Risk Management of a Financial Conglomerate

Article from:

Risk Management Newsletter

March 2004 – Issue No. 1

March 2004 ◗ Risk Management

Risk Management of a Financial Conglomerate

by Luc Henrard, chief risk officer Fortis and Ruben Olieslagers, head of research and development

1. A NECESSITY consistent view on risk measurement across the The emergence of large European financial entire financial services industry. The Joint groups has been one of the principal features of Forum (formerly known as the Joint Forum on the latest banking and insurance consolidation Financial Conglomerates) has been a focal point wave. Financial deregulation, globalization of of the efforts of the international supervisory financial markets and increased shareholder community in meeting this need. pressure for financial performance are the main forces that fueled the mergers and acquisitions The concept of “Economic Capital,” which trend over the past few years. In order to meas- measures risk based on a company’s own unique Luc Henrard is ure, monitor and manage risk and ultimately op- risk profile, is developing as the common meas- general manager and timize risk versus return within a conglomerate ure of risk, sought by many financial conglomer- chief risk officer of at both operating entity and aggregate group ates as well as regulatory bodies. Economic Fortis in Belgium. level, the financial conglomerate needs excel- capital enables financial institutions to estab- He can be reached at lent risk-management processes and internal lish a capital framework that allows for consis- [email protected] control mechanisms. This should also be en- tent translation of risk taking into capital . couraged by the regulatory structures, which requirement, making “apples-to-apples” com- are unfortunately still largely focused on indi- parisons possible. An economic capital frame- vidual operating entities within a group and work does not only allow for the capture of treat each of these as independent silos in set- netting and diversification effects within a fi- ting capital requirements. This silo approach nancial conglomerate, it also addresses many of fails to deal adequately with aggregate risks the current limitations of regulatory capital across different regulated businesses. models (e.g., silo view, standardized risk model- ling approaches). Accurate and consistent risk measurement is a prerequisite for good risk management. Risk 2. THE INTERNAL CHALLENGE: measurement typically starts bottom-up in the SIX STEPS different businesses within a financial con- The development of comparable measures of Ruben Olieslagers is glomerate. As a result, many different ap- capital and value is not an easy task. Fortis, as a Head of Research and proaches to risk measurement have been bancassurance group facing a wide range of Projects at Fortis Central developed between insurance and banking risks, has applied the following six-step ap- Risk Management. businesses and even within each of these areas proach: He can be reached at (e.g., life and non-life insurance). For a finan- ruben.olieslagers@ cial group, especially a conglomerate covering • Define and communicate fortis.com. many business areas, arriving at a common risk your risk taxonomy measure is quite a challenge. • Make sure banking and insurance officers Externally too, the growing emergence of understand each other financial conglomerates and the blurring of • Define the models to be used for each distinctions between the activities of firms in risk type (business, event, credit, etc.) each financial sector had also increased the in a consistent way need for joint efforts to improve the efficacy of • Model each risk and aggregate to arrive supervisory methods and approaches. Basel II at an overall capital figure has focused on improving consistency and • Define a regulatory solvency corridor accuracy of setting solvency requirements • Look at the risk/return “Framework” across banking businesses and now Solvency II will aim to do the same for insurance. A key aim of the regulatory bodies is also to develop a continued on page 14 ◗

Page 13 ◗ Risk Management ◗ March 2004

Financial Conglomerate ity correlated loss) differs significantly from the ◗ continued from page 13 distribution for market risk (high frequency, low severity).

Step 1: Define and communicate 2. Insurance Risk your risk taxonomy • P&C risk: the variability in future claims and loss-adjusted expenses (LAE) paid Many different ways of classifying risk are (whether in size of claims, number of claims or possible, and no single taxonomy is inherently timing of payments) and the variability in the better than another. The classification of risk liabilities for outstanding claims overtime. types often follows the relative importance of • Life risk: the risk exclusively associated risk types to a financial services provider. The with a life insurer. The risk is especially the risk taxonomy used at the level of Fortis Group result of deviations in timing and amount of seeks to establish a common risk language the cashflows due to the (non-) incidence of across the group, while ensuring that all risk death. types are adequately captured. Figure 1 distin- guishes six broad types of risks. 3. Operational Risk Figure 1 • Business risk refers to the risk due to operating leverage (in particular, volatile revenues RISK TAXONOMY and an inflexible expense base). • Event risk refers to the risk of experiencing

Risk vs. Risk vs. one-off adverse non-financial events such as Debtholders Capital Return fraud and punitive damages. Rating Agencies RISK=Volatility Shareholders Regulators Stock Analysts Given that a financial conglomerate is by definition a combination of diverse businesses operat- Operational Investment Insurance Risk Risk Risk ing under a common ownership structure, each of these has a distinct risk profile. From this point of view, an ordering of risks in function of Business Event Credit Market Property & Life the consumption of economic capital is Risk Risk Risk Risk Casualty Risks Risks required, taking into account the fact that a

l Changes in l Fraud l Loans l Equities l Claims for l Life policies conglomerate must not be overcapitalized to the business 'normal' events Unintentional Derivative Bonds Annuities volumes l l l l point where it would cause undue harm to share- errors counterparts l Losses due to l Foreign l Health l Changes in catastrophes/ Legal risk Reinsurance Exchange holders or undercapitalized to the point where it margins and l l natural disasters counterparts costs l Man-made l Real Estate (e.g. earthquakes, would cause undue risk of insolvency to debtors

shocks l Settlement hurricanes, l ALM risk floods, etc) and policyholders. In other words, lower capital l Liquidity risk for a given degree of risk taking will make an institution less solvent, but more profitable, and vice versa. 1. Investment risk • Credit risk: the risk that a borrower/coun- Figure 2 gives an illustrative example of order- terparty will fail to repay the amount owed to ing and is therefore not valid for every business the Fortis Group. within a financial conglomerate because it • Market risk: the potential for loss resulting depends on the relative importance of each of from unfavorable market movements (from the banking and insurance businesses within trading to holding positions in financial the conglomerate. In general, universal banking instruments). Market risk might be treated as activities are mainly dominated by credit risk, one aggregated risk or separately as interest but this is not the case for life insurance activi- rate, equity, foreign exchange, real estate and ties. ALM is invariably the largest consumer of commodity risk. Within market risk we identi- capital in insurance companies (especially in fy ALM risk. Fortis Group is exposed to inter- life) given that insurance risks diversify away in est rate, share price and real estate risk via its large portfolios. P&C activities are mainly dom- investment portfolio. Credit risk and market inated by insurance risk while the non-licensed risk are measured separately because the distri- subsidiary encounters operating risk. bution for credit risk (low frequency, high sever- Insurance risks (mortality and underwriting)

◗ Page 14 March 2004 ◗ Risk Management

Figure 2 nancial con- ILLUSTRATION OF RELATIVE CAPITAL CONSUMPTION glomerate, differences in the First Second Third Capital order order order sector-specific 100% frameworks should be identi- fied and, agree- ment should be reached consistently covering all relevant risks. For example, one 0% ity of the key chal- Bank iting –Bank –Bank – ding Insurance Insurance Tra erwr lenges in a con- – ALM – –Insurance Credit Und ALM ality/morbid Operating Credit glomerate is Mort Operating specifying a uniform time hori- will diversify away substantially in large portfo- zon. In banks, the convention for modeling risks lios because they are not correlated with the and assessing capital is to adopt a one-year hori- other (financial) risks and because a lot of the zon. Alternatively, insurance companies are typi- volatility is already reserved in the provisions. cally capitalized for longer decision horizons. In order to have a “common currency” for risk, a com- Step 2: Make sure banking and Figure 3 insurance officers understand each other Step 2 consists of improving the understand- DIFFERENT APPROACHES USED IN BANKING AND INSURANCE ing by bankers and actuaries of mutual approaches and terminology. Figure 3 Banking Insurance summarizes the typical banking and insurance approaches. The dissimilarities are l Expected loss l Claims Rating masterscale Mortality tables substantial, mainly because of the differ- Terminology l l ences in the dominant risk types that have l VAR l Fair Value traditionally been faced. Furthermore, l RARORAC l Embedded Value/Risk Based Capital banks tend to have assets that are difficult to l Risk l Expected outcome value, whereas insurance companies have Focus l One-year l Multiyear uncertain liabilities. Both also use very different valuation principles. Thus, in order to l Insufficient use of statistics l Insufficient use of modern finance Weaknesses theory (in some countries) make sure that banking and insurance un- l Customer behavior Little use of transfer pricing (ALM) derstand each other, knowledge sharing and l communication efforts should be an important issue in a financial conglomerate.

Step 3: Define the models to mon time horizon needs to be specified, at least at be used for each risk type the group level where risk aggregation across (business, event, credit, etc.) in banking and insurance takes place. Another ex- a consistent way ample is the translation of the one-tailed 99 percent confidence interval for trading risk or 95 Step 3 defines the models to be used for each risk percent confidence interval for specific actuarial type in a consistent way. Those risk types are cred- risk into a 99.97 percent confidence level, which it, market, ALM, life, P&C, business and event is applied to be in line with the Fortis “AA” cali- risk. A common risk measurement framework is bration. the prerequisite to an effective measurement and management of risk and used capital. To construct a common risk language across the whole of a fi-

continued on page 16 ◗

Page 15 ◗ Risk Management ◗ March 2004

Financial Conglomerate 99.97 percent) can be derived from the tail of the ◗ continued from page 15 probability distribution. The distribution illus- trated in Figure 5 represents the probabilities of various earnings outcomes from a loan portfolio over a one-year time horizon against which cap- Step 4: Model each risk and ital must be held in accordance with the desired aggregate to arrive at an overall rating. capital figure Step 4 defines the model for each risk in terms of The process to determine how much capital the amount of value they put at risk to a certain is required in a financial conglomerate can confidence limit determined by the target debt be presented schematically as in Figure 6 (see Figure 4 rating. For next page). Fortis, DISTRIBUTION OF RISK TYPES therefore, Clearly, the probability that the sum of all stand- economic alone capital requirements fails to cover losses capital is for all risk types simultaneously is lower than RISK defined as the probability that only one or a few capital re- the amount quirements for a risk type fall short of covering Investment Insurance Operational of value at losses attributed to the risk type. We are inter- Risk Risk Risk risk to a ested in computing an aggregate capital re- 99.97 per- quirement figure for the group that will cover Credit Market Property & Life Business Event Risk Risk Casualty Risk Risk Risk Risk cent confi- potential group losses up to the desired group dence limit confidence level equivalent to an AA-S&P debt (based on a rating. We would clearly overestimate group AA target capital requirements if we were to add up all the debt rating). stand-alone capital requirements, since that It is calcu- would lead us to a much higher confidence level lated by es- than anticipated. +A+ Source: Mercer Oliver & Wyman timating the 99.9% fair value Instead of adding the stand-alone capital re- now and quirements directly, we must aggregate them comparing it considering the tendency for co-movement with the fair value in one year’s time under a among losses for each of the risk types. If we 99.97 percent worst-case scenario for each risk. know to what degree the losses related to a par- One should be aware that it is not that easy to de- ticular risk type tend to follow the losses related termine the distribution of a risk type because, to other risk types, we can compute an aggregate among other things, a great deal of data is need- capital requirement figure for the group to protect against all losses up to the desired confi- Figure 5 ed. Figure 4 illustrates that a different risk distribution is possible for dence level. every risk type. EARNINGS OR VALUE VOLATILITY Within these aggregation steps, diversification Within Fortis, the is taken into account via a set of correlation stand-alone capital re- estimates. Empirically, diversification effects TAIL PROBABILITIES Probability of Outcome quirements for each of are greatest within a single risk factor (Level 1), AAA AA A BBB several risk types decrease at the business level (Level 2) and 0,01% 0,03% 0,07% 0,30% quantify the value at are smallest across business lines (Level 3). Default probability risk for each risk type Recent estimates suggest that the incremental up to an AA confidence diversification benefits achievable at Level III level over a one-year by combining a bank with an insurance compa- period. The economic ny are on the order of a 5-10 percent reduction capital (after having in capital requirements.

0 Earnings or Value quantified the level of K Mean ∑∑∑ K Α Diversification is a complex issue and it is K risk) to achieve such a ΑΑ Capital required to achieve rating = Economic capital K } ΑΑΑ particular level of sol- understandable that regulators are wary of vency (e.g., AA rating – allowing financial companies to take significant

◗ Page 16

March 2004 ◗ Risk Management

benefit until there is greater convergence on • Solvency corridor floor: a minimum level how it should be measured and managed. of capital Fortis should have. The floor is creat- However, we would argue that there is a very ed to provide an easily understandable and important distinction between netting effects, computable reference point for capital manage- where the same risk can be shown to impact dif- ment. It is derived from the regulatory approach ferent parts of a group in equal and opposite and it can encompass bank and insurance is- ways (e.g., interest rate risk in banking and life sues with specific regulatory and rating con- insurance pools), as well as general diversifica- straints. The Fortis floor for banking is tion. Netting should therefore be analyzed sepa- computed as 4 percent of RWA * 150 percent; rately from more general diversification effects for European insurance it is total capital re- and as we believe, should be recognized in quired * 175 percent. terms of the impact it has on solvency require- Figure 6 ments. SCHEME OF CAPITAL REQUIREMENT Once the correct group-wide capital figure has SCHEME OF CAPITAL REQUIREMENT been computed, it must be re-allocated back to risk types and business lines. However, since II. AGGREGATION III. AGGREGATION IV. DISAGGREGATION I. STANDALONE TO B. LINES OR TO TO THE HOLDING TO BUSINESS LINES OR the group figure will be smaller than the sum of CAPITAL (LEVEL 1) LEGAL ENTITIES (LEVEL 3) TO LEGAL ENTITIES the stand-alone figures, a tailored disaggrega- (LEVEL 2) tion methodology is required. D D Business Risk I I Step 5: Define a regulatory V V Event Risk E E solvency corridor R R S S Credit Risk I I In step 5 the focus is put on the regulatory sol- F F I I vency requirements and the definition of a sol- Market Risk C C vency corridor. Fortis has formulated a A A framework for regulatory solvency that defines Life Risk T T I I an upper and a lower limit of core capital. The P and C Risk O O minimum limit is based on the sum of 6 percent N N of the bank’s risk-weighted assets and 1.75 Diversification includes netting! Group Total times the statutory minimum requirements for the insurance sector. The upper limit comprises 7 percent of the bank’s risk-weighted assets and • Economic capital: the amount of capital 2.5 times the statutory minimum requirements required to cover all the risks faced by a busi- for the insurance industry. ness, analyzed from an economic point of view rather than a regulatory or accounting view. We also note, in addition to the regulatory and Economic capital is calculated in house using economic capital we already have discussed, internal data and methodologies. As a result it that rating agency requirements can not be ig- should be more robust (i.e., reflects the true nored. This therefore leads us to consider four risks in a more tailored fashion) than any views of capital that a financial conglomerate other capital metric. should take into account. •Rating agency driven capital: the amount • Regulatory minimum capital: the amount of capital that the rating agencies expect in of capital to meet the capital adequacy ratio order to feel comfortable about giving a certain stipulated by the regulators to ensure that banks rating. Given the rough rules of thumb used by maintain a certain amount of capital in relation regulators to establish regulatory capital re- to their assets as a cushion against probable quirements and their lacking differentiation for losses. These are currently based on undifferen- the qualitative level of capital adequacy, rating tiated rules of thumb (Basel I, Solvency I) that do agencies have, in some cases (mainly insur- not reflect the real economic risks of the busi- ance), developed their own capital models. One ness, but Basel II and Solvency II have the in- tention to change this to a certain extent.

continued on page 18 ◗

◗ Page 17 Risk Management ◗ March 2004

Financial Conglomerate bonds, “atypical” investments (CDO and other ◗ continued from page 17 structured products) and dynamic hedging strategies in such a framework. The objective is always to push the efficient frontier to the also needs to keep in mind that the rating agen- “Northwest” where you’ll get more return with cies’ decisions on a credit rating are not only less risk. based on quantitative considerations or hard factors but also based on qualitative factors, such as Figure 7 is an illustration of an efficient frontier risk control and management capabilities. analysis for one particular block of group life business. The context is value-based, where re- Step 6: Look at the risk/return turn (Y-axis) is associated with the expected “Framework” increase in value over one year, and risk (X- axis) is defined as the ALM economic capital of In step 6 we have to look at the risk/ that block of business. ALM economic capital return "framework." The accounting view is can be seen as a multiplier times the volatility focused on return-on-assets (ROA) and return- of the changes of fair value over a one-year on-equity (ROE). The regulatory view (Basel I, horizon. The figure below ALM economic capi- Basel II, Solvency II, etc.) is working with re- tal is expressed as a percentage of the turn-on-required-equity (RORE). The risk underlying technical provisions. This is to com- manager view uses concepts such as risk-ad- pare the economic capital requirements with justed return on capital (RAROC). These met- those used in rating agency models, by regulato- rics measure both the return and the capital ry bodies and risk-based solvency frameworks. required on a risk-adjusted, i.e., economic basis, and hence can be viewed as the economic Figure 7 shows, for this particular product group equivalent of the accounting-based ROE prof- and for a fixed percentage of equities in the asset itability measure. mix, that by increasing the duration of the fixed income portfolio we move to the Northwest (less For the insurance operations the risk-return risk more expected return) up to a certain dura- trade-offs are analyzed in the dynamic ALM Figure 7 tion. From there, increasing the duration leads models. In such a model one can test different to more expected return and more risk. If we increase the percentage of equity invest- ALTERNATIVE INVESTMENT STRATEGIES FOR GROUP LIFE PRODUCT ments in the asset mix, we generally increase both expected return and risk (move to the Change in Fair Value Profit sharing based on book yield underlying bonds (1 yr period) Northeast). Internal studies within Fortis show 4,500 40% equities that the shape of the efficient frontiers depends

4,000 30% equities very much on the underlying interest rate

3,500 position (asset minus liabilities) in the product 20% equities 3,000 group. For this group life product the “optimal”

2,500 amount of equities in the asset mix depends on 0% equities 2,000 the risk appetite of the companies selling the

1,500 product and, in practice, also on the competitive Increasing bond 1,000 duration pressures in the local insurance market.

500 Within Fortis, the application of these different 00 0 .0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% 35.0% steps to fix the performance measurement is -500 summarized in the two following schemes. -1,000 Economic Capital consumed by ALM / Technical For the bank pool: • Risk adjusted return = revenue - expenses - EL + capital benefit. asset mixes via a comparison of return and risk • Economic capital is fixed separately in both an earnings and a value-based context. for credit, ALM, trading and The traditional asset classes in such a frame- operational risks. work are equities, bonds and real estate. A major challenge is to incorporate corporate

◗ Page 18

March 2004 ◗ Risk Management

Figure 8 For the insurance companies: • Return = premiums + investment income WIDE VARIETY OF APPLICATIONS + release of reserves - claims - expenses + capital benefit. • Economic capital is fixed separately for credit risk, ALM risk, operational risk, life and non-life.

3. HOW TO PRIORITIZE THE BUSINESS APPLICATIONS Leading banks and insurance companies deploy portfolio management, economic capital and RAROC in a wide variety of applications (see Figure 8).

“Top-Down” Applications: the group will monitor risks as they are assessed at the portfolio level.

1. Reserve and capital adequacy testing The financial system has witnessed consider- able economic turbulence over the last five Figure 9 years. While these conditions have generally not been focused on G-10 countries directly, the THE THEDIFFERENT DIFFERENT APPR APPROACHESOACHES TOTO RISK RISK MANAGEMENT MANAGEMENT risks that financial conglomerates have had to deal with have become more complex and chal- lenging. Financial institutions should frequent- ‘Group Risk ‘Group Risk/ ‘Group Risk Quality ‘Group Risk ‘Group Risk Return ly test and monitor reserves and capital Reporting’ Control’ Monitoring’ Management’ Management’ adequacy, and within Fortis significant re- Operating Operating Group risk and Group risk and Group risk decides sources are put in place in order to measure cap- • companies decide • companies decide • operating • operating • risk appetite their risk appetite their risk appetite companies decide companies decide overall for each ital adequacy from different points of view. Operating Operating risk appetite (at risk appetite (at operating company • companies • companies business line level) sector/risk factor Group risk measure their risks measure their risks as part of annual level) on a dynamic • measures all risks 2. Limit setting how (if) they want Group risk checks planning process basis Group risk decides Group risk • the quality of the Operating Group risk • • • what to do to • aggregates the risk measures, and companies measures all risks Counterparty exposure limits are set to con- change the results issues suggestions measure their risks Group risk • risk/return profile strain the maximum impact of any single default Operating for improvement Group risk analyses the results • Operating • companies decide Operating analyzes the results and makes • companies execute on the capital base of a financial conglomerate. how (if) to manage • companies decide and raises issues recommendations group risk’s risk/return how to manage Operatiing Operating decisions Portfolio risk models allow the calculation of risk/return • companies decide • companies decide how to manage how to maximise the risk contribution of individual counterpar- risk/return, and how return given their to address any risk limits ties or subportfolios taking into account the issues raised by (un)expected losses, correlation effects and group risk thus the economic capital. If risk contributions of certain counterparties are high, senior management could decide to set limits for approval means of secondary loan market, syndicated of additional credits to these counterparties. In lending, credit derivatives and asset-backed a financial conglomerate it is important to apply securities such as CLOs (collateralized the “one obligor” principle which implies that loan obligations). one global vision of all risks on one obligor throughout all entities (no matter the location) “Bottom-Up”Applications: local businesses and risk types (no matter the nature of the un- develop and recommend methodologies of risks derlying risk) should be taken into account. as they are assessed at the individual asset level. 3. Portfolio optimization: buy/sell/ hedge decisions The portfolio managers can optimize the portfolio by using buy, sell or hedge strategies by continued on page 20 ◗

Page 19 ◗ Risk Management ◗ March 2004

Financial Conglomerate Strategic decisions concerning the relative bal- ◗ continued from page 19 ance between corporate and retail banking activities can achieve long-term structural shifts in interest rate risk exposures as well. However, 4. Relationship performance there are limits on how many banking book ex- measurement posures can be transferred to the trading book. Financial institutions have to adapt their When interest rate risk is transferred to the trad- organization and their incentive systems in ing book, usually through transactions that re- order to be successful in the future. semble money market transactions, internal Management must have the incentive to use risk transfer pricing mechanisms are used to deter- information to support better decision making. mine the amount of risk that has shifted between The performance of the relationship of a client books. These pricing mechanisms are highly in- or relationship manager should not solely be stitution-specific. In addition, these mecha- evaluated on revenue and revenue growth rates. nisms do not transfer embedded options and The recognition of capital utilization and return basis risk. on capital are also important.

5. Risk-Based Pricing 4. FORTIS RISK MANAGEMENT Rarely do prices consistently reflect risk. Risk STRUCTURE measurement techniques, in credits for exam- In order to organize an adequate risk manage- ple, can be applied to analyze and price transac- ment structure, the link between central risk tions against the expected loss and required management and local risk management (with- It“ is clear that the economic capital. On the one hand, the narrow- in operating companies) should be clearly de- ing profitability of traditional credit products fined. From this point of view, the following definition of regulatory implies little room for error either in selecting or question arises: Who is in the driver’s seat in the in pricing individual transactions. On the other capital differs greatly measurement and management of the risks and hand, the relative attractiveness of other less returns of each of the activities at a stand-alone traditional but higher margin credit businesses, and aggregated level? between banking and such as project or trade finance can only be eval- insurance environments. uated by taking into account not only their mar- Although the answer to this question will be gins but also their potential impact on the risk of partly influenced by the corporate governance the portfolio. Although the use of internal credit of Fortis, there are two basic principles that will rating models to support the pricing and classi- always hold: fication on a masterscale is a step in the right di- ” rection, it is not sufficient. It is also important to 1) Whether you are at the helm of a bancassur- look at a portfolio level because diversification ance group or a financial holding (with stakes in and timing effects increasingly lead to the dif- banks, life or P&C insurance companies), you ference between profit and loss. must rely on an integrated risk-management framework throughout the whole organization 6. Transfer Pricing (consistent risk-measurement techniques, Transfer pricing, or the price at which one unit of consistent policies: What is my real profile? a firm sells goods or services to another unit of What is the impact of my asset mix on my the same firm, should truly reflect arm’s-length risk-return? How do I monitor and control risk)? prices or the prices at which a willing buyer and a willing unrelated seller would freely agree to 2) The legal structure may evolve over time transact. Banks, for example, use risk manage- (from one bank and many insurance companies ment tools to transfer banking book exposures to to one bank and one insurance holding or even the trading book where possible in order to to one company). It does not matter from a risk hedge interest rate risks internally. For insur- point of view because we have based our risk or- ance companies, basically a comparable ap- ganization structure on the principle of proach is used via replicating portfolios. Unlike "Russian dolls" (from the bottom to the top: banks, life insurance company liabilities are in- business risk committees; central risk commit- tertwined with assets, but this should not pre- tee(s) for the bank and the insurance(s); the vent the company from tracking the Fortis Risk Committee at group level). performance of assets and liabilities.

◗ Page 20

March 2004 ◗ Risk Management

Although we advocate an integrated risk struc- Figure 10 ture, it is up to the financial conglomerates to choose between a centralized or decentralized approach. In Figure 9 we describe the different FORFORTISTIS RISK RISK MANA MANAGEMENTGEMENT STRUCTURE STRUCTURE approaches on how you could organize your structure. Fortis is currently applying the “Group Risk Management” approach. Fortis Risk Committee

Group Fortis Figure 10 summarizes the Fortis Group ap- ALCO Central proach in more detail. The risk organizational Insurance Fortis framework was created to ensure coherent deci- Risk OMPC Risk Bank sion making between the business and group Committee CPC level. Over time, Fortis’ banking and insurance Credit Committee Bank or Bank Mngt. operations have developed risk-management Insurance CCC practices, which support local and tactical decision making. The group objective, howev- Fortis AG Network Banking er, is to build group-wide harmonized risk- Fortis ASR reporting and risk-management structures, FB Insurance Merchant Banking Middle offices which not only integrate practices existing at the Fortis, Inc. Committee Private Professional individual banking and insurance level, but Insurance Mgt. Businesses Investors Services also upgrade the overall approach to include state-of-the-art quantitative risk-management techniques. At the group level, a central risk- management function has been created, reporting directly to Fortis’ CFO. At the business 5. BANKING AND INSURANCE level, each business is responsible for manag- CAPITAL: HIGHLIGHTING SOME ing its risks and ensuring that it has in place ex- DIFFERENCES cellent risk management covering the full risk The purpose of an economic capital/solvency taxonomy. This includes acting within the risk project is to arrive at the capital requirements of policies, guidelines and limits, proactively the group based on the risks taken. This basic identifying, monitoring and managing all of its principle is not easy to implement, taking into risks, holding sufficient reserves to cover liabil- account the different definitions of capital (as ities, etc. All these activities are under the over- mentioned above). all coordination of Fortis Central Risk Management, which: Figure 11 shows the fundamental differences on five crucial items between banks and insurance •helps to ensure the group has and can companies. demonstrate that it has consistently high standards of risk management; Following are two examples that show the •encourages risk/return optimization; impact of some of the previous items: •supports the work of the bank and the insurance risk committees and coordinates Example 1: the capital requirements for the implementation of risk initiatives; "A" rated credit risk •provides support to the businesses on • Banking regulation (Basel I) 8 percent risk-related issues; (minimum 4 percent must be Tier 1). •measures economic capital group-wide; • U.S. insurance P&C (NAIC RBC): 0.3 – •validates the risk models developed by 1.0 percent for investment grade credit. the businesses and by the bank’s • EU life insurance: no explicit focus on credit department; credit risk. •coordinates risk communication with regulators, rating agencies, etc., with the It is clear that the definition of regulatory capi- exception of credit risk in the bank, tal differs greatly between banking and which is communicated through insurance environments. One step in the right central credit management; direction consists of the more risk-sensitive re- •measures and monitors the ALM risk in a consistent way, across bank and insurance. continued on page 22 ◗

Page 21 ◗ Risk Management ◗ March 2004

Financial Conglomerate quirements. In order to bridge the gap between ◗ continued from page 21 banking and insurance, additional efforts will have to be made. We describe this in more detail in the next chapter. quirements set by the New Basel Accord. This trend can also be observed in the 6. ECONOMIC CAPITAL, insurance industry (see Solvency II). These COOPERATION BETWEEN trends will most likely bring regulatory require- REGULATORS AND THE NEW ments much closer to economic capital. ROLE OF THE ACTUARIAL PROFESSION Example 2: Another example of a regulatory mismatch is found in the area of financial guar- As noted earlier in this paper, there is a trend to- antees and their counterpart in the insurance ward more risk-based measures and many world—credit insurance. Certain types of guar- major financial conglomerates are already antees are treated as insurance business if writ- adapting economic capital as the consistent ten by insurance but as banking business if measure of risk within the institution. Designed Figure 11 as a management tool, economic capital, in our view, more closely reflects the real risks of the business in terms of asset/liability manage- A GAP BETWEEN BANKS AND INSURANCE COMPANIES ment. Although developed on the banking side, economic capital has more recently been ex- Basel 2 Statutory Reserves Economic Capital tended to insurance activities. Requirements for Insurance Consumed by a for Banks Companies Bancassurer The reorganization of the supervisors is another Confidence A? / BBB ? None Shareholder's development that could help fill the gap. interval decision Further consolidation of financial entities made Base Line Statutory solvency Statutory solvency Economic solvency policymakers realize that more coordination of regulation and supervision was necessary.

Valuation Statutory Statutory Fair value In addition to this, the actuarial profession must Risk type Excludes Business Excludes Event All risk types also be transformed in order to meet the new coverage risk as well as most Risk of the ALM risk (the needs. As Bob Partridge, a managing director in Banking Book) Standard & Poor’s New York office, states, “Everyone’s paying much more attention to ac- Diversification ?* ?** Yes counting and corporate governance issues these

* Market risks are highly correlated with credit risk. It is not the case however for operational risk. days, but the forgotten issue is the actuaries.” ** The existing European insurance capital requirements assume some “average” level of correlation within one licensed entity. In case Traditionally, actuaries focused on technical in- several such entities form part of an insurance group, any additional diversifications (e.g. geographic diversification) are ignored. surance risks such as mortality, disability, P&C claims risks, etc. Actuaries, who focus on adequacy of reserves, should also be involved in the written by banks, yet the capital needed to sup- whole risk taxonomy and the portfolio manage- port the business is radically different depend- ment of assets and liabilities. This implies that ing on which environment is chosen. For a bank, an integration of ALM and the actuarial depart- the same capital has to be held to support a guar- ment is a necessity. Of course this has conse- antee as would have to be held to support a loan quences for the academic actuarial of the maximum amount guaranteed. In an in- curriculum—transition to a curriculum of all- surance context, we look at an actuarial assess- round financial risk manager, which implies the ment of the amount likely to be paid out. What integration of actuarial science, mathematical we can be sure of is that, unlike in the case of finance, econometrics of financial markets, etc. banks, the amount reserved will almost always be less than the worst case. 7. CONCLUSION There is a need for a more rational and adequate These examples illustrate how differences in framework for responding in an appropriate man- the current regulatory framework for banking ner to the issues and opportunities raised by the and insurance can lead to different capital re- convergence of the banking and insurance mod-

◗ Page 22

March 2004 ◗ Risk Management

els. Within this framework, the actuary Appendix will play a crucial role together with other risk managers. THETHE FOUR FOUR “CAPITAL” "CAPITAL" APPROACHESAPPROACHES It is only in this spirit of cooperation and mutual willingness to learn from REGULATORY CAPITAL AGENCY-DRIVEN CAPITAL ECONOMIC CAPITAL ACTUAL CAPITAL

l Amount of capital required l Amount of capital the l Amount of capital required l Amount of equity capital or each other that we will reap the full to protect the group against rating agencies expect in to protect the group against embedded value actually benefits of convergence. Both Basel II statutory insolvency over order to feel comfortable economic insolvency over held to protect the group and Solvency II are important steps to- a one-year timeframe giving fortis an 'AA' rating a one-year time-frame against economic and l Based on undifferentiated l Based on relatively l Reflects real risks taken statutory insolvency over wards that objective—the uniform rules of thumb that do not undifferentiated rules of in the sense of unexpected a one-year time-frame reflect the real economic thumb (bank), and/or movements in the value economic solvency framework. There simple models (insurance) of assets and liabilities risks of the business and l Accounting result; Not formulaic—other are many issues still to be resolved. To usually based on (relatively) l and on the confidence expanded definition public information factors such as quality of interval management includes hidden solve these, we believe that there is a l Designed to protect policy management and likelihood wishes to tolerate reserves of government bail-out Designed to be a tool need for a well-structured interna- holders and creditors l l Acts as a floor, which are also considered for management tional platform allowing for an open triggers takeover by dialogue between the industry (bank- the regulators ing and insurance) and the regulator (e.g., joint forum). BARE MINIMUM CAPITAL CAPITAL YOU YOU MUST HAVE CAPITAL YOU ARE OUGHT TO HAVE CAPITAL YOU It is also important that regulators and EXPECTED TO HAVE ACTUALLY HAVE rating agencies encourage and support banking and insurance companies to measure solvency requirements based on economic capital (no ment and capital adequacy in financial con- fixed rules of thumb). glomerates, Wharton Financial Institutions Center, p. 55. 8. APPENDIX • KPMG. 2002. “Study into the methodologies to assess the overall financial position of an in- See chart on right. surance undertaking from the perspective of prudential regulation,” p. 249. 9. REFERENCES • Lown, S., C. Osler, P.E. Strahan and A. Sufi. 2000. “The Changing Landscape of the • Allen, F. and A.M. Santomero, 2001. “What Financial Services Industry: What Lies do Financial Intermediaries do?” Journal of Ahead?” Federal Reserve Bank of New York Banking & Finance, pp. 271-294. Economic Policy Review, October, pp. 39-55. • Basel Committee on Banking Supervision. • Morrison, A.D. 2002. The economics of 2000. “Report on the Working Group on Capital capital regulation in financial conglomerates, Adequacy - Consultative Paper” p. 63. University of Oxford, p. 17. • Benoist, G. 2002. “Bancassurance: The • NBB. 2002. “Financial conglomerates,” New Challenges,” Geneva Papers on Risk and Financial Stability Review, pp. 61-79. Insurance - Issues and Practice, 27, pp. 295-303. • OWC. 2001. “Study on the risk profile • Bikker, J.A. & I. Van Lelyveld. 2003. and capital adequacy of financial conglomer- “Economic versus regulatory capital for finan- ates,” p. 43. cial conglomerates,” forthcoming in Banking • Poynton, I. 2003. “Value Creation in a Supervision at the Crossroad. Regulated Market” at the Fifth Annual • Dhaene, J. and S. Vanduffel, M. Goovaert , R. European Insurance Industry Conference Olieslagers, R. Koch. 2003. On the computation Insurance Actuaries. of the capital multiplier in the Fortis Credit • Standard & Poor’s. 2003. “Insurance Economic Capital model, p. 21. actuaries – a Crisis of Credibility,” p. 5. • Hall C. 2002. Economic capital: towards an • Van Lelyveld, I. & A. Schilder. 2002. Risk in integrated risk framework, Risk, p. 5. financial conglomerates: management and • Herring, R.J. and T. Schuermann. 2002. supervision, Nov. 2002. p. 25. Capital Regulation for position risk in banks, • Wilmarth, A.E. 2001. How should we securities firms and insurance companies. respond to the growing risks of financial • Kuritzkes, A., T. Schuermann and S.M. conglomerates,? Working Paper. ✦ Weiner. 2002. Risk measurement, risk manage-

Page 23 ◗