.lse.ac.uk/ pdfs/DP437.pdf, IDStudies, University of Sussex: www.ids.ac.uk/intfinance, summary of article published in The Financial Times, May 13, 2003, p. 15, under the title: ‘A capital idea that will hurt poorer countries’. Instefjord, N., Jackson, P. and Perraudin, W. (1998), ‘Securities fraud’, Economic Policy, 27, October, pp. 587–623. Repullo, R. and Suarez, J. (2004), ‘Loan pricing under Basel capital requirements’, CEMFI, Madrid, mimeo, July. Segoviano, M. and Lowe, P. (2002), ‘Internal ratings, the business cycle and capital requirements: some evidence from an emerging market economy’, BIS Working papers, September, http://www.bis.org/publ/work117.pdf, LSE/FMG: http:// fmg.lse.ac.uk/pdfs/DP428.pdf. Summer, M. (2003), ‘Banking regulation and systemic risk’, Open Economies Review, 14, pp. 43–70. Woodford, M. (2003), Interest and Prices, Princeton, N.J., Princeton University Press.