In the previous chapters, we have discussed methodologies that a bank can use to get the best possible estimate of its required economic capital. Here, we discuss regulatory capital and specifically the recommendations of the Basel Committee on Banking Supervision.
In our previous discussion, the required economic capital was interpreted as the minimum amount of capital that a bank should hold to maintain its desired debt rating. In the economic-capital framework, the amount of capital available is the net present value or market value of the assets minus the liabilities.
As in the economic-capital framework, regulators are concerned with the difference between required capital and available capital. The required regulatory capital depends on the regulator’s assessment of the bank’s risks. The available capital is the regulator’s assessment of the current net value of the bank. It is calculated from the value of the assets minus the liabilities, according to accounting principles. As we will see, the concepts of economic capital and regulatory capital are slowly converging, and by 2006 they should be almost, but not completely, identical.
In this chapter, we focus on credit risk and discuss the history of the capital accords, define the regulatory capital more closely, and discuss the implications of the New Capital Accord proposed in January 2001. We will conclude with a discussion of managing the differences between regulatory and economic capital.
The Basel Committee on Banking Supervision was established in the mid-1980s. It is a committee of national banking regulators, such as the Bank of England and the Federal Reserve Board. It has representatives from 12 industrial countries: Belgium, Canada, France, Germany, Italy, Japan, Luxembourg, the Netherlands, Sweden, Switzerland, the United Kingdom, and the United States. The committee meets in the offices of the Bank for International Settlements in Basel, Switzerland, and is therefore often referred to as the “BIS Committee,” although strictly speaking, the BIS and the Basel Committee are separate entities.
The purpose of the committee is to set common standards for banking regulations and to improve the stability of the international banking system. Its publications can be classified into 3 categories: research papers, consultative documents, and accords. They are available at www.bis.org. The consultative documents suggest new guidelines for accords and allow industry feedback before the publication of a formal accord. The accords lay out the rules to be followed by the national regulators in such matters as setting the minimum capital requirements. The accords are binding, but they give national regulators significant leeway in how they interpret and implement certain parts of the accords.
Although the committee only has 12 formal members, the guidelines have been adopted by regulators in many other countries. These countries adopt the accords because they want to ensure that they are recognized as having a banking system that meets international standards.
The most important publications by the Basel Committee have been the 1988 capital accord,1 the 1996 amendment to the accord,2 and the consultative document issued in 2001 proposing the New Capital Accord.3
The 1988 accord defined Tier I and Tier II capital and set minimum standards for the amount of capital to be held against credit risks. The 1996 accord required additional capital to be held against market risks, and for the first time, allowed banks to measure their own riskiness using internal models, namely Value-at-Risk calculators. The 2001 document suggests a much more refined measurement of credit risks based on methodologies similar to economic capital. It also proposes holding capital specifically for operating risks. The 2001 proposals do not change the definition of Tier I and Tier II available capital; the measurement of market risks is unchanged, and the measurement of credit risk for derivatives is unchanged from the 1988 Accord. No attempt is made to set capital specifically for “interest-rate risk in the banking book,” i.e., asset/liability (ALM) risks.
The 1988 accord was motivated largely by the low amount of available capital kept by Japanese banks in relation to the risks in their lending portfolios. This low ratio was believed to allow the Japanese banks to make loans at “unfairly” low rates. This was because Japanese depositors and shareholders were not demanding high returns to compensate for the increased risk.
The 1988 accord required that all banks should hold available capital equal to at least 8% of their risk-weighted assets (RWA). Let us first discuss the accord’s concept of available capital, and then the calculation of RWA.
If all banks were able to reliably measure the market value of their assets and liabilities, then the definition of capital would be simply the difference between the value of assets and liabilities. However, it is very difficult to get an accurate measure of the true value of illiquid assets such as loans. As a practical alternative, available capital is defined according to accounting measures that are commonly available in all countries.
To explain the calculation of available capital, we must first define some accounting terms. Accounting frameworks start from the face value of liabilities and assets. They then make several deductions and additions to estimate the true value.
The value of the assets is initially estimated as the sum of the face value of the assets (A), minus specific provisions (SP) for assets that have defaulted or are considered to be close to default:
Balance Sheet Assets = A – SP
In addition to the assets formally declared on the balance sheet, there may also be revaluations (RV) and undisclosed profits (UP). Revaluations mostly apply to long-term investments such as equities and real estate. The revaluation is a recognition that the market value of an asset has increased above its initial face value. The revaluation may appear as an item in the formal balance sheet statement, or it may be in a footnote “off balance sheet.” Undisclosed profits are allowed in some countries. They are retained earnings that the bank’s management prefers to treat as a hidden reserve, rather than publish as additional available assets. The estimate of the total assets can therefore be defined as follows:
Total Assets = A – SP + RV + UP
The liabilities have two parts: the actual liabilities, and accounting partitions that divide up the difference between the balance-sheet asset value and the liability value. The difference is divided into three major components: general provisions, equity, and reserves. General provisions (GP) are partitions to account for future possible loan losses, and correspond roughly to the expected loss that we calculate in the economic-capital framework. Equity (E) is the face value of the bank’s shares at the time they were sold. The remaining difference is called reserves (R):
Balance Sheet Assets − Liabilities = GP + E + R
The liabilities can be divided into hard debt (HD) and soft debt (SD):
Liabilities − SD = HD
Hard debt is the most common form. If the bank does not pay its hard debt, it is considered to have defaulted. The soft debt comes in the form of special quasidebt instruments that are halfway between debt and equity; for example, they may be allowed to pay no interest for many years if the bank is unprofitable, but pay very high interest-rates whenever shareholders receive dividends.
From these concepts, we can estimate the net book value of assets in excess of the hard debt:
Net Value = Total Assets - HD
= (Balance Sheet Assets + RV + UP) − (Liabilities − SD)
= RV + UP + SD + (Balance Sheet Assets - Liabilities)
= RV + UP + SD + GP + E + R
From these concepts, the Basel Committee derived their requirements for the minimum amount of capital to be held. The net value was divided into two parts, or Tiers. Tier I capital consists of equity and reserves. Tier II capital consists of revaluations, undeclared profits, soft debt, and general provisions:
Tier I Capital = E + R
Tier II Capital = RV + UP + SD + GP
The 1988 accord required that the sum of Tier I and Tier II capital should be equal to 8% of the risk-weighted assets (RWA), and that the Tier I capital alone must be at least 4% of RWA. A common way of expressing the strength of a bank is its capital adequacy ratio, defined as the available capital divided by the required capital:
Now let us turn our attention to the calculation of RWA.
The risk-weighted assets is the sum of the face value of all assets, weighted according to the asset’s credit risk:
In the 1988 accord, a loan to a private company or individual was considered to have a risk weighting of 100%. Other assets were given reduced weights according to their approximate credit risks as shown in Table 23-1.
The 1988 accord also specified weights for the credit risk of traded and off-balance-sheet assets. For assets such as lines of credit and forward agreements, the risk weighting corresponded to the risk of the underlying company or asset, typically 100%. For derivatives such as interest-rate swaps, the accord gave the option of counting the risk either as a fixed percentage of the notional amount, or as 100% of the current mark-to-market value plus an “add-on,” which was calculated as a percentage of the notional amount. The mark-to-market plus add-on is now the most common approach for major banks. The “add-ons” are shown in Table 23-2.
TABLE 23-1 Credit-Risk Weights Under the 1988 Accord
This simplified counting of risk in the 1988 accord was required so that it could be implemented easily and clearly by all banks. It had the great advantage of improving the capital adequacy of banks and is often used by less-sophisticated banks to calculate a form of risk-adjusted profitability, namely the return on RWA.
As the economic-capital methodology has developed over the last decade, banks have asked their regulators to use more-accurate methods in assessing the capital to be held. This has led to the introduction of the New Basel Capital Accord.
TABLE 23-2 Add-On Values to Calculate the Risk Weight for Derivatives
The suggested form of the new accord was published in January 2001 to obtain comments from the banking industry. The final accord will be implemented around 2006. The new accord retains the same concepts of RWA and Tier I and Tier II available capital, but it changes the method for calculating RWA.
The new accord has three “Pillars:” measurement of the minimum capital requirements, supervisory review, and market discipline. The measurement of capital gives many formulas to replace the simple calculations of the 1988 accord. The supervisory review pillar requires regulators to ensure that the bank has effective risk management, and requires the regulators to increase the required capital if they think that the risks are not being adequately measured. The market discipline pillar requires banks to disclose large amounts of information so that depositors and investors can decide for themselves the riskiness of the bank and require commensurately high interest-rates and return on capital. We will discuss each pillar in turn, but we concentrate on the measurement and data requirements.
The accord allows banks to calculate their required regulatory capital using one of two approaches: the standardized approach or the internal ratings-based approach.
The standardized approach is more complex than the 1988 accord and has sections dealing with many specific cases, but the broad intention is that risk weights should be set according to the credit rating of the customer. In the standardized approach, the credit rating must be made by an organization outside the bank such as a credit rating agency, e.g., Standard & Poor’s. One set of suggested risk weights for governments and banks is shown in Table 23-3. Table 23-4 shows the risk weights for exposures to corporations.
There are two significant implications of setting capital according to credit ratings. It puts great pressure on the rating agencies, and it requires the bank to have a system that can track and retrieve the rating for every loan. Despite these difficulties, it is a great advance over the simplicity of the 1988 weightings.
The approach also allows an additional reduction in RWA for assets that are collateralized by financial instruments, e.g., bonds and equities.
In its most basic form, the risk weight with collateral is calculated from the initial risk weight, the exposure (E), and the adjusted value of the collateral (CA):
The adjusted value of the collateral is calculated by taking its current value (C) and dividing it by “haircut” factors that trim off a little of the value according to the volatility of the exposure (HE), the volatility in the collateral value (HC), and any exchange-rate volatility (HFX):
TABLE 23-3 Risk Weights for Governments and Banks Under the New Standardized Approach
TABLE 23-4 Risk Weights for Corporate Exposures Under the New Standardized Approach
For cash, the haircut is zero. For bonds, the typical haircut is between 1% and 12% depending on the remaining maturity and credit grade of the bond. For equities, the haircut is between 20% and 30%. The haircut for exchange-rate volatility (HFX) is 8%.
As an example, if we were lending a “main index” equity, the exposure haircut (HE) would be 20%. If as collateral we took a BBB-rated, 4-year government bond, the collateral haircut (HC) would be 3%. If the bond was in a different currency than the equity, we would add an additional haircut (HFX) of 8%. The adjusted collateral value would therefore be 76% of the current value:
The standardized approach is relatively easy to implement, but gives inaccurate assessments of risk when compared with the economic-capital measures that we discussed in previous chapters.
For those banks with the tools in place to measure risk more finely, the new Basel accord will allow banks the option of going beyond the standardized approach to measure credit risk using their own internal models. To encourage more sophisticated risk measurement, the Basel Committee intends that the regulatory capital for the internal ratings-based (IRB) approach should in general be less conservative than the standardized approach. This means that if banks choose to go to the trouble of using the IRB approach, they should expect to reduce the amount of required regulatory capital.
At its core, the IRB approach is very similar to economic capital. For loans to banks, governments, and corporations, the starting point is an equation that gives the benchmark risk weight (BRW). The BRW is the risk weight for a $100 loan of 3 years’ maturity and a loss given default of 50%. The BRW is specified to be a function of the probability of default (P):
Here, Φ is the cumulative Normal distribution, and Φ−1 is its inverse. In Excel, these are given by Normdist(x,0,1,1) and Norminv(x,0,1). For a standard loan, the regulatory capital will be 8% of BRW, multiplied by the dollar amount of the loan, divided by 100.
Table 23-5 compares the capital required by the IRB approach with the capital required by the standardized approach. This table is constructed by approximately mapping the probability of default to the implied rating, and then calculating the standardized capital from 8% multiplied by Table 23-4. The results show that for highly rated companies, the IRB approach will tend to give a lower capital. However, for lowly rated companies, the standardized approach gives lower capital. This table is for the benchmark loan and does not take into account the possibility of a loss given default other than 50%.
We can also compare the regulatory capital with economic capital. Notice that the calculation of BRW is very similar to our earlier calculation of the economic capital using the covariance credit-portfolio approach. In Chapter 20, we calculated the unexpected loss contribution of a loan to the portfolio based on its stand-alone UL and its average correlation with the rest of the portfolio:
The required economic capital attributed to a loan was shown to be a simple multiple of ULC. Let us conservatively estimate that the capital multiplier is 8:
TABLE 23-5 Approximate Comparison of Regulatory Capital Required by the Standardized and Internal Ratings-Based Approaches
Furthermore, assume that the customer has an asset correlation of 40% with the rest of the portfolio, then from Table 20-4 we can read off the default correlations. If we assume that the average loan in the portfolio has a probability of default of 100 basis points, we can see from Table 20-4 that a loan with a 10-basis-point probability of default will have an average default correlation of 4.3%. Using this information, we can compare the regulatory capital and the estimate of the economic capital as in Table 23-6. With the mildly conservative assumptions of a high capital multiplier and a high asset correlation, the economic capital comes out to be virtually the same as the regulatory capital.
In this table, the dollar amounts are with reference to a $100 loan. Notice that the formula for BRW implicitly contains the information about the correlation of the assets with the rest of the portfolio. This was a deliberate decision by the Basel Committee because they did not feel that the credit-risk portfolio methodologies were sufficiently reliable to use them for setting regulatory capital. The major problem is the difficulty of back-testing credit-portfolio models. For a further discussion, see the committee’s survey paper: “Credit Risk Modelling: Current Practices and Applications,” Basel Committee on Banking Supervision, April 1999.
TABLE 23-6 Comparison of Regulatory and Economic Capital Results
Once the BRW has been calculated for the standard loan, the accord allows it to be modified depending on the maturity of the loan and its estimated loss given default. This gives the risk weight for the particular loan:
Here, b(P) is a function of the probability, and M is the effective maturity in years. The final step in calculating the risk weight for the asset is to multiply it by the dollar amount of the exposure at default (divided by 100):
RWA = RW × EAD
Once the RWA has been calculated for each asset, they are summed to get the RWA for the portfolio as a whole. Finally, the portfolio’s total RWA is adjusted to account for any concentrations and large loans. This “granularity” adjustment requires summation involving the probability of default, LGD, and EAD of every loan in the portfolio. Importantly, to carry out this adjustment, the bank needs to know the total exposure to each borrower. This is difficult to accomplish because one company may have slightly different names in different geographies or be recorded differently in different branches of the bank.
To use the IRB approach, we need to know values for the probability of default, LGD, the EAD, and the effective maturity (M) for each loan. In estimating these parameters the accord allows for two levels of sophistication: the foundation IRB approach, and the advanced IRB approach.
In the foundation approach, the bank needs to specify only the probability of default. The values for all the other variables are specified in the accord or by the national regulator, depending on the type of collateral, whether there are associated guarantees or credit derivatives, and the type of credit exposure. For example, the exposure of a line of credit is set at the current disbursed amount, plus 75% of the remaining line. In the advanced approach, the bank may specify the LGD and EAD for each product.
The discussion above applies to exposures to governments, banks, and corporations. The procedure for retail exposures is similar, whereby customers are grouped into segments, and then a BRW equation is used to set the capital for each segment. However, the exact form of the BRW equation is still under discussion.
To be allowed to use the IRB approach, the bank must show that it has a reliable process for measuring and managing risk. Specifically, the bank must establish a credit-grading system with at least 6 buckets arranged such that no more than 30% of the portfolio falls in any one bucket. There must be separate grading systems to measure the probability of default and the loss given default. There must also be strong formal control and documentation over the use of models and data. Specifically, the models must be controlled by a group that is independent from the group who profits by making loans, and any changes to the models must be documented and justified.
To ensure that the models are performing well, their predictions must be regularly compared to the actual results. For example, the bank should collect information on all defaults and compare each loss given default with that predicted by the models at the time of granting the loan.
As proof that the bank fundamentally believes in the models that it is using to calculate the regulatory capital, the numbers must pass the “use test.” The use test requires that any numbers (such as LGD) used in the calculation of the regulatory capital must also be used in the models that the bank uses to run its business, namely the pricing models and economic capital calculator.
Although economic capital is not directly used for the calculation of RWA, we will see that if the bank adopts the advanced IRB approach, it is required to disclose its economic capital as part of the market discipline pillar. This effectively means that any bank using the advanced IRB must also calculate economic capital.
The supervisory review pillar is a set of qualitative recommendations meant to fill any unforeseen gaps in the capital measurement pillar. It also seeks to ensure that the bank has a strong risk-management process and culture. It requires that the bank’s board and senior management should be aware and approve of the risk-management process, and that the process should be comprehensive and reliable. Most of the details of the supervisory review are left to the national regulators.
An important and potentially difficult part of the accord is the requirement to disclose information so that other capital-markets players can judge the bank’s credit-worthiness. This provides incentive for the bank’s management to have sound risk management so they will not be charged high interest-rates when borrowing from other banks. The disclosures are expected to be semiannual or quarterly. The list below gives a sample of the information that a bank should disclose if it wants to use the advanced IRB approach.
• For each risk grade, the nominal exposure amount, the effect of collateral, and the weighted average maturity
• For each risk grade, the actual experience of losses compared with the predicted experience, and for LGD the standard deviation of losses
• A comparison of economic capital, actual capital, and minimum regulatory capital
• The risk-weighted assets per risk grade, including and excluding the effects of collateral, netting, guarantees, and credit derivatives
This disclosure is difficult for two main reasons: it requires large amounts of data collection, and it requires the bank to disclose information that it could consider proprietary or the source of competitive advantage. For example, a bank that has invested in sufficient information technology to predict default probabilities and losses accurately will have a competitive advantage when pricing and selecting new loans. Giving away such information will help other banks become equally selective.
There are five major steps that a bank must take to implement the new accord: save historical data, decide the best approach, understand the full data requirements, build models, and build a system to report the results.
The first step in implementing the new accord is to start building a set of historical data. The accord requires that probabilities of default should be based on five years of data, although when a bank first adopts the IRB, it may be allowed to use only two years of data. Without data, a bank limits its future options for adopting one of the IRB options and for building economic-capital models. The data does not initially need to be in a grand centralized database; it can even be on paper, so long as it exists somewhere and can be extracted later.
The types of data to be collected were discussed in Chapter 19. Data needs to be collected on the characteristics of the borrower at the initial time of application, the characteristics of the product (loan or line, type of collateral, seniority, etc.), and the ratings given at that time. Data should also be collected on any significant changes in the customer over time, such as a change in rating. For products with variable balances, such as lines of credit, the exposure should be tracked over time. Ideally, a snapshot should be taken of all customer data at the end of each year.
The data on the customer at the time of application can be used to make loan-application models and pricing models. The snapshot data can be used to make portfolio models to answer the question: given this portfolio at the beginning of the year, how much capital do I need to survive until the end of the year?
To create the models for probability of default, loss given default, and exposure at default, information must be captured on any customer that defaults. This information should include the date of default, the exposure at default, the timing and amount of any repayments, and if available, the trading price of the bond or loan before and after default. There must also be a mechanism for keying the default information back to the original customer information.
The bank must decide which approach it will aim to adopt: standardized, foundation IRB, or advanced IRB. Part of that decision can include a migration plan first to adopt one approach, then to evolve to another. The main factors in the decision should be as follows:
• How much cost and management effort will be required to implement the plan, given the bank’s current systems?
• How much of the work overlaps with the work that will be required for the bank’s current plans to calculate economic capital?
• What will be the saving in required regulatory capital under each approach, given the bank’s portfolio?
• What is the significance of reducing required regulatory capital? Is the bank’s current required economic capital far in excess of the regulatory capital?
• If the bank adopts a more advanced approach, it will be seen by the regulators and the rest of the banking industry as a sophisticated player. How important is this? Will it reduce the burden of regulatory oversight, or increase the bank’s influence in setting regulatory policy? Will its reputation as a sophisticated bank help it attract more customers or increase its share price?
• If the required regulatory capital was reduced, would the bank be able to improve its credit rating?
• If the bank is seen as sophisticated, and has a high regulatory-capital ratio, will this significantly reduce the bank’s cost of raising debt? If the bank raises most of its debt from financially unsophisticated retail depositors, the change in the debt-funding cost will be minimal. If most of the debt is from the interbank market, the change could be significant.
• How much sensitive information would the bank have to disclose? Would the use of that information aid competitors; e.g., could they use high-quality loss data in their models to compete more effectively against the bank?
Table 23-7 gives a comparison of some of the costs and benefits for the three approaches.
TABLE 23-7 Comparison of the Costs and Benefits from Adopting Each of the New Regulatory-Capital Approaches
The data requirements fall into three categories: historical data needed to build the models, live data needed to calculate the required capital, and data needed for disclosure. The data needed for disclosure is very extensive, especially for the IRB approaches, and will only be summarized here.
For the standardized approach, to calculate the capital, the bank needs only to collect exposure and rating information for each of its customers. The rating information must be from standardized sources and be well documented.
In addition to calculating the capital, the bank is required, under pillar three, to disclose the composition of its portfolio in terms of its total exposure to each geography, industry, and type of product (e.g., loans or lines). The bank also needs to be able to disclose the maturity distribution, e.g., what percentage of the loans is for more than one year.
For the foundation IRB approach, the capital is calculated based on the probability of default, which requires the bank to have a system or model to link customer characteristics to default probabilities. If models are used, they will be constructed based on historical default data, which the bank must collect or obtain from an external source. To calculate the granularity adjustment, the bank must know the total exposure to each customer, which means having unified customer codes across the bank. If the bank chooses to reduce its capital by counting collateral, it must also collect collateral information for each loan.
In addition to calculating the capital, the bank must disclose the exposure amount in each probability grade and the effects of any collateral adjustments. It must also disclose information on the performance of its models by publishing the number of loans in each probability grade that later went on to default.
In the advanced approach, the bank must collect and disclose the same information on LDG and EDA. It must also publish statistics on the LGD and EAD that it experienced. Significantly, it must also publish its estimate of economic capital, which implies that the bank must also have a portfolio model such as the ones discussed in Chapter 20 and Chapter 21.
The historical data is used to create models that link borrower and product characteristics to expected probabilities of default, LGD, and EAD. Building the models proceeds in the same way as discussed in Chapter 19. The main difference is that the risk-management organization needs to document all the processes used to create the models. The documentation should ensure that the models are repeatable, independent of individuals, and not open to manipulation. Also, there will be little room to make ad hoc adjustments that are based on the model builder’s intuition.
The reporting process must pull together all the elements described above. This will typically require that a person in each business unit should be designated to be in charge of the data-collection effort to aggregate results for that unit. In consultation with the local regulators, a schedule for reporting will need to be established. For the pillar-three disclosures, this will be semiannual or quarterly. For the calculation of required and available capital, this will be quarterly or monthly. The whole reporting process should be documented with guidelines to ensure that the reporting can be repeated even if individuals change.
In this section, we discuss what the bank should and can do if the required regulatory capital is significantly different from the required economic capital. In this discussion, we use four concepts: the available capital, the required economic capital, the minimum required regulatory capital, and the target required regulatory capital.
In the rest of the chapter we have concentrated on calculating the minimum required regulatory capital. However, if the bank’s available capital falls below this amount, it is in trouble. Therefore, the bank normally has a target of ensuring that the available capital is some margin (e.g., 2%) above the minimum required capital. We call this the bank’s target required regulatory capital.
The bank should keep two sets of accounts: economic capital based and regulatory capital based. In general, the bank should be managed so the required economic capital, the target required regulatory capital, and the available capital are approximately equal. This applies at the highest level of the total bank, but does not need to apply at lower levels, such as individual business units or transactions.
Bear in mind that economic capital is the bank’s best assessment of the amount of capital that it actually needs, given its portfolio, operations, and desired credit rating, whereas regulatory capital, especially under the 1988 accord, is a crude measure of risk that is relatively easy for an external regulating body to calculate and check.
If the required economic capital is more than the required regulatory capital, it means either that the bank has risks that are not captured by the regulatory capital, or the bank has a higher target debt rating than the one implied by regulatory capital. In either case, both the regulatory-capital requirement and the economic-capital requirement will be satisfied if the bank holds capital equal to the required economic capital.
If the target regulatory capital is greater than the economic capital, this means either that the bank is actually safer than the regulators believe, or that it has a low target debt rating. In this case, the bank should reduce the difference between the required regulatory capital and the economic capital.
The bank can reduce the difference by shifting into assets with greater risk but the same regulatory capital requirement. Under the 1988 accord, this can be done simply by giving loans to lowly rated companies and refusing loans to highly rated companies. This can be accomplished by an ad hoc modification of the hurdle rate used in the profitability and pricing calculations to include a penalty if the asset attracts regulatory capital. Such a modification is illustrated in the equation below:
Here, ECi and RCi are the economic and regulatory capitals for the individual transaction, and RCT and ECT are the capitals for the total bank. A refinement to this equation would be to add a scaling factor (s) and only add a charge if the total regulatory capital was greater than the economic capital. If the scaling factor was made larger, the business units would have a stronger incentive to move to assets with a high ratio of ECi to RCi:
Banks can also reduce their regulatory capital by manipulating the accounting. This entails entering into trades that remove assets from the formal balance sheet. Such trades include issuing heavily collateralized asset-backed securities. In this case, the assets are counted as if they have been transferred away from the bank and into a legally independent company that is a “special-purpose vehicle” (SPV). The SPV then issues bonds backed by the assets. Most of these bonds are bought by other banks and investors. The SPV gives the proceeds of the bond issuance to the bank in payment for the assets. On the balance sheet, the assets are now replaced with cash, which attracts a risk weighting of zero.
However, part of the structure of the SPV will be an agreement by the bank to buy the most risky of the bonds issued by the SPV (i.e., the “toxic waste”). These bonds appear on the balance sheet with a small face value and small risk weight, but with great underlying economic risk. By such complex accounting games, the bank can reduce its required regulatory capital without significantly reducing the risk or economic capital. However, regulators tend to view such transactions dimly if they are too blatant.
A third alternative for balancing the capital is to keep the risk the same but increase the required economic capital up to the level of required regulatory capital, by increasing the target debt rating. The reasoning behind this is that if the available capital is greater than the required economic capital, the bank is actually safer than the current rating implies. If this is the case, the bank should be able to convince the rating agencies to improve its rating, thereby reducing the bank’s cost of borrowing and its required rate of equity return.
To increase its rating, the bank would have to make the case to the credit-rating agencies that it is safer than the agencies currently believe. It could make this case by presenting the economic-capital calculations, showing that with the bank’s risks, and the capital that it is holding, it has a very low probability of default, and therefore should be rated more highly.
As the calculation of regulatory capital becomes more similar to the calculation of economic capital, there should be less need to manage the differences.
In this chapter, we discussed required regulatory capital for credit risks with a focus on the choices available to management and the steps required to use the new accord. Next, we explore the capital needed to cover operational risk.
1. International Convergence of Capital Measurement and Capital Standards, Basel Committee on Banking Supervision, July 1988.
2. Amendment to the Capital Accord to Incorporate Market Risks, Basel Committee on Banking Supervision, January 1996.
3. Consultative Document, The New Basel Capital Accord, Basel Committee on Banking Supervision, January 2001.