Set out what the Bank for International Settlements (BIS) recommends supervisors weigh in order to put internal risk measures such as economic capital, which were not built for regulation, to good use.
The idea
Economic capital models were built for internal management, not for regulation, so the BIS asks supervisors to use them with care rather than adopt or dismiss them. The supervisor should first understand the design choices that make the output differ from regulatory capital. These are the confidence level, the time horizon, the risks covered, the diversification assumed and the definition of capital it is compared with. Two banks with similar portfolios can report very different figures for these reasons alone, so figures must be put on a common basis before they are compared. Supervisors should also weigh the quality of model validation, data and governance, and how far the bank really uses the numbers in decisions. Economic capital is then supplementary evidence in the supervisory dialogue, not a replacement for the regulatory minimum. The usual mistake is to read a gap between the two measures as a verdict on risk.
Formulas
Restating normal economic capital at another confidence level
EC₂ = EC₁ × z₂ ÷ z₁
- EC₁
- economic capital at the original confidence level
- z₁
- the standard normal quantile at the original level
- z₂
- the standard normal quantile at the new level
Assumes normal losses and the same horizon. Putting figures on a common basis comes before any comparison across banks.
A practice question
Which statement best captures the BIS view on using internal measures such as economic capital in the supervisory process?
- Economic capital can replace the regulatory minimum, because internal models are more risk sensitive than regulatory rules.Why not: BIS does not endorse replacement; the measures were not designed for regulation and differ in design from regulatory capital.
- Economic capital was built for internal management, so supervisors should understand its design and limits before using it as supplementary information. - correct
- Economic capital should be ignored, because an unvalidated internal opinion adds nothing to supervision.Why not: Dismisses a source of information the BIS regards as useful when its limitations are understood.
- Economic capital can be accepted unchanged once the bank's board has approved the model.Why not: Board approval does not relieve supervisors of assessing assumptions, validation and use.
The BIS stance is neither adopt nor reject. Because economic capital was not designed for regulation, supervisors weigh its purpose, assumptions, validation and governance, and then use it as a complement to regulatory measures.
A harder one
A supervisor compares a bank's credit economic capital of USD 500 million, calibrated at 99.5% one-year confidence, with the supervisory standard of 99.9%. Assuming normally distributed losses and the same horizon, the figure restated at 99.9% is closest to:
- USD 600 million - correct
- USD 502 millionWhy not: Scales by the ratio of confidence levels, 0.999 / 0.995, instead of the ratio of normal quantiles.
- USD 2,500 millionWhy not: Scales by the ratio of tail probabilities, 0.005 / 0.001, which overstates the effect for a normal distribution.
- USD 417 millionWhy not: Inverts the quantile ratio, moving to a higher confidence level yet lowering capital.
Under normality capital is proportional to the quantile. z(99.9%) = 3.090 and z(99.5%) = 2.576, ratio 1.200, so 500 x 1.200 = 600. A different confidence level alone can make two banks' figures differ by a fifth, which is why they must be restated on a common basis.
The hardest tier
Two banks hold similar credit portfolios. Bank A reports economic capital of USD 1,000 million at 99.9% with full diversification across risk types, and Bank B reports USD 1,350 million at 99.97% with simple summation across risk types. What should a supervisor conclude?
- Bank B holds a riskier portfolio, since its capital requirement is 35% higher than Bank A's on a similar portfolio.Why not: Treats the gap as a risk difference; the banks use different confidence levels and aggregation.
- Bank A is undercapitalised, because it should hold at least as much as Bank B.Why not: Assumes the larger figure is the correct benchmark; the figures rest on different assumptions.
- The average of the two figures is the best estimate of industry practice for the portfolio.Why not: Averaging mixes incompatible bases and removes the information the supervisor needs about each bank's assumptions.
- No conclusion about relative risk is possible until confidence level and diversification are put on a common basis. - correct
The gap may come from the higher confidence level and from Bank B's lack of diversification credit, not from portfolio risk. Supervisors should restate both on a common basis, using sensitivity analysis, before comparing.
In the app: 6 practice questions on this outcome, spaced review that brings it back before you forget it.
Start free →More in Range of practices and issues in economic capital frameworks
- Describe the difficulties met in implementing an economic capital framework when defining and computing risk measures, aggregating risk, validating models, modeling dependency in credit risk, evaluating counterparty credit risk, and assessing interest rate risk in the banking book.
- Explain how adopting an economic capital framework helps and changes outcomes in four areas: managing the credit portfolio, pricing for risk, analysing how profitable customers are, and shaping management incentives.
- Outline good practice in governing an economic capital framework and judge the main concerns that arise.