Systemic Surcharges and Measures of Systemic Importance

Berg, Sigbjørn Atle Research Report Systemic Surcharges and Measures of Systemic Importance Staff Memo, No. 12/2010 Provided in Cooperation with: Norges Bank, Oslo Suggested Citation: Berg, Sigbjørn Atle (2010) : Systemic Surcharges and Measures of Systemic Importance, Staff Memo, No. 12/2010, ISBN 978-82-7553-575-5, Norges Bank, Oslo, https://hdl.handle.net/11250/2507742 This Version is available at: https://hdl.handle.net/10419/210215 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by-nc-nd/4.0/deed.no No. 12 | 2010 Systemic surcharges and measures of systemic importance Sigbjørn Atle Berg, Norges Bank Financial Stability Staff Memo Staff Memos present reports and documentation written by staff members and affiliates of Norges Bank, the central bank of Norway. Views and conclusions expressed in Staff Memos should not be taken to represent the views of Norges Bank. © 2010 Norges Bank The text may be quoted or referred to, provided that due acknowledgement is given to source. Staff Memo inneholder utredninger og dokumentasjon skrevet av Norges Banks ansatte og andre forfattere tilknyttet Norges Bank. Synspunkter og konklusjoner i arbeidene er ikke nødvendigvis representative for Norges Banks. © 2010 Norges Bank Det kan siteres fra eller henvises til dette arbeid, gitt at forfatter og Norges Bank oppgis som kilde. ISSN 1504-2596 (online only) ISBN 978-82-7553-5- (online only) 1 Systemic surcharges and measures of systemic importance1 Sigbjørn Atle Berg Norges Bank Financial Stability 8 November 2010 Abstract There is an emerging consensus that systemically important banks should face stricter regulations and systemic surcharges. To make this latter principle operational we need to quantify the systemic importance of individual banks. This paper reviews the proposed measures of systemic importance from the regulatory and research communities and discusses their merits relative to how we would ideally wish to calibrate surcharges on systemically important banks. 1 The views expressed in this paper are those of the author and may not reflect the views of Norges Bank. The author is grateful for useful comments from his colleagues Farooq Akram, Arild Lund and Bent Vale. All remaining errors are those of the author. 2 1. Introduction The recent financial crisis has demonstrated that some banks are considered too important to fail. After the failure of Lehman Brothers in September 2008 rescue operations have taken place in many countries to prevent the failure of their most important banks and in some cases even other financial institutions. The expectation that this will also happen in the future is widespread, and is an important element in the credit evaluations of rating agencies. The stand alone ratings of large banks are typically several notches below the ratings that take expected government support into account; see e.g. Haldane (2010). Moody’s has warned that if “living wills” effectively reduce the likelihood of a government bail-out, many large banks will be downgraded.2 Market perceptions of an implicit government guarantee creates a well known moral hazard problem routinely discussed in banking textbooks; see e.g. Greenbaum and Thakor (1995). The expected private downside for bank stakeholders of any risky bet is limited by the guarantee, whereas the expected upside is still intact. It will be rational for bank stakeholders to take on more risk than they would otherwise have chosen to do. The recent crisis has demonstrated how costly the downside can be to the government, and eventually to the economy. There is work under way to introduce special regulations for banks that are deemed too important to fail in at least some circumstances; see e.g. the progress report from the Financial Stability Board (2010). A special regulatory regime for systemically important banks may have a number of different components, but one key element could be a systemic surcharge on top of the standard capital adequacy requirement. The purpose of such a surcharge should be to mitigate the risk-taking incentives that are implicit in the banks’ importance to the economy. It can be interpreted as a Pigouvian tax, aiming to bring private profitability calculations in line with social profitability. The regulator’s goal would be to calibrate the tax to provide correct private incentives for the risk-taking of bank stakeholders. The regulator thus needs to consider the divergence between the social and the private profitability of banks’ activities. The divergence in incentives will most likely be bank-specific, depending inter alia on the perceived probability that the bank will be rescued in a crisis. To impose a systemic surcharge, regulators need to develop a methodology for calibrating measures of systemic importance across banks. The private incentives for risk-taking at systemically important banks can also be reduced through other forms of regulation. Of particular interest is the work that is now being done on special resolution regimes for banks, including the “living will” idea introduced by Governor Merwyn King of the Bank of England3 2 Press release on 24 September 2009 . “Living wills” are bank-specific recovery and resolution plans that banks will be required to produce. These plans should provide regulators with all the 3 Speech at the Lord Mayor’s Banquet for Bankers and Merchants of the City of London at the Mansion House on Wednesday 17 June 2009. 3 information needed for the resolution regime to work smoothly. The plans could also be used to initiate organisational changes where the existing structures are found to be too complex.4 Ideally, a special resolution regime should enable the regulator to maintain the systemically important part of any bank as a going concern through a crisis, while the bank stakeholders are covering the losses that have been incurred. Such a special resolution regime, facilitated by bank-specific resolution plans, could in principle be a first best solution to the moral hazard problem. It could eliminate the need for other special regulatory instruments: An ideal resolution regime would ensure that all banks can be wound down quickly without significant negative effects on the financial system and the economy, and with losses mainly borne by bank stakeholders. In that hypothetical scenario no bank would be systemically important, and the moral hazard problem related to any perceived guarantee would disappear. Another potentially important regulation measure that is likely to be implemented is to force a large share of derivative contracts to be cleared by a central counterparty clearing house (CCP) and reported to a trade register. This would greatly enhance the transparency of these markets, reduce counterparty risks, and facilitate crisis resolution procedures for banks with a large portfolio of open derivative positions. Both the US and the European Union have recently introduced legislation that requires more contracts to be cleared through CCPs. However, it seems highly unlikely that any feasible reform of resolution regimes and trading infrastructures will be able to fully eliminate the systemic importance of all banks. Even a very good special resolution regime cannot be expected to correct the moral hazard incentives of all stakeholders. It may allow the regulator to impose losses on shareholders and holders of hybrid capital; and possibly also some losses on senior creditors, although the latter is legally much more complicated given that at least part of the bank operations must be maintained without interruptions. The top management can certainly be replaced, but it seems unlikely that their bonuses or other success fees can be reclaimed. The additional complications raised by crossborder banking only reinforce these conclusions. The consensus is that one should also introduce higher capital, liquidity and supervisory standards for systemically important banks, in order to further reduce their risk taking incentives and probabilities of failure, as discussed in the progress report from the Financial Stability Board (2010). But the need for such additional charges will be less the more effective is the resolution regime in place and the more transparent and risk-reducing are the trading infrastructures. Measures of the systemic importance of banks should thus depend on the regulatory regime. In this paper, we first review measures of systemic importance that have been proposed by regulators and researchers. In section 2 we discuss sets of indicators proposed by the US and the UK authorities, as well as the IMF and the European Commission. In section 3 we look at the research literature, where more specific measures have been proposed. We focus our discussion 4 A thorough discussion of recovery and resolution plans can be found in Huertas (2010). 4 on the important contributions by Adrian and Brunnermeier (2009), Acharya (2009), Huang et al. (2010) and Tarashev et al. (2010). In section 4 we discuss how the calibration of a systemic surcharge should ideally be done. We argue that there are four major caveats to the proposed measures of systemic importance which makes them less satisfactory as a basis for calibrating systemic surcharges on banks. First, the existing proposals do not consider what constitutes an optimal trade-off between the costs of regulation and the costs of bank failures. Second, the proposals do not take into account how the costs of bank failures depend on bank regulation and in particular on the bank resolution regime. Third, the measures proposed by the research community assume that outright failures occur, whereas some form of rescue may be more likely in practice. And fourth, all the proposed measures from the research community depend on market data, which are unlikely to provide unbiased information about the social costs of bank failures. Having argued that the expected social costs of a rescue or resolution operation are relevant for determining systemic surcharge, we go on to take a closer look at the determinants of rescue costs in section 5. In section 6 we discuss the implications for how systemically important banks should be regulated. Section 7 concludes. 2. Regulators’ approach to measurement There are parallel international processes going on at the Financial Stability Board (FSB) and the Basel Committee for Banking Supervision (BCBS) to reform financial regulation. Both processes include efforts to develop practical measures of systemic importance. The FSB (2009) has issued a “Guidance to Assess the Systemic Importance of Financial Institutions, Markets and Instruments: Initial considerations.” This guidance provides three key criteria for identifying the systemic importance of individual banks: • Size as measured by e.g. the size of on and off balance sheet exposures. • Lack of substitutability; how difficult it is for other banks to provide the same services. • Interconnectedness; whether failure or malfunction would have substantial repercussions around the financial system. The guidance suggests that regulators should use a scorecard based on these criteria to determine the degree of systemic importance of each bank. Final policy recommendations from the FSB have not yet been presented, though. The BCBS (2009) has discussed pros and cons of a capital surcharge for systemically important banks. One likely solution is to require more buffer capital at such banks. No list of criteria for systemic importance has been published so far, but a scorecard approach along the lines discussed by the FSB seems likely. 5 There are also regulatory processes under way in the major jurisdictions. The US Treasury Secretary Geithner presented his proposals on June 17, 2009 for “Financial Regulatory Reform”. One of the proposals would give the Federal Reserve Board a special responsibility for regulating and supervising systemically important banks, called “Tier 1 Financial Holding Companies” (Tier 1 FHCs) in the language of the report. The Treasury Report indicated that the systemic importance of a bank could be measured by: • The impact that its failure would have on the financial system and the economy. • Its size, leverage and reliance on short-term funding. • Its importance as a source of credit for the real economy and as a source of liquidity for the financial sector. The Treasury Report was vague on the special regulation of Tier 1 FHCs. The Report stated that there should be e.g. higher capital requirements and more rigorous liquidity requirements, but not how much higher they should be or how they should be graduated. The US Congress approved the “Wall Street Reform and Consumer Protection Act” in July 2010. This Bill establishes a “Financial Stability Oversight Council” (FSOC) and authorizes the Federal Reserve to supervise systemic financial institutions. The bill requires all bank holding companies with total assets above USD 50 billion and other financial companies identified by the FSOC as systemically important to produce resolution plans. The Treasury is authorized to let FDIC take receivership of all failing banks, including breaking up systemically important bank holding companies and other large financial institutions. It is not clear how smoothly that can work in practice if one of the largest and most complex banks should be failing. A special resolution regime for failing banks in the UK came into effect in February 2009. The regime is tailored for handling mainly deposit-taking banks, and it is not clear how useful it will be in cases where failing banks are very complex and have cross-border operations, confer Bank of England (2009a). The ability to resolve systemically important banks remains an open issue. A discussion paper from the Bank of England (2009b) looked at the case for cross-section systemic capital surcharges. This discussion paper stated that the goal should be to reduce the default probabilities of institutions whose failure would cause great damage across the financial system. The paper presented an illustration of how the surcharges can be calibrated to equalise banks’ marginal contribution to social risk. The calibration depends on three factors: • Size, measured as total assets including off-balance sheet items. • Connectivity, measured by the interbank liabilities. • Fire sales impact, measured by the value of repo liabilities. Each factor is measured as a share of the system total, and a composite indicator of systemic importance is computed by arbitrarily giving equal weight to each of these indicators. The level of the total surcharge should be set by requiring a “low probability” for system losses above a certain level. This probability is discussed on the implicit assumption that no rescue operations will take place. 6 The IMF (2010) discussed the implementation of a systemic surcharge in its April 2010 Global Financial Stability Report. This report proposed to sort all banks into a limited number of systemic risk buckets. For this purpose it used a network model where contagion is caused by interbank liabilities and common exposures, and measured the maximum loss that the failure of each bank would impose on other banks in any stage of the credit cycle. The discussion assumed that no rescue operation would take place. This maximum loss without government interference was taken to define the systemic importance of the bank. The European Commission is also considering special regulation of systemically important banks. A background paper suggested that the following indicators might be useful for identifying the degree of systemic importance of individual banks: • Total assets exclusive of derivative assets (TA). • Borrowings to other banks relative to TA. • Lending to other banks relative to TA. • Trading book relative to TA. • Fee and commission income relative to TA. These indicators are very similar to those proposed by the FSB. They are meant to represent size, different aspects of interconnectedness, and substitutability. The Commission is still considering whether these indicators will be their final choice. The general impression is that the regulatory community does have clear ideas about the characteristics of systemically important banks, but are less clear on how that can be compounded into a metric for imposing graduated systemic surcharges. Both the Bank of England and the IMF have published illustrations of how systemic importance can be graduated. However, the graduations do not include any explicit cost trade-offs and do not depend on the specific resolution regime in place. Moreover, they are based on scenarios where no government intervention will take place. Also, they are mere sketches and far from ready for actual implementation. In the US, where special regulation of systemic banks has been introduced, the sole indicator used in practice appears to be total assets, together with the subjective judgement of supervisors. 3. A brief review of some measures proposed by the recent research literature Some contributions from the research community have been much more specific. Systemic risk and contributions to systemic risk have been discussed in the research community for a long time. But in the last few years we have seen a number of more specific proposals on how systemic surcharges on banks can become part of a regulatory reform. We shall briefly review some of the main contributions from the research literature, starting with the influential work of Adrian and Brunnermeier (2008), presented at the height of the recent 13 It is conceivable that the government can sell its capital instruments at a positive price at a later stage, and possibly also with a profit, as happened after the Norwegian banking crisis of the 1990’s, see Vale (2004). But the ex ante costs were still substantial with very few investors willing to buy shares along with the government. It can be argued that this ex ante cost is the relevant measure of the rescue costs, because they reflect the subsidy relative to market valuations at the time. In some cases the rescue costs to the government will be minimal or zero. Failures of small banks with standard banking activities can in most cases be handled by the deposit insurance scheme or some other institution that has been given resolution authorities. The solution will normally be a takeover by another bank or an orderly wind-down of the activities with customers given the time to find alternative suppliers. In the US the FDIC is performing such wind-downs on a regular basis and with all costs borne by a fund financed by the banking industry. Even relatively large banks, such as Washington Mutual with a balance sheet of approximately 300 billion USD, were resolved by the FDIC during the recent crisis. If the resolution regime can be expected to work smoothly for a bank, with no disruption in the rest of the financial system, this bank should obviously not face a systemic surcharge on its activities. Turning to larger and more complex banks in trouble, the simple solutions organised by a deposit insurance fund or a similar institution may not be applicable. The need for capital injections may be substantial and beyond the means of an existing fund, and the resolution process may be too complicated to be carried out during a weekend. In these cases the government may need to step in if an outright failure is to be avoided. We shall assume that the government does this by injecting capital. The costs of such a rescue operation will obviously depend on how undercapitalised the bank is, which is again primarily a function of accumulated deficits relative to the initial capital holdings. The size of losses mainly comes down to the size of the bank and the quality of its assets, which are the variables that has been extensively analysed in the academic research on systemic importance. Note, however, that the complexity and interconnectedness of the activities of the bank would be less relevant, given that these activities can be expected to continue as going concerns. The funding of the assets may be relevant, but only because the funding costs determines the earnings of the bank, and thus the expected degree of undercapitalisation. Experiences from previous systemic events may provide some guidance as to what future rescue costs will be, but only to a limited extent. Systemic events are rare, banks are different, and the resolution regime may have been improved. Historical data may still provide some guidance for the future. In the following we shall have a brief look at the resolution costs incurred during the Norwegian banking crisis of the early 1990s. Most of the troubled banks were handled by the deposit insurance funds. Only in a few cases did the government directly inject capital into banks. These 14 included three of the four largest banks of the country, which were all considered systemically important. In addition two medium-sized bank received government capital at an early stage of the crisis, when resolution procedures were not yet firmly in place. A third medium-sized bank received a small injection towards the end of the crisis when the insurance funds had no money left. In the first table below we only include the three systemically important banks. Table 1: Rescue costs for large banks during the Norwegian banking crisis of 1988-93. Numbers are in millions NOK or per cent of pre-crisis total assets. Bank 1987 total assets 1987 capital Accumulated losses 1988- 92 Capital injection from the government; total and in per cent of assets Den norske Bank 170 209 4.2 % 3.9 % 4 750 2.8 % Kreditkassen/Sunnmørsbanken 106 099 4.3 % 7.8 % 8 914 8.4 % Fokus Bank 35 477 4.4 % 12.8 % 1 845 5.6 % Source: Statistics Norway and Government Report 39 (1993-94) to the Storting. The three large banks that were rescued had very similar capital positions immediately before the crisis occurred, and somewhat higher than most banks of their size had prior to the recent crisis. The table illustrates that capital injections were not proportional to the size of the banks. An alternative possible determinant of the rescue costs are the losses incurred during the last years before the crisis. The disproportionately high injection into the second largest bank (Kreditkassen) is related to the fact that it had taken over another troubled bank (Sunnmørsbanken) in 1988, and had probably borne part of its rescue costs. Taking this into account we may consider the data consistent with a view that expected accumulated losses in a crisis event is a relevant determinant of rescue costs. But conclusions can evidently not be drawn from only a couple of cases. Table 2: Determinants of resolution costs during the Norwegian banking crisis 1988-93 Dependent variable: Coefficients t-values Resolution costs 1988-93 (in NOK) Independent variables: Total assets 1987 (TA, in NOK) -0.0048 -0.88 Bank equity 1987 (per cent of TA) -10221 -1.97 Deposits 1987 (per cent of TA) -1535 -2.11 Accumulated losses 1988-93 (in NOK) 0.8485 10.30 Intercept 1575 R-square 0.966 15 Data on the total resolution costs – i.e. for the deposit insurance funds and the government combined - for all the 23 failed banks are also available.7 We use these data together with data for total assets, bank equity and deposit coverage before the crisis and accumulated losses during the crisis to run a simple linear regression, reported in the table above. Given this model specification, with resolution costs, total assets and accumulated losses all measured in absolute values, we find that the resolution costs mainly depended on the losses accumulated during the crisis. Total assets do not come out as statistically significant when accumulated losses are taken into account, and the variable even gets a negative sign. 8 As expected, higher bank equity and higher deposit coverage contributes to reduce the resolution costs. 6. Implications for the surcharge on systemic banks If government rescue of systemically important banks carries a high probability, the expected costs of such rescue operations should be relevant for the calibration of any systemic surcharge. Below we discuss what implications the rescue option may have. It is reasonable to assume that expected losses are the main determinant of expected rescue costs. A multiplicative surcharge on the normal capital adequacy requirement would be an appropriate regulatory instrument for the systemically important banks. The recommendation naturally rests on the assumption that the normal capital requirement correctly reflects the tail risk exposure of the bank, as it is intended to do. A second factor that is possibly relevant for rescue costs is the complexity of the bank. This could determine whether splitting up the bank into systemically important and less important parts will be feasible. The multiplier on capital adequacy could thus depend on some measure of organisational complexity. That is one of the factors considered by the regulatory agencies, but also one that is hard to measure. A better alternative than requiring higher capital for complexity may thus be to impose direct restrictions on the organizational structure of banks. This could be part of the “living will” process. The expected costs conditional on a rescue operation taking place is naturally only one component of the equation. Another essential component is the bank’s probability of failure. This probability has been extensively studied in the research literature, with models for the probability of failure for individual banks generally estimated on accounting data; see e.g. Pettway and Sinkey (1980) for an early example. These early warning models are partial in the sense that they do not take into account the contagion effect from the rest of the financial system. 7 Government Report 39 (1993-94) to Storting (The Norwegian Parliament). 8 Bank size is naturally highly correlated with resolution costs, with a correlation coefficient of 0.85. However, the correlation with accumulated losses before resolution became necessary is even higher, with a correlation coefficient of 0.98. 16 However, one can argue that contagion would be a less relevant issue in a setting where government rescue in some form is the normal solution for a troubled bank. The probability of failure depends in most empirically calibrated models on the capital position, on the earnings and operating costs, on the asset quality, and on the liquidity position of the bank. Most of these variables are the same as those determining the expected cost of a rescue operation. The main additional variable is the liquidity position. The relevance of this variable has been evident during the recent crisis, with spectacular failures occurring in apparently well capitalised banks. But these banks had substantial maturity mismatches between long term illiquid assets and very short funding maturities. Crises were often triggered by banks being cut off from their normal funding sources. Banks that are not individually systemic may still be systemic as part of a herd, as pointed out by for instance Brunnermeier et al (2009). This will indeed be the norm when a systemic crisis erupts: Common exposures will have been built up across a large number of financial institutions. The archetypal example is property lending fuelling a property price bubble, where a major part of the social costs are borne by third parties and not by bank stakeholders, and where rescue costs may not represent the major part of social costs not borne by market participants. Acharya (2009) suggested that exposures to systemic risk factors should be a determinant of a bank differentiated systemic surcharge. But a capital surcharge may not be the best instrument in this case. This threat to financial stability could instead be handled by instruments targeting the exposures directly. In a property price bubble this would mean restricting lending to this particular market. Examples of possible instruments include restrictions on loans to value, imposing higher risk weights on mortgage loans, and making funding of the loan growth more expensive and thus increase loan costs. This should be done across the board with no special treatment of systemically important institutions, because they contribute to the price bubble only to the extent that they lend to or invest in this market. 7. Conclusions The communiqué from the G20 Pittsburgh summit in September 2009 stated that regulatory “standards for large global banks should be commensurate with the cost of their failure”. The G20 also stressed that “systemically important banks should develop internationally-consistent bank-specific contingency and resolution plans”. This reflects the emerging consensus that the proper regulation of systemically important banks is essential for reducing the probability of major financial crises in the future. Proposals from national and international organizations working on financial regulation list a number of characteristics of systemically important banks, with only ad hoc aggregation to one single measure of systemic importance. The implication is that the classification of banks 17 according to importance must be subject to supervisory judgement. The academic literature, however, has tried to be more precise, and a number of measurement proposals have been put forward. The measures are useful to the extent that they help identify and graduate the systemic importance of banks. One very important application of the measures would be the calibration of systemic surcharges. For this purpose the measures have a number of caveats. The optimal surcharge should evidently depend on the trade-off between the social costs of regulation in non-crisis periods on the one hand and the social costs of a financial crisis on the other hand. The social costs of a crisis will depend on what instruments are available to handle the crisis, and on whether the government chooses to use these instruments to intervene. The measures proposed implicitly assume that outright failures will be permitted. The final caveat is that unbiased estimates of the social costs of bank failures are unlikely to be retrieved from market data. In practical life there will be a significantly positive probability that distressed systemically important banks will be rescued by the government. This implies that the social costs of rescuing or resolving a systematically important bank should be a key determinant of any systemic surcharge. These costs will depend on the resolution regime as well as on bank characteristics. An improved resolution regime can make some banks easier to resolve and thus less systemically important, while the improvements may not be applicable to others. The proposed measures of systemic importance never try to take this into account. Given the complexity of large banks and the present resolution regimes available to the regulator, rescue will in many cases mean rescue of the entire bank if some parts are deemed to be systemically important. With improved resolution regimes the splitting up of banks may become more tractable. This may allow the government to retain only the truly systemically important activities as going concerns, and to spin off less important activities. One precondition for this may be stricter supervisory rules for the organizational structure of banks. We found that the characteristics of systemic importance listed by regulatory agencies include most of the characteristics that are likely to explain rescue costs. The list from the academic literature is similar, but with more focus on the asset side of banks’ balance sheets. Our analysis deviates from this literature by assigning a significantly positive probability to government rescue of systemically important banks. We are thus placing less importance on the contagion and interconnectedness indicators of systemic importance, since these characteristics are less relevant when rescue is the expected resolution procedure. Complexity could retain its role as an important determinant of a systemic surcharge, to the extent that it is not eliminated by the requirements for a feasible recovery and resolution plan. Part of the academic literature has stressed the risk exposures of banks to common risk factors as a key indicator of systemic importance and thus as an important determinant of a systemic surcharge. We have instead argued that such common exposure should be contained by more 18 direct regulatory measures. A prime example is lending to the housing and property sector, which could be counteracted by using macroprudential tools such as higher risk weightings on such loans or loan-to-value restrictions. The empirical analysis of this paper is illustrative, but provides some indication that accumulated losses to a large extent determines the costs of government rescue operations. In our framework this implies that tail probability losses, say in a set of relevant stress scenarios, should be an important determinant of any systemic capital surcharge. This is what the standard capital requirement is meant to cover. A systemic surcharge could then take the form of a multiplier on the normal capital requirement, perhaps in the buffer component rather than in the minimum capital requirement. The analysis in this paper is meant to be complementary to the mainstream academic discussion of a systemic surcharge on important banks. We do not argue that rescue should always be the solution to a crisis at an important bank, or that rescue should be a promise to banks paying a high systemic surcharge. We acknowledge that more focus on rescue costs may strengthen market expectations that rescue will be the normal solution, and thus counteract the very purpose of systemic surcharges. But market expectations are already very strong and experience suggests that this option is often used. Then the expected rescue costs must be relevant for the calibration of a systemic surcharge on the activities of systemically important banks. The work now being done on improved resolution regimes and bank-specific resolution plans will hopefully make rescue a less expensive option for the government and a less attractive option for bank stakeholders. References: Acharya, Viral V. (2009), A theory of systemic risk and design of prudential bank regulation. Journal of Financial Stability 5, 224-255. Acharya, Viral. V., Lasse H. Pedersen, Thomas Philippon and Matthew Richardson (2009), Regulating systemic risk. Chapter 13 in: Acharya, Viral. V, and Matthew Richardson, “Restoring Financial Stability: How to repair a failed system.” John Wiley & Sons. Acharya, Viral. V., Lasse H. Pedersen, Thomas Philippon and Matthew Richardson (2010), Measuring systemic risk. Mimeo, May 2010. Adrian, Tobias and Markus K. Brunnermeier (2008), CoVaR. Federal Reserve Bank of New York Staff Report 34 (Revised). Bank of England (2009a), The UK Special resolution regime for failing banks in an international context. Financial Stability Paper No. 5. 19 Bank of England (2009b), The role of macroprudential policy. Discussion Paper. Basel Committee for Banking Supervision (2009), Strengthening the resilience of the banking sector - consultative document. 17 December 2009. Basel Committee for Banking Supervision (2010), An assessment of the long-term economic impact of stronger capital and liquidity requirements. 18 August 2010. Brunnermeier, M.T., A. Crockett, C. Goodhart, A.D. Persaud and H. Shin (2009), The Fundamental Principles of Financial Regulation. Geneva Reports on the World Economy No. 11. Department of the (US) Treasury (2009), Financial Regulatory Reform. Report published on 17 June 2009. Elsinger, Helmut, Alfred Lehar and Martin Summer, Risk Assessment for Banking Systems. Management Science September 2006, 52(9), pp 1301-1314. Financial Services Authority (2009), A regulatory response to the global banking crisis. Discussion Paper 09/2. Financial Stability Board (2009), Guidance to Assess the Systemic Importance of Financial Institutions, Markets and Instruments: Initial Considerations. Report to G20 Finance Ministers and Governors. Financial Stability Board (2010), Reducing the moral hazard posed by systemically important financial institutions. Interim report to G20 Leaders, 18 June. Greenbaum, Stuart I. and Anjan V. Thakor (1995), Contemporary Financial Intermediation. The Dryden Press. Haldane, Andrew G (2010), The $100 billion question. Speech 30 March 2010 published on the Bank of England website. Huang, Xin, Hao Zhou and Haibin Zhu (2010), Assessing the systemic risk of a heterogeneous portfolio of banks during the recent financial crisis. BIS Working Paper No 296. Huang, Xin, Hao Zhou and Haibin Zhu (2009), A framework for assessing the systemic risk of major financial institutions. Journal of Banking and Finance, 33(11), 2036-2049. Huertas, Thomas (2010), Resolution and Contagion. Speech 26 February 2010 published on the Financial Supervision Agency website. International Monetary Fund (2010), Global Financial Stability Report 2010/1. Iragorri, C.C. and S. Ferrari (2010), Measuring the systemic importance of financial institutions using market information. Financial Stability Review 2010, National Bank of Belgium. 20 Pettway, R.H. and J.F. Sinkey (1980), Establishing on-site examination priorities: An early warning system using accounting and market information. Journal of Finance 35(1), 137-150. Reinhart, C.M. and K.S. Rogoff (2009), This time is different. Eight centuries of financial folly. Princeton University Press. Saunders, A., R.C. Smith and I. Walter (2009), Enhanced Regulation of Large, Complex Financial Institutions. Chapter 5 in: Acharya, V. and M. Richardson (eds.), Restoring Financial Stability: How to Repair a Failed System. John Wiley & Sons Inc. Tarashev, Nikola, Claudio Borio and Kostas Tsatsaronis (2010), Allocating systemic risk to individual institutions. BIS Working Paper No 308. Vale, Bent (2004), The Norwegian banking crisis. In Moe, Thorvald G, Jon Solheim and Bent Vale (eds.), The Norwegian Baking Crisis. Norges Bank Occasional Papers No. 33. Zhou, Chen (2009), Are banks too big to fail? DNB Working Paper No. 232.