# Basel 2 – The Three Pillar Approach - Modern Banking

In response to criticism of the 1988 Accord, a number of changes were made, culminating in the 2001 proposal. The original plan was for the proposal to be discussed among bankers and members of the Basel Committee, agreed on by January 2002, and adopted by 2004.

However, over 250 (largely negative) comments from banks, together with the Committee’s three impact studies, prompted it to make substantial changes to the original document. A final consultative document was published in April 2003, with comments invited until the end of July 2003. The new agreement was reached in May 2004, and is published in full (251 pages!) by the BIS on behalf of the Basel Committee on Banking Supervision (2004). The standardised approach will apply to the G-10 countries by the end of 2006 and the ‘‘advanced’’ approaches will take effect from the end of 2007. During the first year of implementation, banks and national regulators are expected to run parallel computations, calculating capital charges based on Basel 1 and 2.

However, US regulators have thrown a spanner in the works. In February 2003 it was announced that just 10 of the most active global US banks would adopt the advanced IRB approach (see below); another 10 or so are expected to abide by the Accord. In addition, it will apply to the largest broker dealers in the USA, according to new rules recently proposed by the Securities and Exchange Commission. The rest of the American banks will continue to use Basel 1. The matter is discussed later, but the decision does undermine the potential impact Basel 2 will have.

It is also worth emphasising that though the Basel agreements apply to the international banks in member countries, many countries require all their banks to adhere to the Basel rules. For example, part of the European Union’s Capital Adequacy Directive II (CAD-II) requires all EU credit institutions to adopt the Basel 1 standards. Basel 2 will be part of the new Capital Requirements Directive, which must be passed by the EU Parliament. Elections were held in June 2004, and included the 10 new countries for the first time. The first opportunity for the Directive to be put to the EU Parliament is in late 2004/5. Once passed by the EU, each of the 25 member states incorporate the Directive into their respective laws. For example, it becomes part of UK law once it is ratified by parliament. Finally, since the membership of the Bank for International Settlements (and the Basel Core Principles Liaison Committee) has expanded to include key developing countries, and regulators from these countries often require their banks to adopt Basel 1/2, many countries with no direct representation on the Basel Committee aspire to treat the Basel rules as a benchmark for their banks.

The new Accord seeks to achieve the following objectives.

1. It moves away from the ‘‘one size fits all’’ approach characteristic of Basel 1. The emphasis is on ‘‘mix and match’’. That is, each bank can choose from a number of options to determine its capital charge for market, credit and operational risk. Table illustrates the choices open to banks.
2. Recognition that in terms of credit risk, lending to banks or corporates can be more or less risky than to OECD sovereigns. As a result, risk weightings have been changed to such an extent that a bank or corporation can receive a lower risk weight than the country where it is headquartered.
3. Basel 2’s Risk Pillar 1: Summary of Approaches

4. Explicit recognition of operational risk, with capital to be set aside, though overall the amount of capital set aside should remain at 8% of total risk assets.
5. Subject to the approval of national regulators, banks will be allowed to use their own internal rating models for the measurement of credit, market and operational risk. Otherwise, banks will have to adopt a standardised approach drawn up by the Basel Committee.
6. In addition to the new ‘‘risk pillar’’, new ‘‘supervisory’’ and ‘‘market discipline’’ pillars have been introduced.

The definition of tier 1 and 2 capital used in Basel 1 is retained. However, over the longer term the Basel Committee plans to review what constitutes eligible tier 1 capital. The minimum requirements to set aside, 4% of capital (for tier 1) or 8% (for tier 2), remains unchanged.

The market risk measure introduced in the 1996 amendment is part of the new risk assets ratio, and was discussed at length earlier. The only change proposed is to have one system for determining the trigger charge. There are important changes in the measurement of credit risk and, for the first time, an attempt to measure and impose a capital requirement for operational risk.

Pillar 1 – Credit Risk Measures

Measures of credit risk have been changed to deal with some of the criticisms of Basel 1. Banks must adopt one of three measurements: Standardised (modification of existing approach); ‘‘Foundation Internal Ratings Based Approach’’ and ‘‘Advanced Internal Ratings Based Approach’’.

The standardised approach to credit risk

Banks lacking sophisticated models for assessing risk will be required to adopt the standardized approach under Basel 2. Even with the standardised approach, the Basel Committee has recognised the need for more flexible treatment with respect to credit risk. The major modification involves the use of a wider band of risk weightings, from 0% for very low risk to 150% for high risk loans. The credit risk weights for loans to countries, banks, corporate and securitised assets are summarised in Table below. There is no longer a distinction between OECD and other sovereigns – a sovereign risk weighting will be determined by external rating agencies or a qualified export credit agency.

Weightings for other assets are as follows.

1. Residential mortgages: 35%
2. Commercial mortgages: 100%
3. Personal loans: 100%
4. Venture capital: the decision of the national regulator
5. All other assets: 100%26

The above changes mean, for example, that if a corporation is rated BB–by a rating agency, the bank may be asked to assign a risk weight of 150% to that asset, so it will have to

Credit Risk Weights Under the Standardised Approach

* For sovereigns, supervisors may opt for the credit scores produced by qualified Export Credit Agencies, in which case the risk weight (in brackets) for ECA risk scores is as follows: risk score = 1 (0%), 2 (20%), 3 (50%), 4–6 (100%), 7 (150%).

setaside 12% of the value of the loan as capital. On the other hand, firms with treble A ratings carry a risk weight of 20%, meaning the bank need set aside only 1.6% of the value of the loan. The changes give banks an incentive to loan to more highly rated corporations, whereas under Basel 1, the amount of capital to be set aside was always the same, independent of corporations’ risk profiles. Small and medium-sized enterprises are generally not rated at all. The German government, among others, expressed concern at higher capital charges imposed on the mittelstand. However, in the final version, the approach taken has satisfied the critics, and the consensus is that Basel 2 is unlikely to result in a reduction in the availability of finance for small and medium-sized enterprises (SMEs) (see below).

Some on-balance sheet netting is to be allowed on the banking book provided it meets specified standards. However, there is no provision that would encourage banks to spread their risks across a group of loans.

Credit risk: foundation and advanced internal ratings based approach

Subject to the approval of the national supervisor, these banks may use their own internal ratings and credit information to determine how much capital is to be set aside for credit

risk. Basel has introduced these options to reward banks with sophisticated risk weighting systems, which should lower the capital to be set aside to cover credit risk. It also increases the likelihood that ratings will be based on economic capital, the capital set aside to cover unexpected losses. This is considered an improvement over regulatory capital, which is set aside based on regulatory dictates such as the Basel 1 or 2 risk weightings. The difference between the foundation and advanced IRB relates to the data supplied by a bank, and the data provided by the supervisor.

A bank must satisfy some minimum requirements to be approved for use of the internal ratings approach (IRB). The conditions include:

• Differentiation of credit risk.
• Clear criteria for the internal ratings system and a complete ratings assignment.
• The probability of default (PD) is estimated for each group of borrowers assigned to internal grades.
• PD data: banks must have at least 2 years of data from the time Basel 2 takes effect, rising to 5 within 3 years.
• LGD: advanced IRB banks are to supply 7 years of data for loss given default (LGD), though they are encouraged to develop a database covering a complete economic cycle.
Banks can use data from other sources, such as rating agencies or other banks.
• Internal validation.
• Risk components: the bank uses its own. Supervisors are to approve the method by which the risk components are converted into risk weights for the computation of risk weighted assets.
• A bank’s internal ratings and VaR must be part of an integrated risk management system.
• For example, while VaR is used to assess market risk and the regulatory capital to be set aside, the risk management system must determine the economic capital (used to set limits), look at performance via a risk adjusted return on capital (RAROC), etc.

• Satisfy the disclosure standards specified under pillar 3.

Risk weights under foundation IRB

Table below applies for all corporate, sovereign and interbank exposures. Once the supervisory authorities approve a bank’s use of the foundation IRB approach, there is the question of how the risk weights will be applied. Basel assigns two risk weights. The first risk weight is a function of PD, which is supplied by the bank; the second a function of LGD. The values for LGD, along with EAD, are supplied by Basel, and will depend on the nature of the exposure.

Basel had intended to include expected losses in the risk weightings but the final agreement (June 2004) replaced this with a requirement that if a bank finds the actual provisions it set aside is less than expected losses, it must be deducted from tier 1 and tier 2 capital, subject to a maximum cap.

For retail exposures, no distinction is drawn between IRB and advanced IRB. All IRB (foundation and advanced) banks are expected to supply internal estimates of PD, LGD and EAD based on pools of exposures. Retail loans are divided into three categories:

1. residential mortgages;
2. revolving retail loans – mainly unsecured revolving credits, such as that incurred by agents who roll over most of their credit card payments; and
3. other retail – non-mortgage consumer lending including loans to SMEs with annual sales of less than ¤5 million. Basel provides the risk weight formula to obtain risk weighted assets in each of the three categories. The risk weight is obtained using a Basel specified risk correlation, and formulae using PD, LGD and EAD.29

The loan loss rates on different types of loans are used to obtain estimates of the loss given default, LGD. Once LGD is known, together with PD, a risk weight is derived. The risk weight for retail exposures is assumed to be about 50% of corporate exposures, based on the reasoning that personal loan portfolios are more highly diversified.

In the original proposals, loans to small and medium-sized enterprises (SMEs – defined as firms with annual sales of <¤50 million) were to be treated like retail loans, but in the final document,30 the IRB risk weight formula for corporates is to be used, adjusted for firm size. The corporate risk weight is adjusted using the formula: 0.04 × 1[(S − 5)/45], where S is the annual sales in ¤ millions. If ¤50 ≥ S ≥ ¤5 million, then the formula is used. ¤5 million is a floor: anything less is treated as ¤5 million, or the firm can opt to have the loan treated as a retail loan. SMEs are treated as retail loans if their total exposure to the banking group is less than ¤1 million–the bank in question treats these loans the same way as other retail exposures.

Securitisation

For IRB banks originating securitisations, a bank must calculate KIRB, which is the amount of capital that would have been set aside if the underlying pool of assets had not been securitised. If the bank is in a first loss position (i.e. in the event of a default on the securitised assets it has to absorb the losses that are a fraction of (or equal to) KIRB), then the position must be deducted from capital. In other words, banks that do not pass on the full credit risk to a third party will have to set aside capital. The amount set aside is determined by a ratings based approach if the security is externally rated. If IRB banks invest in securitisations, a formula is used to estimate how much capital is to be deducted based on the external rating given, or, if they are unrated, other factors. However, in the June 2004 agreement, it was acknowledged that some aspects of the treatment of securitization was under review.

Credit risk mitigation: collateral, guarantees and credit derivatives

Basel recognises collateral, guarantees and credit derivatives as ‘‘credit risk mitigants’’, because the presence of any three may mean that in the event of default, some assets are recovered, which reduces the size of a loss for the bank. However, certain restrictions apply, depending on the risk management approach adopted by a bank.

Collateral

Collateral backs a loan, and in the event of default, is used to recover some assets, Thus, collateral affects LGD – the higher the quality and amount of collateral, the smaller the LGD. Under Basel 2, what is accepted as recognised collateral depends on the approach adopted by the bank.

• Standardised approach: The main components of recognised financial collateral include cash (held on deposit at the bank granting the loan), gold, government securities rated BB− and above or at least BBB (when issued by non-government entities, including banks and securities firms); unrated securities if they are issued by a bank, are traded on a main exchange and qualify as senior debt, equities (or mutuals /UCITS) that are part of a main index (e.g. the FTSE 100).
• IRB: the main components are all collateral under the standard approach, equities (or mutuals/UCITS) traded on a main index, receivables, and some types of commercial/ residential and property.
• Advanced IRB: all forms of physical collateral are accepted, in addition to the collateral listed under IRB.

Guarantees

A guarantee is provided through a backer. For example, another bank can guarantee a loan.

The key risk is the quality of the guarantor. Thus, a guarantee, depending on its quality, will affect the probability of loan default (PD). Ischenko and Samuels (2001) show that for a given expected loss, the risk weight on LGD will be lower than that on PD. It means banks are likely to opt for lending with collateral rather than guarantees, because the risk weight will be lower.

Credit derivatives

Though excluded as a possible credit risk mitigant in the earlier consultative documents, in the third paper (BIS, 2003c), Basel accepted that credit derivatives, in the form of credit default swaps (CDSs), can give a form of insurance against loss. The main issue surrounds what constitutes a credit event, i.e. what constitutes default, and in particular, what types of restructuring constitute default. Basel’s current position is that banks can use them to lower capital requirements provided the credit default swap includes restructuring as a form of default event if it results in credit losses, unless the bank has control over the decision to restructure.

Advanced internal ratings based approach and credit risk

If a bank’s credit risk management system is approved for the advanced internal ratings based approach (AIRB), the bank supplies its own estimates for PD, LGD, EAD and maturity. There are no rules on what factors should be used for the purposes of risk mitigation. Furthermore, all physical collateral is recognised, unlike the limited recognition of property and equity under IRB. Basel 2 proposals reward more sophisticated risk management systems by reducing the amount of capital to be set aside. The reasoning is that their models account for economic capital sufficiently well to satisfy regulatory capital requirements. Ischenko and Samuels (2001) estimated that for some banks, adopting an AIRB will reduce capital requirements by 10–20% compared to IRB.

A more recent publication by Citigroup Smith Barney (2003) concluded there was little difference by way of capital relief if the AIRB was used in place of IRB, but AIRB is significantly more costly to introduce.

Pillar 1 – Operational Risk

Operational risk (OR) is a new controversial addition to the denominator of the risk assets ratio. Recall Basel’s definition of operational risk, which in more recent documents has changed very slightly:

Based on the most recent Basel publications at the time of writing, a bank may adopt one of three approaches (or a variant of the basic standardised approach) in the measurement of operational risk.

• Basic indicator: A capital charge based on a single indicator for overall risk exposure, the average (positive) annual gross income over the previous 3 years. Then the capital set aside is 15% (the alpha factor) of this, i.e. capital charge = (0.15) × (average annual gross income).
• Standardised:33 To qualify for the use of this approach, banks must have in place an operational risk system, which complies with minimum criteria outlined by Basel. This approach requires banks to identify income from eight business lines. The capital charge for each business line is gross income multiplied by a fixed percentage (beta factor), which varies between 12% and 18%. The total capital to be set aside is the sum of these capital charges. The business lines and accompanying beta factors are summarized in Table below. The total capital to be set aside is a three-year average of the regulatory charges calculated for each year. Negative capital charges (arising from negative income) for a given business line can be used to offset positive capital charges from other business lines in that year. However, if the aggregate capital charge for a given year turns out to be negative, it is entered as a 0 in the numerator of equation (XIV). The total capital charge is then:
• KTSA ={Σ1–3 max [Σ(GI1–8×β1–8), 0]}/3 ---(XVII)

where
KTSA: capital charge using the standardised approach _
1–3 : sum over 1 to 3 years
GI1–8: annual gross income in a given year for each business line
β1–8: fixed percentage of the level of gross income for each business line, given in Table below.

• Advanced measurement approaches (AMAs): AMAs are for banks meeting more advanced supervisory standards. Banks use their own methods to assess their exposure to operational risk, and from this, determine the amount of capital to be set aside. Banks are allowed to purchase insurance against operational risk, and use it to reduce the OR capital charge by up to 20%. However, to use insurance, banks must meet certain conditions. The most important is that the insurer is A-rated (by external agencies) in terms of its ability to meet claims. In addition, the insurance coverage must last at least a year, be explicit in terms of the OR it is covering, and may not have any exclusions or limitations arising from regulatory action.
• Operational Risk – Standardised Approach

For banks with global operations and numerous subsidiaries, the final agreement notes that a ‘‘hybrid approach’’ to operational risk may be used. Subject to the approval of a national supervisor, a parent bank with international operations, when employing AMAs for calculating capital to be set aside, can allow for diversification gains within its own operation but is not allowed to include group-wide benefits. Significant subsidiaries can use the head office model, parameters, etc. to compute their operational risk but the amount of capital set aside must be based on the same criteria as those used by the parent bank. Subsidiaries deemed of minor significance to the group’s operations can (subject to agreement by the supervisor) be allocated a charge for OR from the group-wide calculation, or use the parent’s methodology to compute the charge.

Pillar 2 – Responsibilities of National Supervisors

This pillar identifies the role of the national supervisors under Basel 2. Basel has identified four principles of supervisory review:

1. Supervisors are expected to ensure banks use appropriate methodology to determine Basel 2 ratios, and have a strategy to maintain capital requirements.
2. Supervisors should review banks’ internal assessment procedures and strategies, taking appropriate action if these fall below standard.
3. Banks should be encouraged by supervisors to hold capital above the minimum requirement.
4. Supervisors are expected to intervene as early as possible to ask a bank to restore its capital levels if they fall below the minimum.

To fulfil these objectives, an ongoing dialogue between supervisors and banks is necessary. Also, supervisors are likely to focus on banks with a history of taking higher than average risks.

Pillar 2 does not give explicit detail on how supervisors should behave, and is likely to be used to back up pillar 1, and possibly, deal with some of the more controversial aspects of pillar 1. For example, the Committee has recently emphasised the importance of conservative stress testing for banks adopting the IRB approach. Supervisors should require these banks to devise a conservative stress test in order to test how their capital requirements might increase given a particular scenario. Based on the test results, banks should ensure they have a sufficiently robust capital buffer. If capital falls below the necessary amount, supervisors would intervene and require the bank to reduce its credit and/or market risk exposures until it can cover the capital requirements implied by the relevant stress test.

Pillar 3 – Market Discipline

The main purpose of pillar 3 is to reinforce pillars 1 and 2. Providing timely and transparent information, or even knowing they have to provide it, gives the market a role in disciplining banks. Participating banks are expected to disclose:

• Risk exposure.
• Methods for computing capital requirements.
• All material information, that is, information which, if omitted or mis-stated, could affect the decision-making of the agent using the information.
• Disclosure should take place on a semi-annual basis; quarterly in the case of risk exposure, especially if the bank engages in global activities.

The Committee plans to issue templates banks can use to ensure the disclosure principles are adhered to. It considers pillar 3 an important component of Basel 2, especially for banks using the IRB approaches in credit risk, AMA for operational risk and their own internal models for market risk. These banks have far greater discretion in terms of computation of capital charges they incur, and it will be difficult for supervisors to master every detail of the approach they take. Market discipline should discourage attempts by banks to cut corners in their risk assessment.

A Critique of Basel 2

There were numerous criticisms of Basel 2, but some were addressed during the consultative process (e.g. SMEs). The problems with the use of VaR were discussed earlier. Here, the more general problems related to the Basel 2 framework are reviewed. Perhaps the most serious is that it moves with the economic cycle, i.e. it is pro-cyclical. To the extent that the creditworthiness of financial and non-financial firms moves with the cycle, the method for calculating the amount of capital to be set aside in a given year means less will be needed during an economic boom; more during a downturn. The nature of recession (falling stock markets, downgrading of firms experiencing falling profits by independent rating agencies, and higher loan losses as a result of increased default rates) will reduce banks’ risk assets ratios. Since raising capital, even if possible, will be more costly, banks are likely to cut back on their activities (e.g. reduced lending, less trading), which in turn will aggravate the downturn.

Hawke (2001) gives an interesting example of the effect of pro-cyclicality. When Basel 1 was being implemented in the late 1980s/early 1990s, the US banking system was in the throes of a crisis. Banks were facing mounting losses – even the Deposit Insurance Corporation was threatened with insolvency. Many US bank supervisors thought Basel 1 aggravated the crisis as banks struggled to get their Basel risk assets ratios up to 8%, either by reducing lending and/or trying to raise new capital in a depressed market. The Basel Committee addressed this criticism in several ways. Compared to earlier proposals, the risk curve, or the relationship between capital charges and the probability of default, has been flattened for corporate and retail loans. Also, banks have been asked to take a long run view (rather than just one year) when they determine the internal ratings of borrowers. This means the ratings should reflect conditions over a number of years, taking the whole business cycle into account. If banks are estimating their probability of default (which in turn feeds into the capital to be deducted), they are advised to use the full economic cycle. When making loan decisions, banks should note the stage of the economic cycle and employ stress tests to identify economic changes that will affect their portfolio. The information can be fed into the determination of their capital requirements. However, it is often difficult to assess how long a stage of the cycle will last. There is also a more general challenge: to collect sufficient data, especially in the early years.

Arecent study suggests that the external ratings of the creditworthiness of firms could also fuel the problem of pro-cyclicality. Amato and Furfine (2003) reported that it is rare for the rating of a large corporation or bank to change. This finding is consistent with the general claim that credit ratings are not related to the cycle because they are relative measures. A bond rated AAA signals that it is less risky than a bond rated BB. Nonetheless, it has been shown that ratings move with the business cycle,36 though this alone does not necessarily mean the ratings themselves are influenced by the cycle. This is the question Amato and Furfine set out to address, using data on the economic cycle, financial ratios and the ratings themselves. The ratings data include both investment and speculative grade; from Standard and Poor’s monthly ratings of all firms – January 1981 to December 2001. Amata and Furfine report that for small changes in business risk, ratings remain unchanged. However, they find evidence of ‘‘overshooting’’ when a rating is changed. Upgradings were found to be excessive; downgradings too severe. Furthermore, the excessive optimism/pessimism is directly correlated with the state of the macroeconomy, meaning the upgrade/downgrade will aggravate a boom/recession.

Perversely, Basel 2 could raise the amount of systemic risk for banks using the standardized approach. They have little incentive to diversify because they are not rewarded for it, though this was also true in the case of Basel 1.

Recall the original purpose of the Basel 1 accord was to establish a level playing field for international banks in terms of regulatory capital to be set aside. Banks can pick and choose from different parts of Basel 2, which means all banks have an equal opportunity to determine the amount of regulatory capital to be set aside. However, the complex details and/or proportionately higher compliance costs for some banks means the playing field is no longer level.

As was noted earlier, Basel 2 will be used by 10 to 20 of the most internationally active US banks, but the rest of the American banks will use Basel 1. This has important competitive implications. The US banks which do adopt Basel 2 are the ones with sophisticated in-house models, so they will employ advanced approaches to the treatment of credit, market and operational risks, i.e. internal ratings for market risk, advanced IRB for credit risk and AMA for operational risk. Therefore it is likely their overall capital requirements will fall. Furthermore, there are no onerous new compliance costs for the thousands of US banks which continue to employ Basel 1, which may give them a cost advantage if the capital charge based on Basel 1 is lower. This gives US banks a competitive edge over their European or Japanese counterparts. On the other hand, banks adhering to Basel 1 will not experience a reduction in the capital they must set aside, while banks in other countries may. Also, the US sets quite rigorous regulatory standards, which may offset any cost advantage they achieve because they do not adopt Basel 2.

The big European banks which see the major US banks as their main competitors in wholesale markets will have their competitive position further undermined, for two reasons.

First, it was noted earlier that Basel 2 is to be part of the Capital Adequacy Directive III before it is implemented in Europe. According to Milne (2003), contrary to expectations, the fast track Lamfalussy option37 will not be used for the CAD III, which means that most of Basel 2’s technical details will have to be passed by the European parliament, a process that will take, at the minimum, three to four years. US banks which adopt Basel 2 will do so immediately after their regulators approve its use. Their capital requirements are likely to be lower, while the European competitors will have to set aside larger amounts of capital under the old Basel 1 accord. This competitive edge for the top US banks will continue until the Capital Adequacy Directive III is passed. Second, once Basel 2 is part of a European directive, any component of it that dates or is affected by financial innovation will be extremely difficult to update/amend because it is part of a European law.

The problems outlined above will hit London’s financial district particularly hard, and could undermine its leading international position in financial markets. The UK’s Financial Services Authority may be forced to take unilateral action, and require banks in London to implement Basel 2 ahead of the EU’s CAD III.

Some commentators have suggested that there is a danger of banks that are part of financial conglomerates moving their credit risk to another non-bank financial subsidiary to reduce the amount of capital they have to set aside. For example, credit derivatives might transfer the credit risk related to a loan to an insurance company. Or assets could be securitised and sold to third party insurers. However, the final version of Basel 2 (BIS, 2003c; Basel Committee, 2004) has tightened up many loopholes and should prevent some aspects of regulatory arbitrage that occurred under Basel 1. Also, such behaviour is unlikely to be ignored by national regulators: this is an example where pillar 2 could re-enforce pillar1.

A related concern is that the Basel requirements are encouraging banks to transfer credit risk off their balance sheets. The credit derivatives market grew from virtually nothing in the early 1990s to $2 trillion by 2002. These are forms of credit risk transfer: banks originate the loan (agree to lend money to firms and individuals) but transfer the risk from the bank to purchasers of loans or securities. The trend to move loans off-balance sheet began with the issue of mortgage backed securities in the 1970s, followed by, in the 1980s, the sale of sovereign debt, syndicated loans and corporate debt. However, now it is credit risk which is being transferred. Most of the institutional investors assuming this credit risk (as a consequence of securitisation or the use of credit derivatives) do not have in-house credit risk departments and rely on credit rating agencies. The agencies have expertise in assessing personal, firm or country risks, but do not look at the aggregate picture, even though institutional investors typically purchase, or insurance is written for, bundles of loans or bonds. Banks no longer hold risk but are conduits of risks. On the other hand, only a few of the top global banks are active in this market. Recall BIS (2003e) reported that 17 (19) US banks sold (bought) credit protection and only 391 out of 2220 banks supervised by the Office of the Comptroller of Currency held any form of credit derivatives. Risk Magazine reported 13 firms were behind 80% of transactions in credit derivatives. Finally, The Economist claimed roughly 8% of US commercial and industrial loans were insured ($60 billion).40 All of these figures indicate responsibility for the majority of the credit risk associated with lending remains in the banking sector. The emphasis on the use of external ratings raises other issues. To reduce capital requirements, banks using the standardised approach will want to lend to rated firms. Most rated corporations are headquartered in the USA, and to the extent that corporations do business with their own national banks, it gives US banks an additional competitive advantage, at least in the short run. Another problem is the absence of a strong ratings culture in Europe and Japan. For example, Moody’s rates 554 corporates in Europe; 221 of these are in the UK, another 121 in the Netherlands. In France and Germany, the numbers are as low as 43 and 45; respectively. That leaves just 127 other firms spread throughout Europe. In Japan, just 191 corporates are rated.41 However, given the importance Basel will place on rating agencies, it is likely their business will spread rapidly in Japan and Europe.

Regulators will have to identify the most accurate, requiring them to meet a set of criteria to be accepted as a recognised agency.

Small and medium-sized enterprises, and firms located in emerging markets, may find it more difficult to raise external finance because they are not rated. To address this issue, the final document (2004) confirmed the use of an adjusted formula based on the IRB corporate risk weight for SMEs with sales revenues ranging from ¤5 to ¤50 million. Otherwise, if SMEs are classified as retail, they could benefit from the flatter risk curve noted earlier. However, there is no allowance for portfolio diversification through SME exposure.

In the USA, only four agencies (Standard and Poor’s, Moody’s, Fitch IBCA and Dominion Bond Ratings) are officially recognised by the Securities and Exchange Commission (SEC), giving them effective control over the US market. This raises the issue of monopoly power in the ratings sector. A US congressional subcommittee has asked the SEC about its relationship with these agencies. The subcommittee has expressed concern that the arrangement could limit the operation of a free market and prevent consumer interests from being served. Just three of these rating agencies are global players, meaning they are exposed to even less competition outside the USA.

There is also a potential for conflict of interest because increasingly, ratings firms advise banks on their risk management systems. The ratings agency may be tempted to give higher ratings to banks acting on their advice, though this is unlikely provided there are effective firewalls between the ratings agency and its offshoot offering the advice. However, it could increase the number of banks using similar risk management techniques. The degree to which they are correlated will mean banks react in similar ways to changes in the financial markets/macroeconomy, thereby aggravating any boom or recession.

Excessive prescription is another problem. The final agreement (2004) is 251 pages, with detailed instructions given for the implementation of Basel 2, especially the new risk assets ratio. The detailed computations needed if banks adopt either of the IRB approaches could discourage financial innovation and expansion into new markets because of the paucity of historical data necessary to compute PD and LGD. Also there are many recent examples where national regulators have encouraged healthy banks to merge with problem banks to avert a failure. Under Basel 2, any bank with IRB status will be reluctant to agree to such a merger if it means their IRB status is removed for several years because it will take that long to improve the risk management system of the weak acquisition. Thus, regulators could lose a useful tool in the resolution of banking problems, which could increase systemic risk. Milne (2003) identifies another problem arising from too many rules. He argues that regulators may find it difficult to oversee the actions of banks that opt for the advanced approaches and compute their capital obligations in Basel 2. For example, if using the IRB approach they will compute PD and LGD using sophisticated models and a considerable amount of judgement. Independent analysts or supervisors may find it difficult to assess the quality of the risk management input at this level of sophistication, and it will pose a considerable challenge to their resources. There are ways of dealing with this problem, such as requiring external auditors to verify the quality of the capital adequacy requirements as assessed by the banks, or to have supervisors monitor the work of other national supervisors. However, these are costly options. Another possibility is to tighten up disclosure requirements so that banks (after some lapse in time to preserve confidentiality) had to disclose the detailed computations of PD and LGD. But by this time, it might be too late.

The treatment of operational risk (e.g. capital to be set aside based on gross income) is considered unworkable, and OR itself is difficult to quantify. These views are shared by academics and practitioners alike. The Americans rejected Basel 2 for most of its banks because, they argue, it is too costly for them to switch. Also, their regulators believe it is impossible to quantify operational risk, making the resulting capital charge inherently subjective. They argue operational risk should be part of pillar 2 – monitored by regulators, with no explicit charge. European officials want all banks to be able to use insurance on operational risk to reduce the OR portion of the capital charge, independent of the approach they adopt.

Ischenko and Samuels (2001) claim the Basel Committee’s remarks indicate they are focusing on two risks. Rogue trader risk: such as Barings (1995) and Allied Irish Bank (2002).

If banks were required to set aside explicit capital for this type of risk, it would give them a greater incentive to monitor their positions. IT risk: relates to the concern on the reliance of computer systems to complete large numbers of banking transactions. However, there have been no real disasters arising from computer failure, though liquidity has been strained in certain cases; ‘‘9/11’’ is a good example. Back-up systems meant, relative to the scale of the disaster, there was no serious disruption and minimal loss of data.

Ischenko and Samuels (2001) estimated that for some banks, adopting the Advanced IRB will reduce capital requirements by 10–20% compared to IRB. Citigroup Smith Barney (2003) concluded there was little difference by way of capital relief if the AIRB was used in place of IRB, but AIRB is significantly more costly to introduce. More generally, Ischenko and Samuels (2001) argue the banks primarily engaged in investment banking, asset management, proprietary trading, custody and clearing will be the most adversely affected by capital charges for operational risk (OR), because of the emphasis placed on setting aside capital for rogue trading or the collapse of a bank’s IT system. For the more traditional bank with proportionately large amounts of credit related business, the OR charge will be small and the capital savings made from the new proposals for credit risk (especially if the bank adopts an advanced internal ratings approach) could be substantial. Three quantitative impact studies were conducted by the Basel team. The results of the first two indicated higher capital charges (compared to Basel 1) in the majority of cases, and in response to these findings, the proposals were revised. The final quantitative study was

Percentage Changein Capital Requirements

initiated in October 2002, and the results (BIS, 2003b) published in May 2003. 188 banks from 13 G-10 countries, and 177 banks from 30 other countries took part in the third study. Banks were divided into group 1 (globally active, diversified, large banks, with tier 1 capital in excess of ¤3 billion) and group 2 (smaller, more specialised). Table summarises the key findings. After the revisions implemented from the consultation documents, the findings were much more positive, especially if banks adopt one of the two advanced procedures which rely on their own internal ratings system. Table shows that the average capital reduction is between 2% and 6% for globally active banks.

Compared to the previous two quantitative impact studies, the findings were much more positive, especially if banks adopt one of the two advanced procedures which rely on their own internal ratings system. As can be seen from the Min and Max columns, the variation is considerable. Recall the objective of Basel 2: to bring in more sophisticated systems of risk measures so that banks could set aside capital for market, credit and operational risk, but with no change to overall capital burden (compared to Basel 1) or even a reduction in it.

For banks using the standardised approach, column one shows the capital charge is higher, especially for the G-10 group 1 and the ‘‘other’’ category – countries including Australia, Hong Kong, Norway, Singapore and a large number of emerging market countries such as China, Russia, Hungary, the Czech Republic, India, Malaysia, Thailand and Turkey. While most G-10 group 1 banks are likely to have systems in place to qualify for the IRB foundation or advanced approaches,44 this is not the case for many of the banks headquartered in countries from the ‘‘other’’ category.

If the IRB foundation approach is used, the capital charge will fall for most banks, but again, it increases for the G-10 group 1 banks and the ‘‘other’’ category. The report notes that G-10 group 1 banks have, on average, less retail activity than group 2 banks – banks with large retail exposures tended to do better because the new risk weightings are lower compared to Basel 1, with the exception of past due assets. Furthermore, the standardized approach was used by most of the banks to compute operational risk figures; a few used the basic indicator approach; only one used the advanced approach.

Table above shows that for banks adopting the IRB advanced approach, average capital charges will fall. Note, however, that very few banks in the ‘‘other’’ category will have the systems in place to qualify for this approach. Indeed, there were so few banks from the IRB category that they could not be reported because of fears they could be identified.

Basel 2 could prove most onerous for the ‘‘other’’ group, many of which come from emerging markets, not necessarily because they are inherently riskier but because they do not have sophisticated risk management systems in place. The Min and Max columns show the wide variation of impact the Basel 2 framework will have. No matter what approach is used, some banks stand to gain a great deal, while others will suffer a large increase in the capital charge.

For the more sophisticated banks that experience a reduction in their capital requirements, more capital will be released. If, overall, more capital is released than set aside, ‘‘surplus’’ capital will emerge. Too much capital will increase competition, encourage consolidation (capital surplus banks will be looking for capital weak banks), and possibly greater risk taking. The latter outcome would be a bit of an irony: regulators, by creating a situation of surplus capital, end up encouraging the banks to engage in riskier activities. It appears that banks in most of the G-10 countries are quite advanced in their preparations to adopt pillar 1 of Basel 2. In a recent survey,45 over 75% of large (assets in excess of $100 billion) and medium-sized (assets ranging from$25 to \$99 billion) banks in North America, Europe and Australia are planning to be using IRB by 2007 and to have IRB Advanced by 2010. Over 60% of European banks report being at the ‘‘implementation’’ stage, compared to 12% in the USA and 27% in Asia and the emerging markets. Progress on meeting Basel 2 operational risk requirements has been slower, with less than half the North American, European and Australian banks expecting to be using the Advanced Measurement Approach (AMA) by 2007, rising to 70% by 2010. About 62% consider their preparations for pillars 2 and 3 to be ‘‘poor’’ or ‘‘average’’. For the larger banks the cost of complying with Basel 2 ranges from between ¤50 million (60%) and ¤100 million (33%). The majority of medium-sized banks (more specialised) are expecting the cost to be less than ¤50 million.