class notes Economic capital ideas Class Notes is an educational series, designed to pull together the threads of recent developments and thinking about key issues in risk management and derivatives dealing. In this instalment, Charles Smithson explains the concept behind economic capital Economic capital is a measure of the amount of equity capital an enterprise needs to support a risk. More specifically, it is the amount of equity capital necessary to cover losses arising from that risk to some confidence level.1 For example, many of the large international banks define economic capital as the amount of equity capital needed to cover losses 99.97% of the time. In contrast with an accounting view, where capital could be viewed on the right-hand side of the balance sheet, economic capital is a ‘left-hand-side-of-the-balance-sheet’ concept: the amount of capital needed is determined by the riskiness of the company’s assets (including the firm’s business units and activities). While equity capital is the source of the ‘needed’ capital as defined by economic capital, the amount of equity capital required can be reduced by insurance and guarantees or by transferring risk to a third party. Rationale Firms face a range of risks: market risk, credit risk, insurance risk (for example, mortality and morbidity) and operational risk, as well as business risk, reputation risk, compliance risk and liquidity risk. Each risk has its own risk measures. For example, market risk measures include duration and convexity (for interest rates), the Greeks (delta, gamma, vega, theta, and so on) and, more recently, value-at-risk (VAR), while credit risk measures include characteristics of the obligor (probability of default), characteristics of the transaction (exposure in the event of default and loss in the event of default) and characteristics of the portfolio, which do not consider correlation (expected loss and concentrations). For market, credit and operational risk at least, economic capital has become the language of risk. While the risk-type-specific risk measures are still used, market risk, credit risk and operational risk can all be expressed in terms of the amount of equity capital needed to support that risk and, since the measures are in currency units, we have an apples-to-apples measure. In addition to providing a common language, economic capital permits the firm to express its risk appetite. (Above we noted that many large banks use a 99.97% confidence level for measuring economic capital. That confidence level was selected 1 obert Merton and Andre Perold described risk capital as providing a kind of asset insurance against the possibility of lower-than-expected R operating results. risksouthafrica.com 37 class notes know about economic capital, not only for the enterprise as a whole but also for component business units, the calculation of economic capital is normally accomplished using a bottomup approach. Recognising that the different risk types share common determinants, some firms rely on a single model that produces an economic capital value that reflects market risk and credit risk (and, at least theoretically, operational risk).2 Clearly, the benefit of such a model is that the correlations of market, credit and operational risk are captured. However, the more common approach is for the firm to use individual models for each of the risk types. Economic capital for market risk is obtained from some kind of VAR model. Economic capital for credit risk is obtained from a credit capital model – either internally developed or vendor-supplied. Economic capital for operational risk could be obtained using a process approach (which focuses on the chain of activities that comprise an operation or transaction), a factor approach (which relates the level of operational risk to factors that have been identified as significant determinants) or an actuarial approach (which identifies the loss distribution associated with operational risk). The separate economic capital measures are then aggregated. The obvious question is how to deal with the correlation between risk types. Simply adding up the economic capital numbers for the individual risk types is overly conservative, because it assumes perfect positive correlation – in other words, no diversification across risk types. However, estimating the degree to which the different risk types are correlated is an extremely difficult task. Approaches to this task are characterised as bottom-up and top-down. A bottom-up approach can be viewed as a five-step process: Charles Smithson, Rutter Associates because it corresponds 100% minus the historical likelihood of an AA rated firm defaulting in a one-year period. This reflects the risk appetite of senior management and the board in terms of credit rating.) Calculating economic capital at the enterprise level Early work on economic capital envisioned being able to calculate economic capital for a firm using a top-down approach. If it were possible to obtain daily, weekly or monthly observations on earnings, the resulting distribution would reflect not only the effects of market, credit and operations risk, but also the correlation of these risks. Specifying a confidence level determines earnings-at-risk (EAR). To convert EAR to economic capital, the enterprise identifies the amount of equity capital required today to ensure that EAR in future periods is offset by risk-free earnings. Economic capital = Earnings at risk r Risk South Africa Autumn 2008 A top-down approach can be viewed as a three-step process: (1) e stimate the distributions of the individual types of economic capital; (2) define the dependence structure; and (3) generate joint distribution of risk types.3 Risk contributions While the ability to measure economic capital at the enterprise level and report that measure to stakeholders and regulators is valuable, the value of the economic capital concept is potentially much larger when it is applied to some core business decisions: e 2006 IFRI-CRO Forum survey into the economic capital practices of 17 banking institutions and Th 16 insurers in Europe, North America, Australia and Singapore indicated that insurance companies are more likely than banks to incorporate the different risk types into a single economic capital model. 3 The 2005 survey conducted by PricewaterhouseCoopers and The Economist Intelligence Unit found that 33% of the 200 financial institutions surveyed do not incorporate the correlation between risk types. The 2006 IFRI-CRO Forum survey suggested that a similar, albeit smaller, percentage of respondents were not incorporating correlation between risk types: six of the 33 participants reported that they use a simple summation approach. Of the 27 that attempted to incorporate inter-risk correlation, 23 characterised their approach to inter-risk diversification as top-down. 2 where r is the risk-free interest rate. Because of difficulties in obtaining reliable high-frequency earnings data and, more importantly, because firms want to 38 (1) e stimate the distributions of economic risk factor, which might include the term structure of interest rates, credit spreads, equity returns and other macroeconomic factors; (2) relate the economic risk factors to individual risk types; (3) generate marginal loss functions for risk-factor changes; (4) define the dependence structure; and (5) g enerate joint distribution of risk-factor changes and model the impact of these risk factor changes on the loss function for the enterprise. Evaluating the performance of business unit managers. A llocating capital to business units – that is, determining which businesses to grow and which to shrink. n Internal pricing. n Entry/exit decision – that is, deciding which businesses to acquire and which to divest. n n An aggregate measure of economic capital for the enterprise is not sufficient to address these issues. It is necessary to determine the amount of economic capital that should be attributed to individual business units or portfolios or transactions – called risk contributions. This is not a theoretical argument: firms are attributing economic capital to individual business units/portfolios/ transactions. Rutter Associates’ 2004 Survey of Credit Portfolio Management Practices demonstrated that fact in the case of economic capital for credit risk. All of the 44 financial institutions that participated in the survey indicated that they attribute economic capital to individual business units or individual transactions. The complication is that there is more than one way to measure risk contributions. Stand-alone risk contributions The stand-alone risk contribution is the amount of equity capital an individual business unit, portfolio or transaction would require if it were an independent enterprise. (The model or models used to measure economic capital for the enterprise would be run again using the individual business unit/portfolio/transaction as the enterprise.) Stand-alone economic capital does not reflect the diversification benefits that would be obtained if the business unit/portfolio/transaction is part of a larger enterprise.4 Standard-deviation-based marginal risk contributions The standard-deviation-based marginal risk contribution for the ith business unit, portfolio or transaction is: RC i = wi ∂σ E ∂wi where σE is the standard deviation of returns for the enterprise, and wi is the weight (position size) for the ith business unit or portfolio or transaction The measure is referred to as a standard-deviation-based measure because it looks at the impact on the enterprise’s standard deviation of changes in the size of individual business units/portfolios/transactions. It is referred to as a marginal risk contribution because of the use of the partial derivative: we are looking at changes in the enterprise standard deviation with respect to small changes in position size. If the weights are scaled appropriately, the sum of the standard-deviation-based risk contributions will add up to the enterprise standard deviation. 1 Shortfall-based risk contribution Probability Simulated loss distribution Loss corresponding to ‘threshold’ confidence level Loss corresponding to confidence level used to calculate economic capital A diversified risk contribution measures the amount of total enterprise economic capital that should be attributed to an individual business unit/portfolio/transaction when it is viewed as a part of the multi-business-unit/portfolio/transaction enterprise. This approach attributes the diversification benefit to the individual business units/portfolios/transactions that make up the enterprise, in the form of economic capital contributions that are lower than the stand-alone contributions. Consequently, the sum of diversified risk contributions for all the business units/portfolio/transactions will be equal to total economic capital for the enterprise. To provide some intuition for a shortfall-based diversified risk contribution, consider attributing economic capital for credit risk to the different business units/portfolios/transactions. Figure 1 illustrates the simulated loss distribution for the enterprise. The loss corresponding to the confidence level defined by senior management and the board (for example, the 99.97th percentile in the case of the large, international banks) is identified by the up arrow. Select a threshold confidence level lower than the confidence level used for measuring economic capital (for example, if the confidence level for economic capital is 99.97%, the threshold level might be 99.5%). In figure 1, the corresponding loss level is indicated by the down arrow. Identify the iterations of the simulation in which simulated loss exceeded the threshold level. For business unit/portfolio/ transaction A, the conditional loss rate is: Conditional = loss rate Shortfall-based diversified risk contributions Instead of considering how an individual business unit or portfolio or transaction contributes to the enterprise standard deviation, a shortfall-based measure focuses on the amount it contributes to loss when the enterprise is under stress – that is, when the observed (or simulated) magnitude of the loss for the enterprise is in the upper tail of the loss distribution. (Consequently, these measures are also referred to as tail-based measures.)5 Currency units (for example, US dollar) Mean simulated loss 4 5 Conditional number of defaults for ‘A’ Number of losses above ‘threshold’ level onsequently, the sum of stand-alone capital for the enterprise’s individual business units or portfolios or C transactions will be greater than the total economic capital for the enterprise. At least in the case of economic capital for credit risk, the survey evidence indicates that firms are moving from standard-deviation-based to shortfall-based risk contributions. Between the 2002 and 2004 Rutter Associates’ surveys of credit portfolio management practices, the use of standard-deviation-based risk contributions declined (from 50% to 38%), with the use of shortfall-based risk contributions increasing (from 13% to 33%). risksouthafrica.com 39 CLASS NOTES where the numerator is the number of defaults for A, conditional on simulated loss for the enterprise exceeding the threshold level. Then, the conditional attribution to business unit/portfolio/ transaction A – the shortfall-based risk contribution – is: Conditional Conditional loss = attribution to ‘A’ rate ‘A’ x LGD(A) indicated that they were broadly comfortable with the accuracy of outputs. Looking only at economic capital for credit risk, the 2004 Rutter Associates Survey of Credit Portfolio Management Practices found that 43 of the 44 financial institutions that responded to the survey calculated economic capital. Insurance companies Shortfall-based marginal risk contributions As long as the threshold is defined as a percentage (for example, 99.97%), the marginal risk contributions sum to total economic capital for the enterprise (Pearson (2002), pages 161–162). For the 16 insurance companies that participated in the 2006 IFRI-CRO Forum survey, economic capital frameworks had been in place at insurance companies an average of four years. Like the banks, 12 of the 16 insurance companies indicated that they were broadly comfortable with the accuracy of outputs. Writing in Best’s Review, a monthly magazine covering the insurance industry, early in 2007, Robert Stein had a different perspective. He noted that, even though nearly 90% of the respondents to an Ernst & Young survey of North American life insurers had indicated that economic capital is essential to performance evaluation and making strategic decisions about mergers and acquisitions, the majority of US insurers have yet to integrate economic capital into their core decisionmaking process. Which risk contribution measure is right? Investors As with so many things, the answer is: it depends. In this instance, the right risk contribution measure depends on which of the business decisions is being considered. In 2004/05, the HSBC Financial Products Institute received responses regarding risk measurement and management practices from 20 mutual funds, 51 pension plan sponsors, 30 university endowments, 20 foundations and nine hedge funds. These responses indicated that economic capital had not made much of an inroad with institutional investors: only three out of 49 institutional investors indicated they were employing an economic capital measure for market risk and just five out of 115 indicated they were employing an economic capital measure for credit risk. n Instead of considering how a small change in the size of an individual business unit, portfolio or transaction affects the enterprise standard deviation, this measure considers how a small change in the size of an individual business unit, portfolio or transaction affects the expected shortfall (ES): RC = wi ∂ ( ES ) ∂wi I t is generally agreed that stand-alone capital is appropriate for measuring the performance of business unit managers. n It is also generally agreed that marginal economic capital is the appropriate measure for evaluating acquisitions or divestitures. n However, there is less agreement about the appropriate measure for allocating capital to business units and internal pricing. While academics argue that marginal economic capital should be used, practitioners appear to be using a measure of diversified economic capital. Proponents of the diversified economic capital approach point out that marginal capital always under-allocates total enterprise capital and that, even if the marginal capital numbers were scaled up, the signals sent about profitability are potentially misleading. Proponents of the marginal economic capital approach argue that, because economic capital is not additive across business units, any rule that fully allocates enterprise economic capital across the business units will be sub-optimal. n How widely used is economic capital? In October/November 2005, PricewaterhouseCoopers and The Economist Intelligence Unit conducted an online survey of financial institutions in Asia, Europe and the Americas. Fortythree per cent of the 200 respondents reported they were using economic capital: 27% were using economic capital across the enterprise and 16% were currently using economic capital in certain units. Banks For the 17 banks that participated in a 2006 survey by the International Financial Risk Institute (IFRI) and the Chief Risk Officer (CRO) Forum, economic capital frameworks had been in place at banks an average of six years. And 12 of the 17 banks 40 Risk South Africa Autumn 2008 Charles Smithson is a partner at Rutter Associates. He would like to thank Neil Pearson for his help on this article. Email: csmithson@rutterassociates.com References Christian C, 2006 Copula-based top-down approaches in financial risk aggregation Working Paper 32, the University of Applied Sciences of bfi Vienna IFRI Foundation and Chief Risk Officer Forum (CRO Forum), 2006 Insights from the joint IFRI/CRO forum survey on economic capital practice and applications Merton R and A Perold, 1993 Theory of risk capital in financial firms Journal of Applied Corporate Finance Pearson N, 2002 Risk budgeting: portfolio problem solving using value at risk Wiley PricewaterhouseCoopers, 2005 Effective capital management: economic capital as an industry standard? Rutter Associates, 2004 Rutter Associates survey of credit portfolio management practices Sponsored by International Association of Credit Portfolio Managers, International Swaps and Derivatives Association and Risk Management Association Rutter Associates HSBC Financial Products Institute 2004–2005 survey of risk measurement & management practices of institutional investors Stein R, 2007 Using economic capital as a tool Best’s Review