The Financial Modeling of Property/Casualty Insurance Companies by Douglas M. Hodes Tony Neghaiwi, FCAS J.

David Cummins Richard Phillips Sholom Feldblum, FCAS, ASA

THE FINANCIAL MODELING OF PROPERTY-CASUALTY INSURANCE

COMPANIES

Douglas M. Hodes, Tony Neghaiwi, J. David Cummins, Richard Phillips, and Sholom Feldblum

Abstract

This paper describes a financial model currently being used by a major U.S. multi-line insurer. The model, which was first developed for solvency monitoring purposes, is now being employed for a variety of internal management purposes, including (i) the allocation of equity to corporate units, thereby allowing measurements of profitability by business segment and by policy year, as well as analysis of the progression of “free surplus,” (ii) the analysis of major risks, such as inflation risks, interest been difficult to rate risks, and reserving risks, that have heretofore quantify, and (iii) consideration of varying scenarios on the company’s financial performance, both of macroeconomic conditions as well as of the insurance environment. This paper begins with the genesis of the model and with its structure. It moves on to equity considerations and to performance measurement. It then discusses the major risks that have heretofore resisted actuarial analysis, such as interest rate risk (inflation risk), reserving risk, and scenario testing. The paper shows how cash flow financial models can deal with global risks that simultaneously affect various aspects of the insurer’s operations, delineating the resulting changes in the company’s performance.

The

Financial

Modeling

of

Property-Casualty
(Authors)

Insurance

Companies

DouglasM. Hades is a Vice President and Corporate Actuary with the Liberty Mutual Insurance Company in Boston, Massachusetts. He oversees the Corporate Actuarial and Corporate Research divisions of the company, and he is responsible for capital allocation, financial modeling, surplus adequacy monitoring, and reserving oversight functions. Mr. Hodes is a graduate of Yale University (1970) and he completed the Advanced Management Program at Harvard University in 1988. He is a Fellow of the Society of Actuaries, a member of the American Academy of Actuaries, a member of the American Academy of Actuaries Life Insurance Risk-Based Capital Task Force, and a former member of the Actuarial Committee of the New York Guaranty Association. Before joining Liberty Mutual, Mr. Hodes was a Vice President in Corporate Actuarial at the Metropolitan Life Insurance Company, where his responsibilities included the development of a life insurance financial model. Mr. Hodes is the author of “Interest Rate Risk and Capital Requirements for Property-Casualty Insurance Companies” (with Mr. Sholom Feldblum) and of ‘Workers’ Compensation Reserve Uncertainty” (with Dr. Gary Blumsohn and Mr. Feldblum). These papers apply actuarial and financial techniques to quantify risks associated with interest rate movements and with unexpected reserve developments. In addition, Mr. Hodes is a frequent speaker at actuarial conventions on such topics as dynamic financial analysis and risk-based capital.

Tony Neghaiwi, FCAS, is an Associate Actuary with the Liberty Mutual Insurance Company, where he is responsible for the development and the maintenance of the model described in this paper.

Dr. J. David Cummins is the Harry J. Loman Professor of Insurance and Risk Management and Executive Director of the S. S. Huebner Foundation for Insurance Education at the Wharton School of the University of Pennsylvania. Dr. Cummins’ primary research interests are the financial management of insurance companies, the economics of insurance markets, and insurance rate of return and solvency regulation. He is the editor of the Journal of Risk and insurance and past president of the American Risk and fnsurance Association and the Risk Theory Society. Dr. Cummins has written or edited fourteen books and published more than forty journal articles in publications such as the Journal of Finance, Management Science, the Journal of Banking and Finance, the Journal of Economic Perspectives, the Journal of Risk and Uncertainty, and the Astin Bulletin. Among his recent publications are “Insolvency Experience, Risk-Based Capital, and Prompt Corrective Action in Property-Liability Insurance,” Journal of t3anking and Finance (1995); “Pricing Insurance Catastrophe Futures and Call Spreads: An Arbitrage Approach,” Journal of fixed income (1995); and ‘Capital Structure and the Cost of

Mathematics and Economics Equity Capital in Property-Liability Insurance,” Insurance: (1994). His paper, “An Asian Option Approach to Pricing Insurance Futures Contracts,” was awarded the Best Paper prize at the 1995 AFIR Colloquium in Brussels, Belgium. Dr. Cummins has served as consultant to numerous business and governmental organizations. He has consulted and testified on the cost of capital in insurance for organizations such as the National Council on Compensation Insurance and Liberty Mutual Insurance Group. He has conducted research on insurance cycles and crises for the National Association of Insurance Commissioners, and he has testified for several state departments of insurance and the U.S. Department of Justice. He advised the Alliance of American Insurers on risk-based capital in property-liability insurance.

Dr. Richard 0. Phillips is an Assistant Professor of Risk Management and Insurance at Georgia State University. He received his Ph.D. in Finance and Insurance in 1994 from the Wharton School, University of Pennsylvania. Dr. Phillips’s recent publications on topics related to insurance company financial modeling include “Financial Pricing of Insurance in the Multiple-Line Insurance Company” (with J. David Cummins) and “The Economics of Risk and Insurance.” From March 1992 through January 1994, Dr. Phillips was a principle investigator in the Alliance of American Insurers project to develop a risk-based capital cash flow solvency model, which used an earlier version of the model described in this paper. Dr. Phillips is a frequent speaker at research seminars on such topics as cash flow modeling, financial pricing, and insurance regulation.

Shalom Feldblum is an Assistant Vice President and Associate Actuary with the Liberty Mutual Insurance Company. He holds the FCAS, CPCU, ASA, and MAAA designations, and he is a member of the CAS Syllabus Committee and of the American Academy of Actuaries Task Force on RiskBased Capital. He is the author of numerous papers on ratemaking, loss reserving, statutory insurance economics, competitive strategy, investment theory, solvency accounting, monitoring, and finance, which have appeared in Best’s Review, the CPCU Journal, the Proceedings of the Casualty Actuarial Society, the Acfuarial Digest, the CAS Forum, the Journal of insurance Regulation, the Journal of Reinsurance and the CAS Discussion Paper Program. He was the recipient of the CAS Michelbacher Prize in 1993 for his paper on “Professional Ethics and the Actuary.” In addition to the two papers co-authored with Douglas Hodes (see above), Mr. Feldblum is the author of “European Approaches to Insurance Solvency,” which describes the British and Finnish foundations upon which the model described in this paper is based, and “Forecasting the Future: Stochastic Simulation and Scenario Testing,” which describes the use of scenario testing in financial models.

6

however. z Structure: that is. interest rate risks. this complexity renders cumbersome the documentation of the models. the factors that stimulated the development of the model. the types of underwriting and financial operations and the types of time periods with which it deals. along with the progression of LOB surplus and of free surplus. and graphs that are produced by the model. including (i) the allocation of equity to corporate units.” (ii) the analysis of major risks. thereby allowing measurements of profitability by business segment and by policy year. describes a financial model currently being used by a major U. Since this paper is not just a theoretical discourse but also a practical description of a working model. . such as recessions. such as underwriting cycle movements. In the three years since then. Risks that results from an overall change in the external economic environment. multi-line insurer. as well as analysis of the progression of “free surplus. how profitability is measured. and several types of tables. both of macroeconomic conditions as welt as of the insurance environment. or from changes in the insurance industry as a whole.S. The sophistication of complex stochastic simulation is extolled. This complexity is the major stimulus for financial models that consider the workings of the entire corporation. c Equity considerations: how net worth (“economic surplus”) is determined by line of business (LOB) and how the progression of “free surplus” is viewed. it shows the actual inputs and outputs: what variables must be provided by the user. This paper. This paper discusses the following topics: rc Genesis: that is. At times. the numerical exhibits are contained in the appendices. Many multiline insurance enterprises are complex organizations. * Profitability measures: given the actual (past) or expected (future) cash flows. To facilitate the readability of this paper. the model has been greatly expanded and it has been applied to a variety of internal management uses. The first version of the model was developed in 1993 for solvency monitoring purposes. The financial model described here is particularly that are not easily analyzed by other methods: * * important for evaluating three types of risk Risks that simultaneously affect several components of an insurance company’s operations. such as inflation risks. with dozens of distinct yet interrelated parts.THE FINANCIAL MODELING OF PROPERTY-CASUALTY INSURANCE COMPANIES Introduction The existing literature on the financial modeling of property-casualty insurance companies consists predominantly of theoretical discourses seen through the eyes of the research actuary. charts. that have heretofore been difficult to quantify. and reserving risks. in contrast. the practical implementation of the models is rarely considered. such as inflation risks and interest rate risks. and (iii) consideration of varying scenarios on the company’s financial performance. so that the text flows more easily.

which affects both assets and liabilities. 8 . Interest rate risk. For example. The model described in this paper is the practical implementation of the AAA vision: it shows the cash flows of the company under varying future scenarios. Similar analyses should be undertaken for the underwriting risk of new business (“written premium risk” in RBC terminology) and for the risks of reinsurance collectibility. who discusses the “measurement bias” introduced when GAAP or statutory accounting statements are used for solvency monitoring purposes. C. the NAIC developed new risk-based capital requirements both property-casualty and life insurance companies. random fluctuations. Many observers have criticized NAIC efforts from three perspectives: for the A The risk-based capital formulas are based on accounting figures. and (ii) it is intertwined with other risks. @ Meanwhile.” should be replaced by rigorous actuarial analyses of reserve variability. such as inflation risks and reserving risks. Sutsic. should be incorporated into the formula. Some of the RBC charges seem to be “ad hoc” factors lacking actuarial or financial justification. “Solvency Measurement for Property-Liability Risk-Based Capital Applications. Several important risks are not even considered. Volume 61. which are based on the NAIC “worst case year” method coupled with a large dose of “regulatory judgment. B. This paper shows how the financial model deals with these types of risk Genesis of the Model The company’s modeling efforts were stimulated by several developments: 0 From 1990 through 1993. pages 656-690. t Compare especially Robert P. these critics have said that A The statutory financial statements that underlie the risk-based capital formulas should be replaced by cash flow approaches or by market value accounting. both for solvency monitoring by state regulators and for management evaluation of the company’s performance.” Journal of Risk and Insurance. such as reserving risks. Interest rate risk is particularly difficult to model in the NAIC formula. The reserving risk charges in the NAIC formula. covering not just opinions on the reasonableness of loss and loss adjustment expense reserves but also statements on the financial strength of the insurance enterprise under varying longer term scenarios and on the resilience of the company to different types of adverse external conditions.* Risks that depend on complex. Number 4 (December 1994). not an accounting phenomenon.1 B. C. since (i) it is a market value phenomenon. the American Academy of Actuaries has proposed an expanded vision of the Appointed Actuary’s role.

D. pages 87-149.8) Soon after this model was implemented. The required work is primarily accounting. 1994). Chris D. not actuarial. as would be needed for pro-forma financial statements. Devitt. E. E. and M. D. Financial Mode/s of hsurance Solvency (Boston: Kluwer Academic Publishers. B. following the method developed by the British Solvency Working Party in the 1980’s? The cash flow results are particularly important for Appointed Actuary work and for comparing the effects of different scenarios. It required a sophisticated management model. Hey. D. F. 0. Part 2 (1987). G. “Managing Uncertainty in a General Insurance Company. Practical Risk Theory for Actuariss (Chapman and Hall. No. M. and P. 467 (September 1990). the model can generate statutory accounting results. Bernstein. “Assessing the Solvency and Financial Strength of a General insurance Company. Volume 117. Daykin. and acquire other (related) businesses.” in J. 9 . Reynolds. the translation of net cash flows and market values into statutory values is not shown in the exhibits in this paper. pages 173-259. Volume 114. G. F. Devitt. pages 227-310. D. and P. 3 For reasons of space. Daykin. The recent text by Chris D. whereas the model described here uses stochastic procedures for risks that are random and “scenario building” for global risks with interdependent elements. Teivo Pentiktiinen. because of both strong industry profits in its major lines of business and its own favorable performance relative to its peer companies.” Journal of the hsfitute of Acfuaries. and it is not germane to the theoretical framework of the model. I. combines the cash flow approach of the British Solvency Working Party and the accounting approach of the Finnish Working Party. the analyst can determine market values of the insurance enterprise at various points in time B b 2 For a more complete presentation of the British Solvency Working Party approach. B. R. the authors changed their emphasis from solvency monitoring to profitability measurement. D. “The Solvency of a General Insurance Company in Terms of Emerging Costs. develop new products.” Journal of the institute of Actuaries. Coutts. in order to judge both the immediate risks and the long-term uncertainties associated with the new projects. Coutts. Hey. Part 2. Ft. M. These results are useful for analyzing the progression of “free surplus”3 By selecting appropriate discount rates for loss outflows and for investment inflows. Pesonen. Description of the Model The financial model described here provides three types of results: 0 The model itself uses a cash-flow approach. S. 1 (1987). Chris D. No. When insurance companies fare poorly. David Cummins and Richard Derrig. Volume 117. Hey. Smith. or in AST/N Bulletin. It elected to expand into new markets. B. the multi-line insurer using this model fared extremely well. Bernstein. see Chris D. 1989). In addition. W. I. Daykin and G. G. that textbook emphasizes stochastic procedures to develop scenarios. G. as well as the capital needed to safely undertake them. Reynolds. For management purposes. S. Statutory accounting is an important constraint on insurance company strategy. financial models are important for monitoring solvency. pages 85-132. In the early 1990’s. W. Smith. Daykin.

and expenses for practice.4 @ About half of the company’s workers’ compensation business is written on loss sensitive contracts. The anticipated loss and LAE ratios and the anticipated underwriting expense ratios are actuarial projections.0%. The premium payment patterns extend for about ten years after the policy expires. as shown in the exhibits. Two further adjustments are made: 0 The company has large investments in mortgage-backed securities.or under various scenarios. written premium by line of business is taken from the business plans. [The model allows for separate policy year. For instance. which are used in actual work. The assumed stock dividend yield is 2. see Douglas M. of course. existing and of new business. . 10 .1% per annum.0% per annum. Base Case and Alternative Scenarios model. each line of business. .3% per annum. For future years’ operations.75% per annum. and the rate of growth in stock values is 8. Similar adjustments are used for other options. the model uses actual company results. with high prepayments as borrowers change homes or simply refinance their mortgages when interest rates are low.0% and growth in real flows of 6. the cash flows are based on a combination of company business plans and actuarial projections. and the effects are shown in the exhibits.] 4. chain ladder paid loss development for the run-off of existing reserves. losses. “Interest Rate Risk and Capital Requirements for PropertyCasualty Insurance Companies” (CAS Part 10 examination study note). Hodes and Sholom Feldblum. In vary by line. such as call provisions in corporate bonds. 0 The base case scenario assumes an annual inflation rate of exposures of 2. for a nominal growth in underwriting cash These assumptions affect premiums. Past and Future Business These results are important for determining profitability of For past business.75%. stated coupon rates for fixed income securities. and expected dividend yields on common stocks for investment returns. The expected cash flows are adjusted in each scenario for these options. providing an annual return on common stocks of 10. along with . the assumed growth in real exposures will the company’s business plans. two scenarios are shown in the exhibits and To illustrate the power of the financial discussed in the text. These figures are combined with the payment and collection patterns developed from past business to model the cash flows from new business. depending on assumptions by line and by The average pre-tax yield on the bond portfolio held by the company is 8. 4 These expected cash inflows are similar to those required in the new NAIC risk-based capital “supplementary asset schedule” used to measure interest rate risk.

and combined ratios.” American Economic Review.3% * (l-35%).7%.] The growth in real exposures remains 2. losses. For each line of business. the nominal losses grow more quickly than premiums. though the theoretical explanations vary from author to author.The federal income tax rate is 35%. To parameterize our model. and expenses. graphs for ten years of new writings as well. [The market value of existing fixed-income securities. Fama and G. the model uses an after tax discount rate of 5. 5 This decline. is fully recovered in the subsequent two years. for example. of course. the new rates affect premiums. 6 See. however.” Journal of Financial Economics. with the magnitude of the effect depending on the duration of the fixed income portfolio. text discusses a single year of new business (policy year 1995). under each of the two scenarios.0%.7%. I’Inflation and the Stock Market.” Other analysts have replicated the Fama and Schwert results. B.] The expected after-tax yield on the company’s investment portfolio is about 5. common stock prices also drop when interest rates increase. though the magnitudes of the coefficients were dampened. The immediate effect on underwriting and investment results is twofold: A. reflecting the higher returns on the common stocks. as well as of mortgage-backed securities. [5. Market values of bonds and of mortgages. The signs of the coefficients in our analysis were generally consistent with the signs found by Fama and Schwert. tall when interest rates rise. for new business. incurred loss plus loss adjustment expense. The after tax discount rate for insurance cash The summary assumptions for the base case scenario and for the alternative scenario are shown on Exhibit 1 of Appendix A. For the “progression also show the anticipated 1996 written premiums.1%. “Asset Returns and Inflation. Fama and Schwert’s paper uses data which is now 20 years “out-of-date. The exhibits shows ten years For clarity of exposition. falls when the interest rate rises. however. flows now changes to 6. and other underwriting expenses. Since income taxes are explicitly included in the cash flows. the though we show exhibits and of free surplus. Volume 5 (1977).” the exhibits of new business. see.0% with a concomitant increase in the pre-tax bond yield on new investments to 10. pages 839-487.4% to determine the present values of insurance operations (when present values are used in the analyses). since the simultaneous rise in inflation and interest rates causes no change in the real equity value of corporations. When rates first increase. for instance. for a nominal growth in underwriting cash flows of 8. Volume 70 (December 1980). we replicated the Fama and Schwert study using the most recent 20 years of data from the lbbotson and Sinquefield indices. 11 . Eugene F. leading to an initial increase in the loss ratios. as would be used in a “going-concern” valuation. as noted by many investment economists. expense ratios. Martin Feldstein. @ The alternative scenario assumes that the inflation rate increases by 200 basis points to 6. pages 115-146. William Schwert.3%.4% is 8. The federal income tax rate remains 35%. as well as the loss ratios. Initially. Exhibits 2 through 5 of Appendix A show the projected written premium.

the model is substantially more complex than is described here. 0 scenario” has two effects on the market value Since the payment obligations are in fixed dollar terms. This effect occurs even if interest rates do not change. A second portion of the decline in value. causing a further decline Similar analyses are performed for each class of securities. and Lamont. respectively). the market value of these assets decline. refinancings become less frequent. Exhibit 1 of Appendix B shows graphically the effect of a 2% rise in interest rate on the market value of mortgage backed securities. 8 The rise in interest rates under the “alternative of mortgage backed securities. 1986 and post August 6. 7 Because of federal income tax provisions.. discount @ When interest rates rise. become less frequent. when interest rates rise. stems from the higher discount rate. L. 1990). D’Ambrosio’s chapter in J. municipal bonds). L.g. preAugust 6. stems from the fewer prepayments. It is dependent on interest rate changes to the extent that real estate purchases depend on the availability of “affordable mortgages”6 When interest rates decline. 6 On the effect of interest rates on real estate values. the model then securities. Fixed income investments are divided between taxables and tax-exempts (e. incorporates the effects of options. 12 . from the third bar to the middle bar. and the latter category is further subdivided by date of acquisition (i.Asset Returns each group of securities. before consideration of issuer options. Exhibits 4 and 5 of Appendix B shows the overall effects of a 2% rise in interest rates on the entire fixed income portfolio. Second Edition (Warren. The difference is particularly great for securities (which form significant proportions of insurance investment portfolios. Tuttle (editors). 1986). see Charles A. but the appropriate rate rises. such as calls on corporate bonds and pre-payment options on mortgages. Conversely.. from the middle bar to the top bar. they pre-pay the mortgages. many homeowners refinance their mortgages. The dominant portion of the decline in value. Gorham. The model requires cash flows by type of security for each scenario. not the stated mortgages and for mortgage backed life insurance and property-casualty reasons: The financial model uses the expected cash flows from cash flows. for two 0 As borrowers move to different homes. Maginn and D. One begins with the stated cash flows from each category of For each scenario. The effect is divided into two pieces. refinancings in the market value of the assets. Managing investment Portfolio: A Dynamic Process.7 Exhibits 2 and 3 of Appendix B show graphically the effects of a 2% rise in interest rates on the cash flows and remaining balances from mortgage backed securities.e.

and for the base scenario versus the alternative scenario. The casualty liabilities do not have the complication of policyholder options. investment inflows. Since federal income taxes have a great effect on net income. Exhibits 1 and 2 of Appendix C show the cash flows in future years from the run-off of existing workers’ compensation business under each of the two scenarios mentioned above. showing whether the income of the company from its investments will These cash flow suffice to meet benefit obligation under sever future interest scenarios. loss outflows. are combined. with the appropriate tax rates applied to each. and expense outflows. and they are incorporated as inputs into the model. such as premium inflows. For instance. See below in the text for the manner in which these are handled. compensation make a chain ladder paid loss development reserving procedure accurate for this line of business.” signed by a Fellow of the Society of Actuaries. Exhibits 5 through 10 show the combined cash flows for assets versus liabilities for 35 years. Exhibit 11 shows the loss and loss adjustment expense cash flows for the two scenarios. since the cash outflow patterns for loss reserves are relatively stable. [Exhibits 3 and 4 show the individual cash flows separately for each component of the company for two years. Combining Operations over other analyses of an insurance The model described here provides numerous advantages company’s operations: 0 The different components of the company. but they have other complexities. for the pure run-off case versus the run-off with one additional policy year. Since the company uses primarily paid loss retros for its large accounts. for the pure run-off case versus the run-off with one additional policy year.The model performs this analysis for each type of security. For the property-casualty model. Since the company has a large life insurance subsidiary. projections must take into account both issuer options on the asset side and policyholder options on the liability side. 13 . The dominance of retrospectively rated contracts in this company’s workers’ compensation book of business causes retrospective premium payments for about ten years past the expiration date of the policies. The expected cash flows from assets are projected by the company’s Chief Investment Officer and his staff. such as inflation sensitivity and dependence on macroeconomic conditions.“s Existing 0 vs New Business Equally important as the combination of the different components of the company is the differentiafion of blocks of business. the model separates (i) the cash inflows and outflows from the run-off of the current business from (ii) the cash inflows and a Life insurance companies must have an “asset adequacy analysis. and for the base scenario versus the alternative scenario. we use the same cash flow projections on the asset side. a cash flow approach is most useful.] Workers’ compensation provides a particularly good example of the model’s operations. In particular. these cash flow projections under different interest rate scenarios are needed for the “asset adequacy analysis. so The statutory benefits in workers’ we use this example repeatedly in the paper. the analysis is performed separately for taxable versus tax-exempt bonds.

projecting future inflation rates that are tied to the assumed future interest rates (which affect the asset values). policy provisions. and what is the implied change in profitability for the run-off of existing business? 2A. since most of the premium has already been collected whereas most of the losses remain to be paid. with a This is interest rate consider the issue 6 The more sophisticated analysis described below in this paper quantifies more carefully how loss liabilities are affected by inflationary changes. Exhibits 5 and 6 of Appendix C shows the investment and the net liability cash flows for the base case scenario for (i) the run-off of existing business and (ii) the run-off of existing business plus one year of new business. Previous analyses often asked: “Is the company’s workers’ compensation book of business profitable?” This question is not just simplistic.outflows resulting from new business. and it demonstrates the power of the financial model. In the former case. let us conceptually. not business already earned. and premium rates. For example. however. the net liability cash outflow is small.lf external conditions change. the future earning of premium and accrual of losses from the most recent policy year is included in the “run-off” section. All the exhibits in this paper differentiate between the effects on the existing business and the effects on new business. For instance. Rather. 1 B. or from additional policy years of business? As noted above.lf external conditions change. as we have in the two scenarios. Thus. which the financial model quantifies. risk. the net liability cash outflows greatly exceed the investment income cash inflows in the first subsequent calendar year (1995 in the exhibits). What are the net cash flows from the run-off of the existing book of business? We are concerned with business already written. by calculating “real dollar” link ratios for the chain ladder loss development procedure. since the premiums from the new business nearly equal the total loss and expense payments in that year. the actuary must ask two sets of questions: 1A. how would the net cash flows change. External changes have vastly different effects on the run-off of existing profitability of new business. what are the expected net cash flows from the new business?” 2B. it misses the point. In the latter case. how would the net cash flows from the new business change. and what is the implied change in profitability of this business? The difference between questions 2A and 2B is crucial for the valuation analysis. The present question ties the valuation analysis to the current underwriting procedures. as well as the earnings from the unexpired portions of policies already written and the expected loss accruals from these policies. this “runoff” includes both l l the future (retrospective) premium collections from exposures already earned and payments for losses already incurred. an important external property-casualty insurance company profitability is a movement concomitant movement in interest rates. We are asking: “Based on the company’s current business plans and our actuarial projections.What are the net cash flows from an additional policy year of business.9 business versus the change that affects in inflation. determining the “inflation 14 . First.

Workers’ compensation benefit payments consist of indemnity (“wage-loss”) benefits and medical benefits.We must consider the effects on the run-off of previously written business versus the effects on new business. COLA adjustments make certain indemnity benefits inflation sensitive as well. would severely sensitivity” of each reserve component (such as workers’ compensation medical benefits versus workers’ compensation indemnity benefits). reserves paid out one year hence will rise in nominal value by 2%. In about half the US. A 2% rise in inflation. Given the steady benefit outflows in workers’ compensation. a common asset liability management strategy would call for high grade corporate bonds or mortgage backed securities with an average maturity of ten years or longer. [Permanent total claims. [As noted earlier. Since medical benefits are paid more quickly. For simplicity. the true effects are easily seen. or “lifetime pension” cases.] Because of the high retention levels of workers’ compensation business and the generally upward sloping yield curves. a 2% rise in inflation causes nominal loss costs on previously written business to rise by about 1%. Generally. the increase in nominal value of reserves exceeds 2%. The explanation in the text. tc The textual explanation given here over-simplifies the effects. most companies will choose asset portfolios with longer maturities than the average maturity of the loss reserves. generally about 7 to 8 years. with a corresponding 2% rise in interest rates. 6. is simplified. reserves paid out in two years’ time will rise by about 4%. of course. jurisdictions. however. Since the average payment date of the reserves exceeds one year. medical benefits 3. 15 . A loss paid three years after the valuation date would increase by about 3%. The actual model. 2. the COLA adjustments in these jurisdictions apply only to long-term indemnity benefits. Workers’ compensation loss reserves have a long average payment date.] 5. Medical benefits are fully inflation sensitive. 4. thereby greatly lengthening the duration of reserves compared to the duration of incurred losses. (in nominal terms) will be 2% higher. such as 5% per annum. to highlight the differences between existing business and new business. The “50%” inflation sensitivity is used in this explanation for heuristic purposes only. Incurred losses are about 55-60% indemnity and 40-45%‘medical. That is. Since the financial model tracks the cash flows. let us assume that overall workers’ compensation reserves are 50% inflation sensitive. and the adjustments are often capped at a relatively low amount. We have the following characteristics of workers’ compensation business: Run-Off Valuation and Interest Rate Risk 1. separately quantifies indemnity and medical workers’ compensation benefits and their respective cash flows. and then calculating the cash outflows each year. and so forth. the reserves are about 65-75% indemnity and 2535% medical. the 1% rise applies to losses paid one year after the valuation date.10 If inflation increases by 2%. In other words. form a higher proportion of reserves than they do of incurred losses. such as benefits two years or more after the accident.

and interest rates show a corresponding 2% jump. higher for young and inexperienced workers. firms hire young and Workers’ compensation accident workers. since their nominal value rises (as the liabilities are 50% inflation sensitive) and their duration is shorter. firms lay off recently hired and inexperienced workers. The change in timing of loss payments and tax payments In addition. In sum. leading to no net change.11 2.depress the market values of these corporate bonds or mortgage backed securities. Similarly. For simplicity. A recession has two effects on workers’ compensation benefit costs 0 During recessions. work decreases. There is almost no change to the expected value of new business from interest rate risk. Consider the effects of a recession on a workers’ compensation carrier. Discounting at the new interest rate shows the loss resulting from interest rate risk. Both medical and indemnity losses will be 2% higher. 16 . since the cash inflows from the existing bond portfolio do not change in nominal terms. Alternatively. the difference between the slightly affect the results for the alternative scenario. suppose that inflation increases by 2% just before the new business is written. particularly when they are and overtime inexperienced frequency is working long workers’ of new 11 Since this is new business. The model shows the cash flows from assets and for benefit payments. the financial model shows this explicitly. but since the discount rate is also 2% higher. Again. 1. interest rate risk has a great effect on the run-off of existing compensation business. whereas the cash outflows from the reserve portfolio increase in nominal terms. their economic value does not change. There is a 2% increase in the cash inflows from assets and a concomitant 2% increase in cash outflows for benefits. their economic value does not change. Conversely. the assets purchased with the newly collected premiums either have higher coupons (for newly issued bonds) or higher yields to maturity (for bonds bought in the secondary market). increase in workers’ compensation benefits and the assume increase in inflation slightly increases the expected profitability. during prosperous years. Recessions There is no reason to assume that changes in the external environment affect only the valuation of run-off business but not the value of new business. the coupon rate on newly issued bonds will be 2% higher. 7. and overtime work increases. The market value of the liabilities would decline by a lower amount. but little effect (if any) on the expected profitability business. New Business and Interest Rate Risk The situation is entirely different for the new business. 3. The financial model shows this explicitly. but since the discount rate is also 2% higher. discounting both sets of flows at the new (higher) discount rate shows no change in the present value of either cash flow.

one can expect a 1% increase in loss costs from lengthening durations of disability. workers’ compensation reserves are dominated by permanent total cases. (i) first as payrolls decline and the demand for workers’ compensation coverage decreases. Sept. “Workers’ Compensation Ratemaking. seniority effects on job retention patterns. the decline in interest rates raises the value of fixed-income assets the reserves more than it raises the value of compensation benefit obligations. for fear that there may not be a job to return to when they have fully recovered from the injury. permanent partial cases. there is no effect from a decline in claim frequency. For new business. 64 During recessions. the effect of a recession on the value of the insurance enterprise holding a block of workers’ compensation reserves is unclear. The decline in written premium raises expense ratios and reduces overall profits. the interest rates would not affect the valuation of new business.” (Casualty Actuarial Society Part 6 Study Note. the financial model may show either a net increase or a net decrease.13 12 For the effects of macroeconomic conditions on workers’ compensation claim frequency and durations of disability. the model will generally show an increase in the value of the insurance enterprise.12 For new business. these two effects are offsetting. though the effects of claim frequency are stronger. Moreover. The exact magnitudes depend on a host of factors. A general rule of thumb. Depending on the input assumptions. accident frequency is higher during prosperous years than during recessions. 13 Numerous items that we have not discussed in the text have opposing effects. Thus. In addition. For reserves. for an overall 1% decline in loss costs. the increase in durations of disability is particularly noticeable. and medium term temporary total cases. workers are often reluctant to file workers’ compensation claims for less severe injuries. the collectibility of premiums receivable 17 . A recession causing a 2% decline in loss frequency and a 1% increase in loss severity (duration of disability) for new business would cause a 1% or greater increase in reserves. In sum.hours. 1993) and the references cited therein. For the latter two types of cases. the claim filing frequency is lower during recessions. though. some workers are afraid that if they do file a claim during a recession. For the run-off however. Moreover. Recessions decline in business. Thus. Group health insurance studies of longterm disability coverage show that as unemployment increases. and the relationships between experience levels and injury rates. unemployment rises and durations of disability lengthen. see Sholom Feldblum. is that for every 2% decline in claim frequency during recessions. overtime practices. apparently because there may be no job to return to. however. the employer will look less favorably upon them during promotion and advancement decisions. durations of disability lengthen. unemployment levels. and (ii) second as carriers compete more strenuously for the remaining business. supporting are generally accompanied by declines in interest rates. This phenomenon is equally true for workers’ compensation: during recessions. such as the type of industry. written premium declines during recessions. Moreover. disabled employees tend to remain on disability for longer periods. For instance. even for the same accident frequency levels. As discussed above. during recessions.

the implications are sometimes counter-intuitive. the insurance l To properly measure the returns on surplus and the progression decreases. 18 .” These models tell us little about the actual profitability of the insurance enterprise. in December 1995. such as by setting a target return on equity that the insurance operations must provide. Volume I.” in Philbrick’s terms) from the capital requirements (or the “broad risk margin” in Philbrick’s terms). In fact. Indeed. Some do so indirectly. The financial model described in this paper uses two methods to measure performance. as employers find it difficult to meet their payment obligations. the model will simulate the expected return from policy year 1996 workers’ compensation business. Most of these models use equity assumptions. The relationship between the narrow risk margin and the broad risk margin. an internal rate of return model with a fixed reserves to surplus ratio may imply that a company with poor underwriting experience and high reserves is using more surplus. pages 303-347. Some do so directly. depends on the relationship the risk-free interest rate and the desired return on equity. l For the operating performance of distinct blocks of business.” Casualty Actuarial Society Forum (Spring 1994). For instance. pages I-90.14 For the internal rate of return models often used in workers’ compensation rate setting. The latter method is used by Stephen Philbrick to determine a pricing risk margin (or “narrow risk margin. see his “Accounting for Risk Margins. the company has less surplus. “An Illustrated Guide to the Use of the RiskCompensated Discounted Cash Flow Method. scenarios about interest rates and inflation rates. Surplus needed to “support” insurance underwriting is “tied up” in the “day to day” operations of the company. the model uses a return on (economic) surplus.Surplus and Profitability Many insurance profitability models deal with returns on equity.” Casualty Actuarial Society Forum (Spring 1990). The insurer’s management asks: “How much ‘free surplus’ does the company have?” “Is this ‘free surplus’ increasing or decreasing?” “What operations of the company are contributing to the increase or decrease?” of free surplus. which is precisely the item we are trying to measure. the desired return on equity becomes the internal rate of return that the projected premium must achieve. Both of these effects must be incorporated into the financial model to accurately ascertain the expected results from a recession on the profitability of new business. (The calculation of the needed economic surplus is described below. and analysis of the economic surplus needed to support this business. 14 The former method is used by the Fireman’s Fund risk-adjusted discounted cash flow model. such as by using the target return on equity to determine a risk adjustment to the loss reserve discount rate or to determine a risk margin in the premium. Butsic and Stuart Lerwick. given assumptions about 1996 compensation underwriting experience. such as “assumed premium to surplus” ratios or “reserves to surplus ratios. in Philbrick’s method. see Robert P.] For instance.

along with the long duration of these patterns. and then translated into target reserves to surplus leverage ratios. combined with target expected policyholder deficit ratios are used to set reserves to surplus leverage ratios and premium to surplus leverage ratios for the surplus supporting the run-off 15 Throughout the surplus allocation process described here.. Statutory accounting. this surplus supports the risk from poor reinsurance arrangements. such as surplus supporting overseas expansion. not at statutory figures.e. using target expected policyholder deficit ratios or target probability of ruin percentiles.g. 63 Free surplus. since the economic surplus needed is less than the interest cushion in the undiscounted reserve. a monoline workers’ compensation carrier with steady underwriting results may feel forced to hold significant statutory surplus to support these reserves because of the NAIC’s 11% riskbased capital reserving risk charge. with past workers’ compensation reserves and a The stochastic simulation analyses new policy year of workers’ compensation business. business growth. as well as the asset risk as the newly collected premium is held in the investment markets before the losses are paid. see below in the text. is a constraint on insurance company operations.. from random loss fluctuations.” Consider first a monoline insurance company. @ Surplus supporting fhe new business.company’s economic surplus (i. makes the “implicit interest margin” in undiscounted workers’ compensation reserves far exceed the capital required to safeguard the company against even highly unlikely adverse scenarios (i. which differ for each line of business. since we are looking at economic values of assets and of liabilities.e. or (c) it may be pure “surplus surplus. Once again. the economic net worth) is divided into three components:15 0 Surplus supporting the run-off business.. This is the company’s economic surplus that is not needed to support its insurance operations. we have actually used a combination of surplus determined by the economic allocation described in this paper and surplus as determined by the NAIC’s risk-based capital formula. we are concerned with “economic net worth. The amount of surplus needed is determined by stochastic simulation analyses. In other words. show positive surplus. The exhibits in this paper. and asset risks (such as default risk or market fluctuation risks) on the investments supporting the reserves. In addition. (b) it may be required for regulatory purposes (e.” The distinction is particular important for the surplus supporting workers’ compensation reserves.. stemming from underwriting cycle movements. however. as well as credit risk from reinsurance recoverables. the amount of surplus needed is determined by stochastic simulation analyses. 19 . which differ for each line of business. low probabilities of ruin or low expected policyholder deficit ratios). For allocating surplus by line of business. or an investment company subsidiary. and then translated into target premium to surplus leverage ratios. the statutory surplus needed to support workers compensation reserves is negative. using target expected policyholder deficit ratios or target probability of ruin percentiles. it may be needed to achieve a high risk-based capital ratio).” not “statutory surplus. The stability of workers’ compensation loss payout patterns.e. (a) It may be used for other operations. unexpected adverse loss development). This surplus supports the variability in underwriting results. and from natural catastrophes. This surplus supports the variability in the indicated reserves (i. For instance. however.

see below in the text. where the return on surplus allocated to reserves stems from the difference between a risk-free rate used for assets supporting the reserves and a risk-adjusted rate used to value the reserves themselves. since the investment portfolio includes also common stocks. Meanwhile. As is true for surplus supporting reserves. and mortgage-backed securities. we use an “discounted reserves” to “economic surplus” leverage ratio.” Return on Surplus and Surplus Progression As the year progresses. using a stochastic simulation with 10. In addition. see his “Pricing to Optimize an Insurer’s Risk-Return Relation.ls 0 16 These leverage ratios use market value accounting. use statutory leverage ratios. which is the yield rate on Treasury securities of short to medium maturities. Daniel Gogol uses a similar procedure. not a “statutory reserves” to “statutory surplus” leverage ratio.] For an illustration of the method in this paper. 18 The two returns . The company’s remaining economic surplus at the inception of the new policy year is “free surplus. Since the greatest value of the existing consumer base is the retention of insureds and the expected future profits. see Sholom Feldblum.business and the surplus supporting the new business. insurers 20 . May 1992) for the standard workers’ compensastion procedures.ls The leverage ratios determine the amount of surplus needed at the inception of the new policy year. “Pricing Insurance Policies: The Internal Rate of Return Model” (Casualty Actuarial Society Part IOA Examination Study Note. persistency rates are high in most casualty lines. the financial model uses a “riskfree” discount rate. Winter 1996 Edition (Casualty Actuarial Society. since the discount rate for loss reserves is generally less than the investment yield on the surplus funds. the difference in the two yield rates times the reserves to surplus ratio is an additional return on these surplus funds.000 runs of workers’ compensation reserves along with a 1% expected policyholder deficit ratio. respectively. The investment yield of the company is somewhat higher than this rate.are not independent. who must file rate revisions with state insurance departments. corporate bonds.” CAS Forum. [Pricing actuaries. In addition. the actual return on the surplus supporting new business includes a third piece: the favorable or adverse underwriting performance of this business. in contrast. The expected return on the surplus supporting new business is also composed of two pieces: The assets corresponding to this surplus earn a return in the investment market. For instance. 0 there are three “returns.the return on the run-off of existing business and the return on new business . pages 213-242. 17 To determine the economic value of the loss reserves. the projected underwriting gain or loss on this business is an additional return on these surplus funds.17 The actual return on the surplus supporting reserves includes a third piece: the favorable or adverse development on these reserves.” The expecfed return on the surplus supporting reserves is composed of two pieces: The assets corresponding to this surplus earn a return in the investment market. such as personal automobile and workers’ compensation. 1995).

so the investment yield on the aSSetS supporting the reserves does influence the return. the financial model asks: “How has the amount of free surplus progressed over the course of the year?” The three profitability measures overlap. and the new business becomes run-off business. The model accrues profit as the premium is written. [The choice of investments does affect the insurer’s returns. however. The result is that many of the same policyholders occupy the company’s existing book of business as well as its future book of business. The are reluctant to implement (and regulators are equally reluctant to allow) large rate changes. For expected results. reserve developments. given the assumptions for underwriting results. The Butsic and Philbrick models seek to align the return with “uncertainty. the two pieces are largely independent. they are not independent. The third profitability measure shown by the financial model asks: “How will the amount of free surplus progress over the course of the year. The initial reserves decline over the course of the year.” The total surplus of the company increases by the returns on investable funds plus underwriting gains (or minus losses) minus federal income taxes and minus the unwinding of the interest discount on economic reserves. the investment return on the surplus supporting the reserves.” and + - This is different from the approaches used incorporates an “implicit reserve margin” risk-free rate. For instance. both leading to lower surplus requirements. The financial model described in this paper seeks to align the return with the insurer’s operating decisions. Expected underwriting profits are included in the new business section. The return on the run-off of existing business {if interest rates do not change) is the investment return on the assets supporting the reserves. the insurer’s action do not much affect the random reserve developments that determine the actual return. there is a new year of new business. good expected underwriting results for new business will raise the return on surplus supporting this business and also result in an increase in free surplus. Once the policy has been written and earned.6 Finally. held “above the line.” As long as there is uncertainty in the ultimate loss payments. Thus. Butsic using a reserve discount rate lower than the “explicit reserve margin” (his “narrow risk policy premium. with appropriate adjustments for inflation and expected real business growth.] 21 . They are used for different purposes. The free surplus at the beginning of the year minus the surplus needed to support both the run-off business and the new business is the initial free surplus. and investment performance?” Similarly. there is a progression of “free surplus. must earn a “risk-compensated” return. the insurer. not as the losses are paid. Philbrick incorporates margin” or NRM) that is embedded in ultimately paid to equityholders. Conversely. with additional surplus requirements. once the year has actually transpired. unexpected favorable or adverse results on the run-off of the existing book may portent corresponding favorable or adverse results on the book of new business. the amortization of discount on the market value of the reserves. We assume that the company expects to write a similar volume of business one year from now as it is writing in the new policy year. or the insurer’s owners. by by an the Robert Butsic and by Stephen Philbrick.

Greater reserve variability means that the insurer must hold more surplus to support the reserves. In fact. [See Appendix D. Industry-wide workers’ compensation results have been good 1990’s. underwriting in the early ratio. 0 For the base case. and the discount rates used.6% to -7. and the company projects an 80% loss and loss adjustment expense company’s direct writing distribution system (a salaried sales force) provide underwriting expense ratio of 18%. It does not mean that the insurer “earns” more by holding these reserves.] The long payment lag in workers’ compensation anticipated return on surplus of 42. such as expansion into new markets. with a 2% rise in inflation and interest rates.” Exhibits 3 through 7 detail the results for each line of business. and it therefore has less surplus for other uses. not a “risk adjustmenY for reserve variability. the alternative scenario. and the loss reserve discount rate is 5. The return on surplus supporting new business is driven primarily by expected gains and losses.8% to 42. exhibits 6-12. there are differences in the other lines of business.1s With a three to one “reserves to surplus” leverage ratio. particularly the duration of the assets supporting the compensation reserves. [See Appendix A for these figures. not the appropriate discount rate. primarily because of the drop in the market value of investments. yielding a combined ratio of 98%. 22 . from 42. Scenarios and Returns on Surplus Exhibits 8 through 12 of Appendix D show the various measures of return described above. the after tax investment yield is 5.6% return. with only partially 19 This difference reflects the lower yielding bonds supporting the reserves. The reserve variability is used to determine the target leverage ratio. Exhibits 1 and 2 of Appendix D illustrate graphically the connection between the return on surplus and the progression of “free surplus. The progression of free surplus shows how much money is available for other corporate functions. for both the base case scenario and for the alternative scenario.6%. the return drops sharply. from 6.] In fact. The magnitude of the change in the return is driven by several items.return on surplus supporting the new business shows the profitability of the insurance operations.7%. for the total returns. causes only an insignificant change in the return. the return is slightly higher than the expected after-tax return on assets. the high 1994 and 1995 returns in workers’ compensation result from strong benefit reforms in many states along with a movement to “managed care” program. depending on the “discounted reserves to economic surplus” leverage ratio used for each line. the sensitivity of workers’ compensation retrospective premiums. @ In the alternative scenario. Consider first the return on surplus supporting the run-off business. the inflation sensitivity of the losses. The for a low Exhibit 3 of produce an This return is not much affected by changing interest rates or inflation rates. for an analysis of the sources of return.4%. as discussed earlier. for instance. using an expected policyholder deficit analysis. the federal income tax implications. resulting in a 6. For workers’ compensation.3%. However. and exhibits 13-19.7%. the 30 basis point difference in yield contributes 90 basis points to the return on surplus.

0 Some analysts use the same leverage ratio. decreases in 1996 (as in Massachusetts) will reduce the anticipated return on new business.” or a “three-to-one reserves to surplus ratio. more capital is moved to “free surplus.] The capital requirements used in the model are now being updated. though they will have no effect on the anticipated return on surplus supporting existing business. allocation of existing capital. Past actuarial attempts to assess surplus requirements have several failings. 20 Because of the competitive characteristics of the commercial insurance market. it is insufficient to adopt simple rules-of-thumb. reducing the expected return on equity. not simply an Similarly. The financial model described here takes a different approach.” such as a “two-to-one premium to surplus ratio.20 Rate level anticipated. such as a premium to surplus ratio or a reserves to surplus ratio. If the company is profitable. whereas the return on surplus supporting new business does not change significantly. [These are the leverage ratios which are reproduced in the exhibits in the Appendix. one may expect workers’ compensation rate levels to decrease in line with costs. for the entire underwriting risk. This approach demands a method of quantifying the capital truly needed. Actuarial science has developed several methods of quantifying the needed capital. Such assumptions simply beg the question of how much capital is actually needsd. analysts use an “assumed surplus requirement. with large reductions in workers’ compensation claim frequencies.” to model the needed capital.” which the company may use for other purposes. such as the Kenney rule used by some regulators of a “two-to-one premium to surplus ratio” or the ad hoc “reserves to surplus” ratios often used by pricing actuaries for workers’ compensation ratemaking. If the company is profitable. which are avoided in the financial model described here.The incentive effects of the benefit reforms were greater than offsetting rate reductions. Another common approach has been to take the company’s existing surplus and simply This allocate it to lines of business based upon premium volume or reserve volume. using an “expected policyholder deficit” analysis by line of business from simulation analyses. 23 . types of liabilities. with additional detail showing types of assets. The financial model described here began with leverage ratios determined from “probability of ruin” analyses. Surplus Requirements The measurement of expected profitability requires an assessment of the capital needed to support the business. 0 Most commonly. This approach has two shortcomings. Exhibits 13 and 14 of Appendix D show the overall company results under both the base scenario and the alternative scenario for (i) surplus supporting the run-off versus (ii) surplus supporting the new business. procedure skirts the issue entirely. it allocates more capital to each line of business. as has already occurred in several states in 1996 and 1996. The return on surplus supporting the run-off business drops sharply. and federal income taxes.

like life annuities. which are based on ten years of Schedule P data. Wall’s estimates. And no insurer would say that the great uncertainty in pollution payments necessitate a low discount rate for determining the present value of the reserves. this question is: “How do these different scenarios affect the progression of ‘free surplus’?” 93 For a rough estimate of the payout pattern of workers’ compensation reserves. (Casualty Actuarial Society 1987 Discussion Paper Program). But insurers are not holding pollution and asbestos reserves in order to earn a high return on surplus. The needed surplus to support these reserves may be estimated either by a reserves to surplus leverage ratio or by other actuarial techniques. pages 446-533. are severely understated. it fails to recognize that there are two distinct risks. as used in some internal rate of return pricing models.” since workers’ compensation loss reserves. Wall. a premium to surplus ratio is irrelevant. is an assumed ratio of This approach compounds all three errors statutory reserves to statutory surplus. the proper type of leverage ratio. (a) The underwriter seeking to sell new policies or the pricing actuary seeking to make rates for new business must quantify the variability of loss costs over the coming policy period.” Financial Analysis of insurance Companies. based on 24 .is appropriate. since the lifetime pension cases. (b) To assess the surplus needed to support the run-off of pollution and asbestos claims. thereby leading to a high expected return on surplus. one must know the volume of exposures. approaches ignore the implicit interest margins inherent in undiscounted reserves.l First. differs for these two types of risk. For instance. which form the bulk of workers’ compensation reserves for older policy years. a standard measure of surplus needed to support workers’ compensation writings. the insurer’s management asks: “How do different scenarios relating to environmental liabilities affect the company’s net worth?” Rephrased in the terms used by the financial model. Insurance companies want to know the effects of alternative environmental scenarios on the company’s performance. see Richard G. These For instance.22 21 Insurer liabilities for environmental exposures highlight the difference between returns on surplus and the progression of free surplus. the degree of diversification. have an extremely slow payout pattern. as well as the needed analysis. new business volume . discussed above: l Undiscounted workers’ compensation loss reserves contain an enormous “implicit interest margin.as considered in a premium to surplus ratio . to assess the surplus needed to support a new policy year of Homeowners writings. “Insurance Profits: Keeping Score. and the adequacy of reinsurance arrangements. Rather.21 $ Past approaches often use leverage ratios of statutory figures to statutory surplus. (b) The corporate accountant seeking to complete the company’s financial statements or the reserving actuary seeking to estimate the company’s loss obligations must quantify the variability of adverse reserve developments on the company’s existing reserve portfolio. A reserves to surplus leverage ratio is irrelevant. (a) For new business. Second. have slow but steady payment patterns combined with long durations. which are important for different insurance personnel. The authors’ own analyses.

23 This statement should be qualified. are concerned with the same issue. 25 . show an average time to payment of about eight years for workers’ compensation reserves. are shown in Exhibits 2 and 3 of Appendix F.e. Neither has much theoretical foundation. which is not affected by the magnitude of the post-insolvency loss. Most of the deficiencies of past analyses are solved by these leverage ratios: l l l The ratios are determined by probability of ruin analyses. All measures use “economic surplus” and “economic reserves. based upon 2.l The major risks in workers’ compensation business are not the fluctuations in loss reserves but the uncertainties in new business. the appropriate measure of financial strength should depend on the use of the model. not by tradition or by rule of thumb. The supporting exhibits for workers’ compensation. Policyholders are indeed concerned about the amount of loss. or the prospects for rate level increases. Similarly. and the former has a short actuarial tradition. The latter has a long regulatory tradition. The EPD ratio analysis says that 0 The appropriate measure of solvency is the ratio of the expected policyholder obligations to policyholders (i. These ratios are shown in Exhibit 1 of Appendix F. particularly when different risks are being considered in combination. However. Thus. The appropriate measure of solvency for regulators and for policyholders is the expected policyholder deficit ratio.23 The corollary to this is that the appropriate deficit to the amount of capital needed to guard against any 25 years of paid loss experience from 10% of the industry’s business. insulates investors from the magnitude of the loss after bankruptcy. industry underwriting cycles. which serve to protect policyholders. whether by the random occurrences of accidents or by macroeconomic conditions. the expected premium levels. or the regulatory climate that affect the claim frequency rate.” The financial model described here is now being refined by means of the expected policyholder deficit (EPD) concept developed by Robert Butsic. Regulatory measures of solvency. the expected losses).000 runs of a stochastic simulation analysis. Expected Pollcyholder Deficit Approach l The The original leverage ratios used in the financial model were developed from a probability of ruin analysis. there are other advantages of the EPD approach. The “corporate shield.. so we present the results from this approach. The appropriate measure of financial strength for investors and for company management is the probability of ruin.” however. management and employees are concerned with job security. A “three-to-one” reserves to surplus ratio has no more actuarial support than a “twoto-one” premium to surplus ratio. Separate ratios are used for the run-off of existing business (reserving risk) and for the writing of new business (premium risk).

expressed in terms of EPD ratios. as shown in Exhibit 1 of Appendix H. we normalize the results to $100 of average undiscounted reserves. for accident years 1970 through 1994. Each link ratios (one for each column). forthcoming). should be uniform across risks. Sholom Feldblum. “Workers’ Compensation Reserve Uncertainty” (1996 Casualty Loss Resewe Seminar discussion paper program.000 simulations produce 10. the expected deficit is the average deficit over all simulations. and Gary Blumsohn. When the future loss obligations are less than the funds held by the insurance company to meet these obligations. the capital needed to guard against workers’ compensation reserving risk should produce the same EPD ratio as the capital needed to guard against personal auto resewing risk. The Expected Policyholder Deficit To properly quantify the surplus needed for each type of risk.000 reserve amounts. In the analysis here. That is.000 sets of stochastic simulations. We ask: “How tight is this 24 A more complete description is contained in Douglas M. The “expected policyholder deficit” is the average deficit over all scenarios. The Stochastic Simulation How should we measure the uncertainty inherent in the loss reserve estimates? We use stochastic simulation of the experience data to ensure statistically meaningful results. simulation produces 24 “age-to-age” age factors that drive the actual loss link ratios to lognormal curves. separately for indemnity and medical benefits. @ Capital requirements. From these data we develop 24 columns of paid loss “age-to-age” link ratios. weighted by the probability of each scenario. We fit each column of “age-to-age” “sigma” (o) parameters for each. When the future loss obligations are greater than the funds held. then the insurer need hold funds just equal to the reserve amount to meet its loss obligations. The 10. We illustrate this for workers’ compensation resewing risk. 26 . The insurer holds surplus to ensure that the loss obligations will indeed be met. which discusses the stochastic simulation. with simulation parameters that are based upon the actual experience of the company. determining “mu” (u) and We perform 10. Hodes. and the influences on reserve uncertainty. the curve fitting considerations. the “deficit” is zero. For ease of presentation.risk is the amount of capital needed to reduce the EPD ratio to a predetermined figure. We begin with 25 years of countrywide paid loss workers’ compensation experience. These are the age-topayments.24 The calculations are as follows: Were there no uncertainty in the future loss payments. we combine a simulation analysis with an expected policyholder deficit approach. for which we have completed the full analysis. funds equal to the expected loss amount will sometimes suffice to meet future obligations and will sometimes fall short. the “deficit” is the difference between the two. Since future loss payments are not certain. each of which is equally weighted.

distribution of reserve amounts?” *

We answer in two ways.

We show the standard deviation, the mean, and two other percentiles of the distribution (5% and 95%). For instance, the table below shows that for discounted reserves with no adjustments for inflation, the mean reserve amount is $5.27 million, the standard deviation is $3.4 million, the 95th percentile is $58.7 million, and the 5th percentile is $47.9 million. To facilitate the comparison of reserve uncertainty with other types of risk used in the financial model, we use the “expected policyholder deficit (EPD) ratio” as a yardstick. We ask: “How much additional capital must the insurer hold to have a 1% EPD ratio?” The table below shows that for discounted reserves, the required capital for a 1% EPD ratio is $2.4 million. Average Reserve Amount Undiscounted Discounted: 6.75% / 100.0 52.7 1 Standard Deviation of Reserve 19.5 3.4 1 95th Percentile of Reserve 135.3 58.7 5th Percentile of Reserve 1 .74.0 ) 47.9 1 1 Capital Needec for 1% EPD Ratio 31.0 2.4

*

Reserve

Discounting

We are primarily concerned with the economic values, or discounted values, of the reserves, not with undiscounted amounts. Much of the variation in statutory reserve requirements stems from fluctuations in “tail factors,” This fluctuation depends in part on inflation rates. For discounted reserves, the effects of changes in the long-term inflation rate are offset by corresponding changes in the discount rate. Moreover, tail factor uncertainty has a relatively minor effect on the present value of loss reserves, even if the discount rate is held fixed. Thus, the distribution of discounted loss reserve amounts is more compact than the distribution of undiscounted loss reserve amounts. Because statutory accounting mandates that insurers hold undiscounted reserves, we show analyze results both for discounted and for undiscounted reserves. Moreover, the difference between the discounted and undiscounted reserve amounts is the “implicit interest margin” in the reserves, which is important for assessing the implications of the reserve uncertainty on the financial position of an insurance company. Length of the Development

The paid loss development for 25 years is based on observed data. Workers’ compensation paid loss patterns extend well beyond 25 years. For each simulation, we complete the development pattern as follows: 0 Given the 24 paid loss “age-to-age” link ratios from the set of stochastic simulations on the fitted lognormal curves, we fit an inverse power curve to provide the remaining

27

“age-to-age” factors.25 This fit is deterministic. 4) The length of the tail is chosen (stochastically) from a linear distribution years.

of 30 to 70

Let us suppose first that the company holds no capital besides the funds supporting the reserves. We ran our analysis For the discounted analysis, the average reserve amount is $52.7 million. About half the simulations give reserve amounts less than $52.7 million. In these cases, the deficit is zero. The remaining simulations give reserve amounts greater than $52.7 million: these give positive deficits. The average deficit over all 10,000 simulations is the expected policyholder deficit, the EPD. The “EPD ratio” is the ratio of the EPD to the expected losses, which are $52.7 million in this case. Clearly, if the probability then the EPD ratio will needed reserve amounts reserves - then the EPD distribution of the needed reserve amounts is “compact,” or “tight,” be relatively low. Conversely, if the probability distribution of the is “dispersed” - that is, if there is much uncertainty in the loss ratio will be relatively great.

We now “fix” the EPD ratio at a desired level of financial solidity and determine how much additional capital is needed to achieve this EPD ratio. We use a 1% EPD ratio as our benchmark, since this is the ratio which Butsic uses for risk-based capital applications. Suppose the desired EPD ratio is 1%. If the even if the insurer holds no capital beyond that EPD ratio may be 1% or less. If the reserve must hold additional capital to achieve an EPD the greater the required capital. Results The results for the base case, with discounted reserves, are shown in the table above.26 The average discounted reserves are $52.7 million, and additional capital of $2.4 million is needed to achieve a 1% EPD ratio. The corresponding full value reserves are $100.0 million. The company uses tabular discounts on the indemnity portion of life-time pension cases at a 3.5% discount rates, which is the rate used in the NCCI unit statistical plan. The resulting statutory reserves, normalized to a $100 reserve distribution is extremely compact, then required to fund the expected loss payments, the distribution is more dispersed, then the insurer ratio of 1%. The greater the reserve uncertainty,

2 5 On the use of the Smoothing, and Interpolating Society, Vol. 71 (1984), pages Mohrman, vol 72 (1985), page

inverse power curve, see Richard Sherman, “Extrapolating, Development Factors,” Proceedings of the Casualty Actuarial 122-192, as well as the discussion by Stephen Lowe and David 182, and Sherman’s reply to the discussion, page 190.

26 We ran our simulation for several cases: (i) discounted versus undiscounted reserves, (ii) with and without various adjustments for medical inflation, and (iii) with and without consideration of loss-sensitive contracts. The “base case” in our analysis uses discounted reserves with no adjustments for medical inflation or for loss-sensitive contracts. For a full description of the analysis, see Hodes, Feldblum, and Blumsohn, “Workers’ Compensation Reserves Uncertainty” (op. cit.).

28

million undiscounted

reserve, are about $92 million

The difference between the perspective in the financial model described here and the “received The common view is that workers’ actuarial wisdom” warrants further comments. compensation reserve estimates are highly uncertain, because of the long duration of the claim payments and because of the unlimited nature of the insurance contract form. This uncertainty creates a great need for capital to hedge against unexpected reserve development. In fact, the opposite is true. There is indeed great underwriting uncertainty in workers’ compensation, and regulatory constraints on the pricing and marketing of this line of business have disrupted markets and contributed to the financial distress of several carriers. But once the policy term has expired and the accidents have occurred, little uncertainty remains. The difference between the economic value of the reserves and the reported (statutory) reserves, or the “implicit interest margin,” is many times greater than capital that would be needed to hedge against reserve uncertainty. These results have important implications for our financial model.
l

There is no “leverage ratio” between statutory reserves and statutory surplus, since one needs negative statutory surplus to support workers’ compensation reserves (if undiscounted reserves are indeed held). capital requirements, force Regulatory requirements, however, such as risk-based companies to hold more surplus to guard against “reserving risk” than they actually need.27 compensation as well as a imposed by ratios used in analysis.*a

l

Thus, our financial model assumes that the implicit interest margin in the reserves provides the full economic surplus needed to support the reserves substantial amount of “free surplus.” However, because of the constraints statutory accounting and by the NAIC’s risk-based capital formula, the leverage the financial model are still lower than the implicit leverage ratios from the EPD Inflation and Interest Rate Risks

For the past twenty years - ever since the dramatic rise in inflation during the late 1970s casualty actuaries have debated the effects of inflation on the economic worth of insurance companies. Because there are multiple and simultaneous effects, several of which are difficult to quantify without a sophisticated financial model, past analyses of this issue have often been

27 For instance, the current NAIC risk-based capital formula uses an 11% reserving risk charge for workers’ compensation, before company-specific adjustments, such as the company’s average reserve development, the percentage of business written on loss-sensitive contracts, and loss concentration factor; see Sholom Feldblum, “Risk-Based Capital Requirements” (Casualty Actuarial Society Part 10 Examination Study Note, Second Edition, July 1995), for a complete description of the NAIC risk-based capital reserving risk charge. 28 The final surplus requirements by line of business and by operational unit used by the company were determined in part by management discretion, with consideration of rating company expectations, peer company practices, and NAIC constraints, not solely by the actuarial analysis reviewed in this paper,

29

incomplete. Inflationary changes have several effects on the financial solidity of an insurance For the “alternative scenario” discussed earlier, we assume that the inflationary matched by a corresponding change in interest rate (the “Fisher effect”). enterprise. change is

They reduce the market value of fixed income securities. The effects on payment patterns and market values of fixed income securities have been studied by both investment analysts and by actuaries (primarily life actuaries). A full analysis requires consideration also of issuer options, such as calls on corporate bonds and prepayments on mortgages and mortgage backed securities. The effects on the company’s portfolio of mortgage backed securities are shown in the exhibits and they are discussed further below. They temporarily reduce the market value of common stock investments, though not of real estate investments. The effect on all equity investments, however, is generally short-term. Over the long-term, equity investments serve as a “hedge” against inflation. Good data on the magnitude of the relationships are lacking. Our model uses in-house studies based on the stock market experience of the past twenty years, as described further below. They reduce the market value of fixed liabilities, such as some workers’ compensation indemnity payments or personal automobile no-fault compensation payments. They increase the nominal value of most casualty loss reserves. In other words, most casualty loss reserves are “inflation sensitive”: if inflation increases, the nominal amounts increase as well, with little change in the market value. Quantifying the last of these effects is particularly difficult without a financial model. In workers’ compensation, for instance, inflation affects medical benefits through the payment date. In about half of the U.S. jurisdictions, indemnity payments that extend beyond two years have COLA adjustments that depend on inflation. To properly assess the effects of inflation on the company’s loss reserves, we “strip” out past inflation from the historical loss triangles, determine the paid loss “age-to-age” link ratios, then restore expected future inflation to the indicated (future) link ratios. In other words, we make the following adjustments to the loss reserve development analysis used to project future loss payments: 0 We convert the paid losses to “real dollar” amounts by means of an appropriate inflation index. For workers’ compensation medical benefits, we used the medical component of the CPI. We select a future inflation rate that is consistent with the scenario being analyzed. For instance, if we project future inflation at 6% per annum, we may select 7% per annum as the medical inflation rate.

8

B) Finally, we combine the projected link ratios and the projected inflation rate to determine the expected loss outflows.

30

the AAA recommendations would show a slight interest rate risk charge. “Economic values. The reason for this difference risk-based capital formula. The risk-based capital formula picks up this adverse development in the reserving risk charge. however.” when needed. The resulting differences between the statutory charges developed by the AAA task force on risk-based capital for interest rate risk and the economic effects quantified by the financial model described here are large. are determined by means of market interest rates. For a two point increase in interest rates and in inflation rates. not by a “worst-case year” approach (see above). For NAIC risk-based capital purposes. they do not enter the underlying analysis. The inflation sensitivity of casualty reserves. The surplus needed to support the reserves is determined by an “expected policyholder deficit” analysis. The NAIC formula uses a “worst-case year” approach to determine the surplus needed to support reserves. is a scenario based model. The financial model used here. l is the overstatement of the reserving risk charge in the NAIC The NAIC formula uses a flat 5% discount rate to determine the implicit interest margin in the reserves. Results A comparison of the results from the financial model with the recommendations on the American Academy of Actuaries task force on risk-based capital regarding interest rate risk is most instructive. must be explicitly incorporated into the results. which causes higher nominal cash flows when inflation increases. To reflect it also in the interest rate risk charge would be “double-counting” the same risk. and then restoring the projected inflation to the assumed future payments. There are no valuation rates. determining link ratios in real dollars.Exhibits 1 through 5 of Appendix G shows the method of “stripping” inflation out of the loss payment triangle. The financial model described here deals with cash flows. @ The risk-based capital charge must consider the interplay of reserving risk and interest rate risk. The financial model described here uses either an “expected policyholder deficit” analysis or a “probability of ruin” analysis to determine the surplus needed to l 31 . Accounting conventions serve only as constraints. if it would show any at all. The financial model described here the anticipated cash flows. The statutory interest rate risk considered by the AAA task force differs in two respects from the economic effects of a simultaneous shift in interest rates and inflation rates: 0 The statutory effect must consider the valuation rates used for assets and liabilities. this is the rate implicit in amortized values for fixed income assets and a flat 5% rate used for loss liabilities. Increases in interest rates that are accompanied by increases in inflation cause adverse development of undiscounted losses. For most companies. the economic return on the surplus supporting reserves changes from a positive return to a significant negative return. since this is the rate used to discount losses for the reserving risk charge. The difference between these rates may provide either a cushion or an extra charge for interest rate risk.

on the company’s financial position?” Rather. Scenarios may be divided into two categories: economic scenarios and insurance scenarios. For instance. which affect bond prices and stock prices (which are similar for most insurers) as well as claim frequency and claim severity (which vary by line of business). The financial model described here shows the expected results under a variety of future scenarios. economic recessions are often characterized by falling interest rates and high unemployment. Many observers have pointed out that the NAIC reserving risk charges and written premium risk charges seem to be ad hoc numbers only marginally related to the actual risks faced by companies. Thus. modeling the effects of scenarios is a two step process: * The analyst must translate the scenario into a set of model assumptions. elements as the underwriting cycle. a drop in inflation would decrease the (nominal) required reserve in some lines of business. a drop in interest rates would increase the market values of fixed income securities.support reserves. characteristics and run * . a drop in inflation. industry competition. These economic assumptions must be applied to the company’s through the financial model. each of which may affect multiple elements of an insurance company’s operations. a fall in stock market prices. and the rise in unemployment would probably decrease workers’ compensation claim frequency but increase workers’ compensation claim severity (or “durations of disability”). or prosperous years. Scenarios are composed of interdependent elements. a recession scenario may be characterized by a drop in interest rates. or of an underwriting cycle downturn. The AAA interest rate risk recommendations were hampered by the need to fit into an existing formula that did not accurately reflect the actual risks. Economic scenarios posit changes in the macro-economic environment. some are less serious. 32 . or state regulation. high Insurance industry scenarios posit changes in such unemployment. Scenario Testing The proper management of an insurance enterprise requires consideration of the overall environment. An insurance executive does not ask: “What is the effect of a 200 basis point drop in interest rates. though the economic value may increase because of the decline in interest rates. We have included this comparison of the financial model described here with the NAIC riskbased capital charges to highlight the importance of actuarial analysis. Valuation actuaries must be careful to consider the risks highlighted by financial models of the type described here. he or she may ask: “What would be the effect of an economic recession. For instance. on the company’s performance?” The actuary must translate the scenarios into model assumptions and rerun the cash flow projections to answer such questions. . . Some risks are more serious than those implied by the NAIC risk-based capital formula. and a rise in unemployment. such as recessions. or a 10% rise in personal auto claim frequency. not of isolated risks. For instance.

For instance. Unemployment and the assumptions relevant for the Economic environment 1. 29 For a more complete discussion of the process of scenario building. medical) 3. 4. The types of scenarios. an underwriting cycle downturn would involve changes in premium revenues (because of changes in rate adequacy). an economic recession and an underwriting downturn. such as growing involuntary markets.” in Society 1995 151-177. pages 33 . changes in expense ratios (because of changes in the percent of the market turning to self-insurance). 3. 5. interest rate movements and changes in the cost of capital may affect the course of the underwriting cycle.29 The economic environment and the insurance environment are not independent. Casualty Actuarial Discussion Paper Program (Landover. such as high unemployment.In other words. I rate adequacy premium revenues loss ratios expense ratios We have applied this analysis with two scenarios. 3. 1. bond prices stock prices claim frequency claim severity premium revenues I Assumptions . interest rates 2. Insurance scenarios are equally complex. Inflation rates (wages. the effects of the economic scenario depend on the composition of the investment portfolio. the lines of business written. Feldblum. Underwriting cycle 2. For instance. 2. Peer competition 4. changes in involuntary market sizes and “burdens. are clearer in this line of business than in most others. for the company’s workers’ compensation book of business. see Sholom Testing.” and changes in the expected loss ratio. financial model are shown in the graph below. Maryland: Colortone Press. 1995). “Forecasting the Future: Stochastic Simulation and Scenario incorporating Risk factors in Dynamic Financial Analysis. Workers’ compensation is particularly appropriate for scenario testing for two reasons: l The effects of economic conditions. } Insurance environment L 1. and of insurance industry conditions. 4. Specifying the attributes and associated characteristics of a scenario requires a keen understanding both of the macroeconomic environment and of the effects of external factors on company operations. 2. their distinguishing attributes. Consumer demand 1. and the nature of the loss reserves. Involuntary markets 3.

(a) Workplace accidents are noticeably greater for young and inexperience workers. lest there be no job to return once they have recuperated. During the late 1970s. make cash flow projections essential for proper performance measurement. The effects on claim frequency affect new business results. 34 . not the results of new business. they affect separate components of the insurer’s business (new business versus run-off business). As noted above. Carefulness does not abide well with fatigue.30 [Note the distinction: working employees are less likely to file claims during recession. (b) In addition. (3 Users often have different opinions about the components of scenarios and about their effects on company operations. Translating the projected scenario into model assumptions. and allocate surplus (or economic net worth) backing each line of business. (c) Finally. They have no effect on the run- Conversely. the analyst must also allocate the assets supporting the reserves of each line of business. Disabled employees are less likely to recover from disabilities during recessions. The financial model described here separates the effects of the projected scenario on these two components. recessions cause a lengthening of durations of disability. when unemployment rates rose. makes the problem almost intractable at first glance. An injured worker is unlikely to declare himself healed if there is no job to return to. running both the base assumptions and the revised assumptions through the model. claim frequency rates can be quantified by monthly reports. In addition. and workers’ compensation claim frequency rises. and comparing the resulting cash flows. In truth. during recession. overtime work increases. so neither need by quantified. During prosperous years. where a rise in unemployment may cause a lengthening of durations of disability whose financial effects are spread out over thirty years. Needed rate increases or decreases for the new book of business are should not be offset against unexpected gains or losses on the existing reserve portfolio. the model allows the analyst to separate individual lines of business or blocks of business. off of existing business. enables the user to see the effects of a changing environment.Consider several effects of a recession on claim frequency and claim severity in workers’ compensation. Some analysts err by assuming that the effects on claim frequency and claim severity (durations of disability) offset each other.] This relationships affects primarily the run-off of existing business. and claim frequency increases even more. For instance. The long duration of workers’ compensation business. who are frequently unfamiliar with the hazards of the machines or equipment used. The use of a financial model for scenario testing has two advantages: 0 The interdependence of the various scenario components. durations of disability under these plans rose concomitantly. and the complexity of their effects on insurance operations. users may have different views on the expected stock 30 This influence is based on the experience of employer provided group health insurance. Durations of disability can be quantified only over periods of years. In other words. workers are loath to file claims. firms hire new workers.

This paper documents the cash-flow financial model used by a major commercial line insurance company. Property-casualty insurance companies Actuarial practice is outracing actuarial literature. It discusses the uses of the model. And it should inform all actuaries that financial modeling is here to stay. the analyst can show the cash flows resulting from different model assumptions. with fluctuating rate adequacy. Conclusion Corporate financial models are less important for a company writing short-tailed lines of business with an assured consumer base. But financial models are essential for companies writing long-tailed lines of business. as well as the sensitivity of the results to changes in the assumptions. The NAG and the major rating agencies have developed solvency models to help ascertain companies’ resilience to adverse future conditions.market movements during a recession or on the effects of unemployment on workers’ compensation claim frequency. and the scenarios that it analyses. Dynamic financial models take various forms. With the financial model described in this paper. and it would confront experience actuaries with the complexities of reserving risk. and little competition. and volatile consumer bases. 35 . It should acquaint new actuaries with the many components of dynamic financial analysis. adequate rates. there are no “cookbook” approaches to serve as benchmarks. severe competition. interest rate risk (inflation risk). are adopting models originally designed for life insurance companies and adapting them to their specific risks. corresponding to the types of business written by the insurance enterprise and the issues that they address. Unlike traditional ratemaking or reserving techniques. and scenario testing. the types of risks addressed.

75% Federal income tax rate 35. Pretax Yield on Bonds Stock Dividend Yield Rate of growth in value of stock Resulting annual return on stock 8.08% For each line of business.75% 8. expenses. After-tax rate is appropriate because income taxes are explicitly mcluded in the cash flows. resulting m the overall mcrease in loss ratio. the abovc rates affect equally losses. the above rates affect equally losses.12% For each line of business. Resulting annual return on stock 12 Z”.00”” 6.. Additional increase m mtercst rates 2.OO’Y Resulting annual inflation: Additional growth in real exposures 0. and premium. when rates first mcrease. the market value of stock declines by five percentage pomts for each percentage point increase m interest rata.7510 Rate of growth in value of stock# 10 00”.00% D b.0070 10.00% Resulting growth m real exposures 2. # J’he above growth rate for stocks is the long-term rate In thr short term.00% Growth in real exposures 2.00% Resulting growth in underwriting 6. However. In our alternahre scenarm.00” G Resultmg new bond FE-tax bond yield 10.00’% Resulting growth m underwriting 8. crpcnscs.30% Stock Diwdend Yield 2.00% Discount Rate for Insurance operations cash flow’ 5.40% * This rate is the after-tax yield on bonds. Alternative Additional mcrease in inflation: Scenario: 2.70% Note: Actual input in the model had finer line detail. the market value of stock reacts negatively to increase m interest and inflation rates. 36 . This decline was assumed to be eventually rccovcred after tn-o years Federal income tax rate Discount Rate for Insurance operations cash flow 35. we assumed that mitially.SamDie Financial Model Appendix A Exhibit 1 Notes to Underwriting Assumptions Base Case Scenario Annual inflation rate 4. and ~remiutn.30% 2. nominal losses grow at a higher rate than prenxum.

038 5.000 7.274 1.349 2.806 4.521 1.671 2.588 3.425 7. GL& CMP 513 544 577 612 650 689 731 776 823 873 Net of Reinsurance Comm Auto & Total Pers Auto &HO 2.864 6.415 2.822 7.087 2.244 3.155 4.613 1.490 2.189 3.108 Involuntary.236 7.883 3.347 2.086 2.834 3.650 3.787 5.043 1 482 511 542 575 610 647 687 728 773 820 I Note: Actual input in the model had finer line detail 37 . GL& CMP 505 535 5M) 602 639 678 719 763 809 656 Expense ($ Millions) Total Pen Auto &HO 1.718 2.599 7.434 1.518 2. Work Involuntary.356 8.sn 8.212 2.058 3.562 2.431 6.352 1.220 6.711 1.387 5.201 1.152 3.491 2.974 3. GL& CMP 118 125 133 141 149 158 lb8 178 189 201 Net of Reinsurance Comm Auto & Other 124 131 139 148 157 166 176 187 198 210 Total Pers Auto &HO 478 507 538 570 605 642 681 722 766 813 1.815 1.855 1.801 2.Sample Financial Model BASE CASE SCENARIO Appendix A Exhibit 2 Voluntary.926 2.864 9.872 4.676 8. Work Comp 2.062 6.441 3.382 3.374 2.143 Net of Reinsurance Comm Auto & Other 462 490 520 551 585 620 658 698 741 786 Projected Other Expense ($ Millions) Voluntary.7l5 6.078 5.214 2.643 2.972 3.974 562 596 633 673 712 755 801 850 902 956 t 1 Projected Loss and Adjustment Voluntary. Work camp 1.403 9.344 Involuntary.006 3.804 2.966 2.641 2.968 2.

Sample Financial Model BASE CASE SCENARIO Appendix A Exhibit 3 Projected Loss and Adjustment Voluntary. GL& CMP Liab 23% 23% 23% 23% 23% 23% 23% 23% 23% 23% 22% 22% 22% 22% 22% 22% 22% 22% 22% 22% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% Projected Underwriting Voluntary. Expense Ratio Total Pers Auto &HO 78% 78% 78% 78% 78% 78% 78% 78% 78% 78% 82% 82% 82% 82% 82% 82% 82% 82% 82% 82% Total Pers Auto &HO -I Work Involuntary. Work Comp Involuntary. GL& CMP 98% 98% 98% 98% 98% 98% 98% 98% 98% 98% Net of Reinsurance Comm Auto & Other 82% 82% 82% 82% 82% 82% 82% 82% 62% 82% Net of Reinsurance Comm Auto & Other 81% 81% 81% 81% 81% 81% 81% 81% 81% 81% Voluntary. 38 . CL& CMP Ratio Total Pers Auto &HO 98% 98% 98% 98% 98% 98% 98% 98% 98% 98% Net of Reinsurance Comm Auto & Note: Actual input in the model had finer line detail. Work Camp Camp 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% Involuntary.

241 1341 1.981 3.319 3. Involuntary.111 Involuntary.067 3.366 4.699 1.685 8. Involuntary.143 Net of Reinsurance Comm Auto & Other 496 536 580 628 680 735 796 861 931 l.776 4.058 6.316 I Projected Voluntary.421 2.839 1.304 8.128 3385 3.421 562 608 658 712 770 834 902 976 1.451 1.963 Total camp 5.Sample Financial Model AL-IERNATIVE SCENARIO Appendix A Exhibit 4 Projected Written Premium ($ Millions) Voluntary.316 8.672 2.o+lS 1 Pm Auto &HO 1.570 1.107 1.153 2.989 2.944 3.w 3.564 7.891 3.085 Net of Reinsurance Comm Auto & Other 589 637 690 746 807 874 945 1.662 3.204 4. Work Camp 2.445 Note: Actual input in the model had finer line detail 39 .607 6.056 1.922 -r Total Comp 6. Net of Reinsurance Work Comp 501 541 586 634 686 742 803 869 933 979 GL& CMP 122 132 143 154 167 181 195 211 229 243 Camm Auto & Other 129 139 150 163 176 190 206 223 241 256 Pers Auto &HO 490 529 573 620 671 726 785 850 918 968 I Total Comp 1.490 3.620 2.382 12.003 1.736 10.093 7.103 7.175 2.949 2.519 11.282 2.534 Projected Other Expense ($ Millions) Voluntary. GL& CMP 534 578 625 676 732 792 857 927 1.469 2.591 3.035 4.729 4.720 2.109 2.067 6.023 1.182 5.353 2546 2.555 7.675 8.755 2.885 4.185 3.086 4.724 5.835 3.446 3.985 9.198 Pers Auto &HO 2.514 2.549 4.321 2.722 10. GL& Work 1 Comp CMP 2.9!% 9.

57 0 84% 84% 84% 84% 84% 84% 84% 84% 84% 84% Net of Reinsurance CommAuto& Other Work COIIIQ Comp 20% 20% 20% 20% 20% 20% 20% 20% 20% 19% Voluntary. GL& CMP Liab 23% 23% 23% 23% 23% 2376 23% 23% 23% 22% 22% 22% 22% 22% 22% 22% 22% 22% 22% 27% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% 20% Projected Underwriting Voluntary. Work Comp 106% 106% 106% 106% 106% 106% 106% 106% 106% 106% Involuntary. Involuntary.Sample Financial Model ALTERNATIVE SCENAFflO Appendix A Exhibit 5 Projected Loss and Adjustment Voluntary. 40 . GL& CMP 128% 128% 12890 128% 128% 128% 128% 128% 128% 128% Ratio Total Pers Auto &HO 101% 107 % 101% 101% 101% 101% 101% 101% 101% 100% t 1 Net of Reinsurance Comm Auto & Other 106% 106% 106% 106% 106% 106% 106% 106% 106% 105% I 106% 106% 1067" 106% 106% 106% lob% 105% Note: Actual input in the model had finer line detail. Work Comp 87% 87% 87% 87% 87% 87% 87% 87% 87% 87% Expense Ratio Total Pets Auto &HO 81% 81 % 81% 81% 81% 81% 81% 81% 81% 81% 86% 86% 86% 86% 86% 86% 86% 86% 86% 8690 Total Pers Auto &HO Camp Involuntary. Net of Reinsurance Comm Auto & GL& Other CMP 105% 105% 105% 105% 105% 105% 105% 105% 105% IO.

300 6. Value of Mortgage Backed .400 6. ! I .900 6. I I Securuites I i I 1 / f I I I I . as a result of the / I higher interest rates. / New Rates and new pattern I / I / New Rates but old pattern Difference is the result of motgage holders 1 pre-paying at a slower rate.000 6.700 5.600 Value in Millions .200 6. I I I I I I / / Base case 5.800 5.100 6.Sample Financial Model Appendix B Exhibit 1 Market 1 .500 6.

200 1.Sample Financial Model Appendix 0 Exhibit 2 Mortgage Backed Securities Cash Flow Effect of Prepayment from interest rate increase 1.000 800 --------600 Alternative Scenario I i 400 Base Case Scenario 200 0 .

000 E 0 rz 03 3.000 1.Sample Financial Model Appendix B Exhibit 3 Mortgage Backed Securities Remaining Balance Effect of Prepayment from interest rate increase 7.000 0 -.000 6.000 5.000 0 2. .000 g E 4.

Sample Financial Model Appendix B Exhibit 4 Total Fixed Income Securities Cash Flow Effect of Prepayment from interest rate increase --------- Alternative Scenario Base Case Scenario Year .

000 / 1 --------I / j I Alternative Scenario Base Case Scenario 3! 4.000 2.Sample Financial Model Appendix I3 Exhibit 5 Total Fixed Income Securities Remaining Balance Effect of Prepayment from interest rate increase 16. \ *.000 ‘.000 0 -t Year .000 14. ‘. \ *\ 12.

477 136.695) (63.334 452.871 542.956) (2.793 131.494) (1.130 116.120 Other Expens 160.739 1.371) (1.481 1.685 82j569 75.303 67.503) (3.049) (209.514) (191.126) (44.524 199.853) (16.621) (1.224 360.530) @VfY (74.360 158.Sample Financial Model Workers Compensation Insurance Operations Cash Flows (Base Case) 12/94 Runoff Business Only Appendix C Exhibit 1 Loss & LAE 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2m5 2016 2017 2018 2019 2020 2021 Total 819.219) (168.117 106.780) (454.201) (247.641 327.299) (102.650 3.586) (134.517 0 0 0 0 0 0 0 0 0 0 0 0 2.065) (974) (872) (873.178) (156.504 46.969 16.033) (1.650) (191.610 679.693 170.702 65.309 390.623) (105.170) (225.329) (66.972 874.497 97.254) (7.607) (145.525) (199.481 55.907) (179.362) (89.858 344.754) (1.126 253.182) (2.050 1.568) (96.130) (6.612 38.216.346 197.155) (1.352 23.202) 46 .747 9.502 89.450 101.534) (73.311 1.559) (185.881) (124.962) (175.648 30.161) (124.565.002 9.976) (77.706 80.152) (334.961 FIT from Underwriting (126.247) (88.185 9.703) Net Insurance Operations (957.892) (1.744 4.467) (394.115 182.284) (11.640 147.812 213.003) (762.704 308.280.199) (23.293 959 529 310 199 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 209.256) (1.741.831) (278.635 126.509) (114.212) (33.854 231.509.945) (54.930 279.

040 24.170 230.131) (10.252.786) (3.842) (210.231) (102.Sample Financial Model Workers Compensation Insurance Operations Cash Flows (Alternative) 12/94 Runoff Business Only Appendix C Exhibit 2 Premium 1995 19% 1997 1998 1999 2004 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2m3 2m4 2015 2016 2017 2m8 2019 2020 2021 Total 917.460) (34.171) (21364) (15.538 207.804 133.338) (138.304) (218.012 FIT from Underwriting (132.327 Loss & LAE 1.931) (237.499 84.802) (393.608) (56.833) (69.823) (219.600) (284.958) (92.629 1.191 196.700 116.685) (256.628) (4.815 352.966 142.816 31.466) (lim) (1.970) (122.554 152.829.222 341.862) (206.220 163.170 n9s44 582.002) (212.105 323.438.624) (12.673) 47 .621) (216.937) (225.708 57.984 3.@5) (114.837) (9.075 106.034) (147.890 10.662) (166.580 4.599 9.279) (44.624) (8.205) (2.958 125.762 277.427) (445.559 432.354) (186.067) (7X7.218 298.025 492.687) (162.281 375.079 47.030 67.473 16.633 La81 1.478) (746.808 43 0 0 0 0 0 0 0 0 0 0 0 2.167 912.467 259.446 138.804 105.021 386.284.160 664 405 263 42 30 25 20 18 3 0 0 0 0 0 0 0 0 0 0 0 221.234) (338.376 209.582 1.747 244.500) (7.277 218.164 185.538) (215.767.366) (196.664 39.404) (27.752) (157.220) (81.468) (6520) (5.116 Other Expense 167.876 10.162 174.408) (135572) Net Insurance OpeEbXlS (885.540) (176.988) (3.338) (mo97) (107.129) (104.018) (129.

000) $0 $2.000 $4.000) ($2.000 .000) ($33 I ) ($2. Income Tax” Net r ($6.Insurance Operations Base Case Cash Flows First Year and Second Year Comparison Runoff Only [3 L&LAE Exp.000 $4.027) f ($4.ow (s2.000 Runoff and One New Policy Year Premium L&LAE EXP Income Tax* Net r ($6.128) 1 (52.000) ($2.562) r ($2.000 # $6.000) $0 $2.596) I ($4.576) r Appendix C Exhibit 3 [23 First year Second Year I I $6.OJM -J 5233 I$207 ($4.

202) m~.u-.795 $6. ($6. .000) ($2. (%324)C ($1.ooO) $0 $2. ~!. Income Tax* Net -1 1$1.ooO) ($2.i ($4.000 woo0 $6WQ $8.T Base Case Premium L&LAE Exp.000 one new year Alternative Scenario Premium L&LAE Exp.000) $0 1 $8.-.000) ($4.182)= -J$264 ]$I19 ($2.152) ) .000 $6.468) \ ($167) ..c~&.:.000 $ZW $4.Insurance Operations Cash Flow in First Year *%Z. Income Tax* Net ($6.048 ($5.ooO .000) ($4..

000 z Y) .E z C 5 J 1. / 0 Year ’ n Net Cash for Insurance Operations m Cash from Held Investments I 1 .000 e : 2.000 Cash from Held investments does not include any future reinvestments.Sample Financial Model Appendix C Exhibit 5 Investments and Net Liability Cash Flows Cash Flow from Runoff Only Base Case Scenario 3.

Sample Financial Model Exhibit 6 Investments and Net Liability Cash Flows Cash Flow from Runoff and One New Policy Year Base Case Scenario 3.000 z 5 k ii. f 1. il/ I 0 r m Net Cash for insurance Operations a Cash from Held Investments i I .000 7 I e g 2.000 d + L /Cash from Held investments does not include any future reinvestments.

--~--ICash from Held investments does not include any future reinvestments.Sample Financial Model Appendix C Exhibit 7 Investments and Net Liability Cash Flows Cash Flow from Runoff Only Alternative Scenario 3.000 ______--.: --( Year Net Cash for Insurance Operations 1 __Cash from Held Investments n .___-_____ .

: . .Sample Financial Model Appendix C Exhibit 8 Investments and Net Liability Cash Flows Cash Flow from Runoff and One New Policy Year Alternative Scenario 3.000 i jCash from Held investments does not include any future reinvestments.r I 1 n Net Cash for Insurance Operations E Cash from Held Investments .

...Sample Financial Model Net cash flows under the two Scenarios Cash Flow from Runoff Only Yearly cash for base case ___..- Yearly cash for alternative scenario j .

._--Yearly cash for alternative scenario Yearly cash for base case .E b t c 2 0 0 _________~.000 ..Sample Financial Model Appendix C Exhibit 10 Net cash flows under the two Scenarios Cash Flow from Runoff and One New Policy 2.. 1. r -~-- Year .000 Year ..

Z w 3.000 Yearly cash for base case _ _ .000 i 20._ _ _ _.g .000 .c 3 k f J 2..000 1..000 e ..000 4..Sample Financial Model Appendix C Exhibit 11 Loss and LAE cash flows under the two Scenarios Cash Flow from Runoff Only 5.Cumulative cash for alternative for base case scenario -~~~ _I___ J . Yearly cash for alternative scenario Cumulative cash .

651 I Assets supporting surplus $66 Free Surplus $1.056 $163 -w 1995) $1.229 (Millions of Dollars) *This combination makes up the 12/95 runoff surplus .Return on Capital Illustration Base Case Beginning Surplus (l/95) 12194Runoff (1994 and prior) -New Year (PY 1995) Appendix D Exhibit I Return on Surplus 1995 Assets supporting liabilities Ending Surplus Runoff* t----l $26 Assets supporting SUrplUS $3.

775 New Year (1996) New Year (1996) Free Surplus (Millions of Dollars) Note: Excess runoff surplus is released first to support the new year. . and the remainder is released to free surplus.Release of Capital Illustration Base Case 1996 Beginning Surplus Before Capital Reallocation 1215Runoff (1995 and prior) Appendix D Exhibit 2 Surplus Release at (l/1/96) 1996 Beginning Surplus After Capital Reallocation 12/5 Runoff (1995 and prior) $2.

226.6% return ending Fllmlus 3.055.749.495 22.465 1.749.764 ending surplus before mw war UP& 3.6% 5.7% 12.744.404.055.224 1.416 Runoff new year free surplus Total ab!Eg&k 5175.1% -new 375.7% 5.Summary of return on Capital and Base Case release of Capital Appendix Exhibit D 3 total beginning Runoff new year free surplus Total as!u!b 2.162 (1.162.745 2.7% 6.7% 6.026.228.775.224) 650.226 1.404.303 1.404.303 1.119.679.720 1.565 1.951 6.225 1.7% Surplus allocated at the beginning capital 2.951 6.400.973 5660.495 375.52<527 of the following year Capital capital rrrlaaseri 2.416 .669 6.936 reallocated 2.921 additional 6.749:224 4.951 6.650.226.?45 1.416 6.6% 26.4% 5.667.775.

396 Runoff new year TOW .956 u beginning Runoff new year Total sur!JhE 493.070 of the following year capital ay&& 736.655 203.Summary of return on Capital and Base Case release of Capital Appendix Exhibit 0 4 w beginning Runoff new year Total 1.22i.292.976 213.097 2.679) 309.7% 6.226.r.516 6.7% 6.5% -0.021 696.007.6% 42.917 215.047 2.532 622.976 214.2% 6.2% E%ttn!u 524.050) ending &.9% Surplus allocated Runoff new year Total at the beginning capital oaadad 1.6% 17.1% additional return from new year (1.3% 5.6% 1.532 640.549 (215..676 ending surplus before new vear Uffl 524.6% 5.1% 11.LQum 6.153) 50.536 739.466 ?k.091 of the following year capiial ay&&lQ 2.166 ending surplus before new vear UPA! 1.679 1.0% Surplus allocated at the beginning capital JIQBW 472.365.466 566.226.466 capital 265.916.3% from216316 216.365.635 (623.153 666.629 additional l?LL&um 6.212 623.299.466 736.696 1.050) (1.047 capital 933.047 6.666.&m -0.466 736.416 return ending 37.3% 5.515 2.

640 247.0% oew 159.298 (141.9% 186.927 additional return foxn new ymr 3O.7% 5.553 .9% additional -new (2.096) -2.587 141.1% Surplus allocated Runoff new year Total at the beginning capital oee.670 (87.616 85.745) 26.0% 5.092 228.7% 28.1% 5. 58 capital 104.mLm 6.i36 30.065 141.340 327.7% 5...488 87.363 capital 168.587 171.545 ending surplus before ?!Lreim 6.935 233.380 of the following year capital 245.363 %.158 245.893 133.7% 6.Summary of return on Capital and release of Capital Base Case Appendix D Exhibit 5 cm beginning Runoff new year Total 150.9% 159.0% 3.5% -0.98) (2.542 245.616 87.810 of the following year capital 358.4% 16.745 331.892) 16.2% new 186.778 Runotf new year Total CA beginning Runoff new year Total 174.dad 190.156 return ending ?LLcQuQ 6.776 358.8% 9.256 Surplus allocated at the beginning capital 140.752 308.645 ending surplus before Y!izaum 6.610 82.363 358.436 ending % return 22.

946 (469.7% 5.212 capital 647.953 31.832 ending new 5.266 489.159 ending surplus before flew Year UN 340.130 14.505 capital (1.343 813.2% -68.9% -54.9% 24.8% additional from(10.YAwle 9.505 9.2% 18.599) (15.008 additional 7.557 19.9% Surplus allocated at the beginning capital 10.008 632.2% 5.mml 6.247) (14.609 of the following year capital ElYabuQ 972.Summary of return on Capital and Base Case release of Capital Appendix Exhibit D 6 Other Bus.2% 36.351 of the following year capital 8.2% 5.752 14.8% 487.945 827.505 return ending YL.603 Runoft new year Total I .343) 158.182) -73.7% 5.7% 6.b 340.2% -48.259 return ending s!4tP.212 972.212 7.259 144.182) (10.599 25.848) Runoft new year Total beginning Runoff 317.5% Surplus allocated at the beginning capital naada 324.3% from144.687 surplus before ?!Qmhlm 6. beginning Runoff svlplus 4.204 972.130 41375 9.

613) 104.108 630.603 176.714 171.694 625.498 .6% return ending ycplus 453.7% -new (4.002 additional IYLrawa 6.389 166.216 of the following year capital 625.714 625314 capital reiaasarl 281.7% -2.111 (176.820 ?!iJQtm 6.894 176.Summary of Return on Capital and Base Case Release of Capital Appendix Exhibit D 7 HO 6.288) -0.5% 5.?luRb 426.5% 6.5% 3.613 521.1 % (4.2% 5.288) 5urplos allocated Runoff new year Total al the beginning capital tEedad 344. Other beginning Runoff new year Total t.043 ending surplus before new vex UMI 453.654 593.

7% 5.158 _ .4% 0.172.684.951 6.226 1.495 ending w 22.055.522 5.7% additional return tom new VW 375.585 I .225 1.495 375.744.2% 0.141 1.720.784 ending surplus before oew 2D487.585 ending surplus before new v*ar UNV 3.921 ?!Lceml 6.1% Alternative beggining Runoff new year free surplus Total 2.228.119.055.8% 0.318.867.850.297 1.680.028.714 2.162.8% -6.745 2.6% 28.4% Case additional return from new veilc 402: 58 % return 24.973 5.524 5.1% ending zzua!ds 2.377 ?ILxmuo -73.6% 5.4% 25.745 1.7% 6.4% 7.228 7.720 6.068.4% 402.7% 3.Comparison of return on Capital between base case and alternative case Appendix D Exhibit 8 Total Base Case beggining Runoff new year 2.867.8% 0.535 -13.650.774 7.481.172.

299.978 213.488 588.6% 42.698 1.798.299.3% 5.047 zLLL&Ko 6.Qm “SW ma r4hcetum (1 .0% .418 return ending 1.9% 0.516 before z4-rs&n 6.021 696.205.698 1.5% -0.097 2.016 41.486 yQLeuL0 6.385.0% Alternative beggining RU”O!f new year Total 493.186 ending surplus before -7.418 218.007.978 214.855 203.6% Alternative beggining Runoff new year t2 Total s.515 2.227 634.Ja!m -7.Comparison of return on Capital between base case and alternative case Appendix Exhibit D 9 beggining Runoff new year TOtal s!J@4s 1. (1.3% 245.532 840.855 203.434 839.8% -8.876 ending surplus before !T6l&ro -12.6% 5.888.205.9% new 430.8% 13.8% 17.315 -12.1% 11.ia!IE 1.3% 0.2% 6.2% ending §!ami 524.865 245.886.538 739.226.016 593.3% 42.1% additional return fr.3% 5.050) -0.oso.688 634.3% additional -new 218.021 696.488 738.418 Tk.6% -9.9% 0.533 1.776 Case addiiional -new (461) (461) -0.488 588.186 ending surf~lus before z!-zdam 6.549 Case additional -new return ending svrplus 1.7% 6.8% -4.2% -0.044.7% new 1.629 37.876 ending surplus oew 524.7% 6.6% 8.532 622.088 204.0% 2.1% return ending fil!m!m 430.9% oew 13385.885 a Base case beggining Runoff new year Total !2lLc&2 493.088 204.

0% new 152.545 ending surplus before T!YumKo -9.0% 0.617 220.6% .2% -new 30.436 22.850 286.7% 5.400 217.9% -1.4% 16.6% sYuQi!s 152.7% 6.935 233.0% 5.8% 9.587 141.5% new 136.545 ending surplus before !cQauLn 6.821 z!%Lwum -13.776 358.1% sJGl!a174.363 T&.0% 3.LQul 6.542 245.720 322.Comparison of return on Capital between base case and alternative case Appendix Exhibit D IO Runoft new year Total beggining surplus 150.9% 22 -3.&8) (2.8% 11.0% 27.368 z4-wum -9.1% 0.8% -5.256 -2.752 308.640 247.7% 28.1% -3.158 k!kaAMO 6.693 133.935 233.870 35.0% new 159.645 new zL?xmm 6.616 87.4% I I Runoff new year Total CA Base Case baggining ending surplus 186.752 308.217) return ending surnlus 136.9% Alternative beggining Runoff new year Total svrplus 150.7% 5.585 Case additional from(3.870 ending m 26.9% Alternative beggining RU”Ofl new year Total suozlus 174.610 82.645 ending surplus before !h-mkla -13.101 134.968 80.968 83.8% -7.927 before additional return ending ?L!mlG 166.098) return ending ylrplus 159.587 171.217) (3.9% additional from(2.616 85.1% 5.101 170.610 82.1% -6.951 Case additional return m new “*a 35.436 30.340 327.893 133.5% -0.6% 4.

8% Case additional -new (7o.204 972.9% -54.632 ending w surplus before war Q& 5.5% return ending 2%!da!s 340.6% -61.259 31.3% 0.Comparison of return on Capital between base case and alternative case Appendix Exhibit D 11 I 5 Other Bus.375 9.2% 16.5% 0.807) -78.541 667.110 465.tmb 317.062 13.008 632.ao7) (10.869 17.130 14.082 7.951 surptus belore z!dQbdm -15.651 before !l?cdan -36.110 606.945 827.9% Alternatlvs beggining svrplus 4.159 461.2% 36.6% PA Ease Casa beggining Runoff zJJ.949 return ending 2%uB!Ys 202.5% 18.3% 31.259 144.7% 5.mb 317.159 461.749 778.749 776.l82) (10.8% additional return tromnewvear (tO.776 18.008 487. Base Case beggining Runoff 4.2% -68.490 808.3% Case additional tromnew 140.7% 6.687 yQL?m 6.953 before 7.1% .7% 5.3% 3.9% -3.062 3.2% 5.949 140.505 ghrefvrn 6.9% -77.4% -58.8% 30.6% new year Total Alternative beggining RU”Off new year Total i5.8% -14.144 -15.2% 5.557 19.182) ending I?!LEam -73.908 ending surplus 202.2% -48.1% return ending Ysm2!@ 4.608 ending oew 4.600 !4sIuul -36.832 new Total year 13.130 4.3% additional fromnew 144.9% 24.908 ending m surplus 340.212 %x!bm 7.

6% -12.&g 6.288) ending ?Lc&r~ -2.106 630.8% -13.288) Case additional return from new veaf (6.714 n 6.063) Alternative beggining Runoff new year TOhI 2Lurms 426.960 513.5% (4.654 593.&K0 -17.1% 5.023 519.392 y&.4% ending YdeUr13 -3.6% -1 .7% s!JaLLs 453.002 Q.694 176.694 171.2% additional return from new vex (4.043 ending surplus before new war u/w 351.329 m -17.6% -2.O% suD!us 351.654 593.369 168.4% .389 166.820 625.043 ending surplus before new war urn 453.063) (6.6% 0.5% 5.369 166.369 161.6% -0.Comparison of Return on Capital between base case and alternative scenario Appendix Exhibit D 12 HO 8 Other Ease Case beggining Runoff new year TOElI ctut!G 426.5% 3.7% 6.

787) (1.681 14.681 R‘Zhll7l Rate of Return 12.0% 5.257) Discount Factor Item Premium Loss & LAE Other Expense Net Return from New Policy Year 5.&h 6.4% Market Value 0 12195 Fixed income Securities Stocks Unrecognized FIT on Investments Other Assets Collectable Premium Loss & LAE Other Expense Net Market Value Net Value of Investments &Other Net Value of Insurance Op WdiOIlS 786 125 (55) (6) 162 (699) (19) 55 348 849 W) Additional Return from New Policy Year (1995 PY) Nominal Value Dee-95 Market Value 5.6% 88.8% Summary of Results 1 Market Value Q 12194 All Items 5. (360) lsm 5.4% 5.1% 5.404 724 Note: Actual input in the model had finer line detail 69 .864 (4.4% 5.942 1.4% 5.210) 4 375 99.839 (4.200 (79) m (lag.7% Market Value 0 12195 6.201) &4 (81) (1.6% 10.9% 100.Sample Financial Model Appendix D Exhibit 13 Illustration of Return Overall Company Base Case Scenario I item Return on Runoff Before any New Business Market Value @I 12194 13.7% 5.973 (9.292) Return Rate of Return 5.4% 7.

4% 0.5% 0.) (34 414 (362) 123 hw (9.292) Additional Return from New Policy Year (1995 PY) Nominal Value Dee-95 Market V&le 6.182) (1.418) (1.681 14.3% 100.973 RetuiTl Rate of R&UTl Market Value 0 12/95 13.3% 85.318 15.058 (5.942 1.201 259 (96) (12Z) (394) l&s 5.1% 7.2% (6) 336 (1.016 (4.960) (360) l‘Ql8 5.5% 4Qzxz -6.241) xXxX (365) 99.w9 (12. 70 .8% 5.732 1.3% 9.4% Note: Actual input in the model had finer line detail.246) sn 402 Discount Factor Item P1emiUm Loss & LAE Other Expense Net Return from New Policy Year 6.207.778) Fixed Income seturities stocks Unrecognized FIT on Investments Other Assets Collectable Premium Loss & LAE Other Expense Net Market Value Net Value of Inveshnents &Other Net Value of Insurance Operations (W 1 339 -1.096 (9.Sample Financial Model Appendix D Exhibit 14 Illustration of Return Overall Company Alternative Scenario Return on Runoff Before any New Business km Market Value @ 12/94 13.200 (79) w9 3.2% 9.0% 11.

160 1. and other miscellaneous items.298 million.651 1.163 (9. Liabilities SUl-plUS Runoff New Year Free Surplus 12. we allocate assets to support the runoff surplus. and free surplus.867 1. is allocated to the runoff surplus. with a starting market value of $9. This allocation procedure results in the following “summarized balance sheets”. Second.651 1. Why? By definition.292) 0 0 2. How are assets and liabilities underwriting): allocated? Recall the stated Results (Before new year To get an explanation as to why the return on the runoff surplus is different from the return on the other surplus. The second is because the after-tax rate return on investments is different from the after-tax rate of discount in the reserves. both of which need to exist simultaneously. Composition of the Beginning Surplus Investments Ins. The rest of the assets are then allocated to supper! each of the remaining surplus items: new year surplus. two of which go to the runoff surplus. we allocate enough assets to support the net insurance operations liabilities. These two portfolios are allocated to the runoff surplus. can thus be broken into four parts. Ops and Misc. one needs to look at the composition of assets and net liabilities in each of these segments. First.Sample Financial Model Appendix D Exhibit 15 Analysis of Returns (Before New Business) Illustration of Base Case Scenario for the Overall Company The difference in returns results from the combination of two factors. The first is because of how assets and liabilities are allocated. Investments. the runoff surplus is dedicated to support all of the insurance operations that were generated prior to the evaluation date. all of the net liabilities.163 71 . At the beginning of the year. which is the net liabilities from the insurance operations.

72 .4%) * (9.681 (6) new year surplus (7) free surplus [(S) total surplus Note that a real quick way to double-check the reasonability of the results: Return on runoff surplus = 5. then the return from the assets supporting the liabilities. Reserves are discounted at 5. the after-tax yield rate of a selected bond portfolio.651 1.681)) = 6.292 / 5.5.6% 5.7% return as the return on assets supporting surplus.7% 5. Return on total surplus = 5. we shall call the 5. the difference between the investment rate of return and the rate of discount in the reserves.7% . However since the rate of return on investments is higher than the discount rate. If the rate of return on investments and the discount rate were equal. The table below reconciles the returns. For convenience.867 1. The following are examples for some selected lines.163 5.1%.l%j (9. would be exactly offset by the unwinding of the discount in the liabilities.7% .4%) as the return on assets supporting liabilities. However. the actual investment portfolio was projected to yield 5.7%. One needs only the ratio of liabilities to surplus (all on a market value basis).292 Dollar !?%eum 527 (501) 26 163 189 94 66 348 Rate of lL6sm.3% (5.4%) * (9. in addition to the above returns.292) 0 2.Sample Financial Model Appendix D Exhibit 16 We now turn to the second factor.7% .7% + ((5.4% 5.867 2.292 / 2.4%. the unwinding of the discount will not completely offset the return on investments. and the 0. The above short cut can be used to confirm the computation for each of the lines. Value of m (1) (2) (3) (4) (5) Assets supporting liabilities liabilities Net = (1) + (2) pure runoff surplus runoff surplus (3) + (4) 9.867)) = 6.7% 6.5.5. mainly because this portfolio includes common stocks with a projected after-tax return of 6.7% + ((5.7% 6.6%.n 5.7%.7% 5.

Sample Financial Model Leverage Ratios Market Values of Insurance Allocated Surplus Operations Workers Compensation General Liability Personal Auto (1) Return on Investments (2) Insurance Operations Discount (3) Return on Assets Supporting (4) Return on Assets Supporting 4.3% 7.745 1.9% 0.6% 6.429 1.7% 0.7% 5.2% Workers Compensation General Liability Personal Auto * Return on Assets Supporting Liabilities yield of assets supporting liabilities l 3.3% Appendix D Exhibit 17 Liability-toSurplus Ratio 3.408 2.299 494 317 5. 73 .4% 5.7% % Return on Surplus 6.7% 5.7% 5.(2) Return on the Runoff (Base Case) Liability-tosurp1us Ratio Return on Assets supporing supporing Liabilities surplus I *t 0.503 Rate Surplus = (1) Liabilities = (1) .260 5.6% 1.260 5.503 = product ratio and the net * Return on assets supporting surplus = net yield of assets supporting surplus.5% of liabilities-to-surplus 5.408 2.116 1.

In the illustrated scenario. True. The only exception is that a portion of workers camp indemnity reserves. except for acquisition expenses. All the changes above will impact future earned premium and incurred loss and expenses. Company managers do not increase and reduce the reserve levels by such levels just because inflation has gone up. From a market value perspective. held reserves affect only the timing of income taxes. In our scenario. and should similarly increase. do not calculate reserves under different inflation scenarios. The change of 200 basis points in interest rates causes a change of 130 basis points in the discount rate. since the discount rate is an after-tax rate.Sample Financial Model Analysis of Change in Inflation and Interest Rates Impact on the Rehnn on Runoff Surplus Appendix D Exhibit 18 In the illustrated alternative scenario.8% in the alternative scenario. we assumed that both inflation and interest rates jump by 200 basis points and remain at the new level. All other expenses. First. in the non-COLA states. The increase in premium is dependent on the retrospective contracts. The tax implications from these earnings and losses are considered in the model. The standard actuarial techniques do not consider inflation implicitly. Inflation is explicitly projected in the standard techniques. the retrospective premiums are sensitive to loss experience. the scenario above assumes that the nominal held reserves increase at the rate of inflation. There are other consequences that are considered in the model. are not sensitive to chnages in inflation. to 0. rather the difficulty is in attempting to mimic the real world. Unfortunately. to loss and was heavily age at which the loss development Second. but not necessarily at the same rate. We had discussed previously the impact on investments from a 200 basis points increase in interest rates. since one can easily quantify such impact. WC nominal reserves ultimately grow 14% because of the 200 basis points jump in inflation. 74 . one needs to consider the actual reaction of held reserves to the change in inflation outlook. Overall. grow at the rate of inflation. Third. as a result. Because of these considerations. the sensitivity of retro premium to loss experience dependent was based on the company’s on the relative historic premium sensitivity occurs. the after-tax return on investments drops from 5. The difficulty stems not from theoretical reasons. but these techniques average out inflation throughout the life of the policy or accident year and. The model assumes that the nominal paid loss and adjustment expenses increase at the 2% annual rate. the reaction of held reserves to changes in inflation are not easily quantifiable.7% in the base case scenario.

9% 4. FI-I 411 Premium Workers Compensation 158 L&LAE General Liability Personal Auto Total All-Lines 34 WA (1.9% (278) (78) W) (59O)l Total Change in Return Insurance Operations Investments Workers Compensation General Liability Personal Auto Total All-Lines Total Change as Change % of Surplus (38) lb (278) (78) WJ) (590) UW (95) (138) (574) -13.0% 75 .5% -20. Exp.2% -43. Change in Nominal Operations 0th. Exp.320) (158) (132) (11) (1) (1) (15) 44 47 556 7 227 (L8C’o) Operations Change in Market Values of Insurance Premium L&LAE 0th.Sample Financial Model The following exhibits will illustrate the change in return on the runoff Values of Insurance Appendix 0 Exhibit 19 for some of the lines.160 -4. FlT Workers Compensation General Liability Personal Auto Total All-Lines 112 27 1 (247) (74) (78) (503) (11) (1) (1) (15) 243 32 40 359 174 Change in Market Values of Investments Beginning Investments % Change in Return Change in Return Workers Compensation General Liability Personal Auto Total All-Lines 5.728 1.9% -4.9% -4.062 12.610 2.9% -19.

....Market Value (Alternative) ---__---Statutoty Surplus (Alternative) .Smaple Financial Model Appendix E Exhibii 1 Statutory and Market Value Surplus Runoff With No New Policy Years 12 12 0- l 1996 1997 1998 1999 2000 2001 2002 2003 1995 i-o 2004 2005 Year-end Market Value (Base Case) Statutory Surplus (Base Case) .

.Smaple Financial Model Appendix E Exhibit 2 Results 18 Statutory and Market Value Surplus Reflect Ten Additional Years of Underwriting 18 6 6 2 J- .- .- 4 2 1995 1996 1997 1998 1999 2000 Year-end 2001 2002 2003 2004 2005 Market Value (Base Case) ~ Statutoty Surplus (Base Case) -s--m Market Value (Alternative) --------- Statutory Surplus (Alternative) 1 1 I .-.A m.__ _/ ---.+ _ .- ~cm---~-------:-.-bp-L .

192 CMP 150.316 Other Lines * 922.986 Total 2.1 3.898.960.821 696.361 l/l/96 Information WC 6.146 5.630 2414.697.930 170.960 for l/1/95 Capital Allocation GL CMP 1523.460.1 Other Lines * (Mixed) (Mixed) T&II (Mixed) (Mixed) Appendix F Exhibit 1 Capital Allocation WC GL Dsic.790 Capital Allocation for l/1/96 Capital Allocation CMP GL 1.121 l&%.775.451 Allocated Capital GL 472.005.462 82.Sample Financial Model Capital Allocation Rules for Base Case Scenario Rules CMP 2.863.892 228.780 180. .879 1.866.8 2.109 12/94 Discounted Reserves 1995 Planned premium Other Lines * 4.368 775.070 12/95 Discounted Reserves 1996 Planned premium Other Lines 4.527 Other Lines Capital Ratios varied by specific lines.686 522.415 l&48.538 Total 2.303 1.749.9l6.114.213 588.918 Runoff Business New Policy Year Total Capital l/1/95 WC 1.524.935.298.241 363.676 417.854 233.300 1.917 215.091 CMP 140.220.966 4.423 l Total 12.513.508 390.371 202.444 Runoff Business New Policy Year Total Capital l l/1/96 WC 1.691.864.335 Allocated Capital GL 493.561.3 4.119.547 3.945 342.723333 6.919 Total 12. Reserves-to-surplus ratio Premium-to-surplus ratio 5.292.738 2.488 87.212 623.1 1.153 688.170 1.7 l/495 Capital Allocation Information WC 6.224 4.380 Other Lines * 869.895 2.

Sample Financial Model Appendix F Exhibit 2 Workers 10% 9% Compensation Variability in the Discounted LOSS & LAE Reserves 90% 195th Percen$ 80% - e~-:~ /’ -- I 50% Ratio of Simulated-to-Expected Results .

Sample Financial Model Appendix F Exhibit 3 Workers Compensation Variability in the New Year’s Results 4% 3% 2% 1% 0% Ratio of Simulated-to-Expected Results .

233 1.193 2.042 2.~ 3 4 .Impact of Medical Inflation Illustration based on a Fifteen-Year Triangle where Me Dollar amounts and the payment patterns are the Fixed Basis for Further Simulations Appendix G Exhibit 1 1 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 19989 1990 1991 1992 2 3 4 5 Payments m Nommal Dollars ($M’ IlliOnS) 6 7 8 10 9 11 0 0 0 -rw 0 . j I- : .502 1.g.434 263 114 285 125 311 138 346 153 383 171 427 186 46.409 1.g+ 1 2 131 56 / .5 197 493 no 524 226 564 240 601 256 640 -.323 2.311 1.161 1.600 UC 1.-O O 0 0 _. 12 13 0 0 0 0 .878 2. 0-r-~ 14 0 TF 0 0 0 O ’ 15 I . 276 689 J 301 327 817 351 877 372 929 389 974 r J 50 55 61 68 74 79 90 22 24 27 30 32 34 38 10 11 12 13 13 14 16 4 5 5 5 6 6 7 2 2 2 2 2 1 1 1 1 0 0 0 .fi 1 1 1 0 .067 1. 0 0 0 1993 1994 600 6. .58 Tl3 778 864 956 1.

OGO 1.ooO 100.000 rJEJ -.053 1.050 1.OOOJ-iXS 1.ooO 1.087 96.512 304 3.004 1.4% 0.0% Y-28.O% 100.045 3.434 1.002 1.0% 0.227 92.2% 1.015 1.0% 100.547 1.0% 0.134 1.001 l.296 2.l.000 1.002 mn l.050 1.429 1.000 LOCO 1.002 l.127 1.052 1.0% 0.009 1.cOl l.049 r1.O@l 1.424 1.152 2.035 98.9% 0.0% 100.435 1.M)o 1. Pay Pattern 1.4% 99.7% 99.447 122 3.020 1.132 1.815 3.020 1.306 3.022 2.444 1.oou 1.266 2.052 1.8% 0.090 1.000 Loo0 1.5% 24.048 1.136 1.Dol I ~-1.0% lM).128 1.004 -1.044 3.ooO 1.OOQ l.426 1.000 1.006 99.002 l.434 1.CQP l.001 .004 1.547 0 I I .130 1.ooO J i Pay Pattern 57.0% 100.020 l.048 1.0% 0.000 1.151 0 2.022 0 1.000 1.000 1.460 1 2.000 1.0001 1.132 1.004 1.OO+l 1.ooo l.048 1.876 0 1.004 1.000 1.impact of Medical Inflation Illustration based on a Fifteen-Year Triangle where the Dollar amounts and the payment patterns are the Fixed Basis for Further Simulations Appendix G Exhibit 2 1-2 1976 1977 2-3 3-4 Loss Development Factors Based on Nominal Payments lo-11 4-5 5-6 6-7 7-8 8-9 9-10 11-12 12-13 1.001 1.876 1.002 l.5% 1.000 l.753 81.437 1.008 1 -1.Ooo 1. 1.569 3.008 1.627 3 2.129 1.438 1.051 1.000 ix00 1.009 l.049 l.811 8 2.1% 57.002 1.5% 1.435 1.049 1.004 ::ztz .818 2.021 1.000 1.022 1.OiM 1.000 l.001 1.630 2.436 1.cQ4 1.cJMxM1.426 1.127 1.000 1.0% 0.026 19 2.000 1.COO 1.002 1.X 1994 1993 1992 Calculation of indicated Reserves (Nominal Dollars) 1991 1990 1989 1988 1987 1986 1985 1984 1983 1982 1981 1980 Ultimate Paid-to-date Reserves otal Keserves 4.Doo 1.430 1.053 1.000 1.127 1.020 1.009 1.OGU l.1% l.443 1.000 -- l.707 0 1.431 _11.009 1.001 1.004 1.010 1.000 r J 1.132 1.832 3.020 1.128 1.297 748 3.9% 0.430 1.002 1.001 l.LKlb 4.001 l.O2O_J1.000 1.425 1.020 l.419 r r1.047 [ 1.125J 1.risz 1.1% 1.135 1.022 1.003 1.OOl l.oof3of3?xi4 J 1.5% 2.425 1.707 1.004 1.0% Five Year Averages 1.OQO 1.002 r1.020 1.000 1.009 1.000 l.461 2.295 0 2.0% 10.022 1.000 1.002 1.001 99.000 1.0% 0.022 1.050 1.004 rmC 1.000 1.021 1.446 1.5% 1.020 pc9 1.049 1.004 1.021 1.ooaa 13-14 14-15 1978 lY7Y 1980 1981 1982 1983 1984 1985 1986 1987 1988 1P89 1990 1991 w 1992 1993 1994 Link Ratio Factor to Ultimate Cum.129 1 1.004 J 1.009 1.6% 4.000 1.090 Loco 1.137 1.050 1.010 1.257 49 3.009 1.008 1.129 I.

192 i.854 1.973 1.727 2.727 1.620 1.048 1.127 2.412 1.192 1.521 1.296 1.280 2.29a .415 3.192 1.ooo 1.545 2.127 2.192 1.521 .2% 1.110 1.048 1.620 1.E~ J 1.296 1.620 4 3.727 1.000 r 1.973 1.110 I.048 1.4121 1.OaO J 1.Impact of Medical Inflation Illustration based on a Fifteen-Year Triangle where the Dollar amounts and the payment patterns are the Fixed Basis for Further Simulations Appendix G Exhibit 3 1 1976 1977 1978 1979 1980 1981 2 3.096 1.727 1.280 2.727 1.973 1.412 1 1.096 1.110 1.048 1.727 1.856 1.856 1.110 I.110 1.620 1.110 1.192 1.545 2.856 1.973 1.973 1.296 5 6 Inflation Factors (19% $Dollar) 7 8 9 10 11 12 13 14 15 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 4.oQo I ~-~ 1.545 2.048 1.545 2.521 ~_~ 1. I .701 3.048 1.192 l.280 2.110 1.817 2.973 1.412 1 1.620 1.@30 1.856 1.521 1.127 2.412 j 1.521 2.973 1.:.192 1.521 1.110 1.415 3.048 l.192 1.412 1 1.620 2.856 1.:.048 l.1.521 1.$A!.521 1.096 1.521 1.110 1.817 2.521 I.192 1.280 2.280 2.973 1.280 2.727 1.048 1.296 ~.192 1.296 1.048 .048 1.192 1.412Jx296m 1.817 2.1.z 1.280 2.412 1.000 r I 1.110 1.856 1.110 1.412 1.817 2.620 1.701 3.727 1.m 1 1.545 2.545 2.856 1.4giis1.521 1.521 1.048 3 3.412 j 1.094 1.727 1.127 2.096 1.096 1.620 1.727 1.415 3.296 ' 1.817 2.620 2.110 1.620 1.856 1.z .620 1.096 1.620 1.096 1.856 1.2% 1.057 3.973 1.727 1.048 1.192 1.

0 i I : ’ i i .434 974 2.434 974 2.434 974 2.434 974 2.434 21434 974 974 974 2.Impact of Medical Inflation Illustration based on a Fifteen-Year Triangle where the Dollar amounts and the payment patterns are the Fixed Basis for Further Simulations Appendix G Exhibit 4 I Year 1976 t- 1 2 3 4 5 Payments in 1994 Dollars (5millions) 6 7 8 9 10 11 12 13 14 15 2. 0 I . 0 0 ro.434 974 2.434 974 2.434 1 974 2.434 974 2. 1 1 .434 974 2.434 974 2.434 974 2.434 2.434 974 2.434 974 2.434 1 389 389 389 389 389 389 389 389 389 389 389 389 1 389 389 389 389 389 ) 156 156 156 156 156 156 156 156 156 156 156 1 156 156 156 156 156 1 62 62 62 62 62 62 62 62 62 62 1 62 62 62 2s 25 25 25 25 25 25 25 25110 1 ~25 25 25 25 10 10 10 10 10 10 10 10 10 10 4 4 4 4 4 1 i 4 4 4 I 2 2 2 2 : . I ..434 974 2.434 974 2. 2 I 1 1 1 0 0 0 0 : E.

001 p0 1.114 1.114 1.057 4.0% 9.400 1.000 1.006 1.041 1.000 [ l.623 L.002 99.041 1.000 l.002 1.400 1.lM~j -.041 1.oco l.016 1.8% 99.000 1 1.010 1.2% Reserves 0.9% 1.PayPa&m Pay Pattern 60.002J 1.001 1.016 _fl.000 l.ooO 1.041 r 1.UUU 100.016 1.8% 1.001 l.000 1.ooo l.0% 1.016 1.OiXJ 1.050 4.400 l. 12-13 13-14 14-l: 1976 1977 )I978 1979 1980 1981 1982 1983 1984 1985 1986 1.057 7 1987 4.057 0 1980 Ultimate Reserves ‘otal Kesen’es 2.041 1.057 0 1982 1981 4.0% 0.001 l.oOa l.114 1.016 1.000 1994 Link Ratio Factor to Ultimate Cum.114 1.057 260 3.000 1.0% Five Year Averages 1.006 1.002 1.016 1.057 649 3.114 1.400 r 1.016 1.000 1.400 1.m 1.016 1.001 1.016 1.C@l 1.6% 3.mo l.057 1.002 1.om 1.000 I-1.0% l.057 42 4.041 1.041 I.400 1.a57 0 4.114 1.001 looa 1.@36 1.ooil 1.007.006 1.0% 0.054 4.016 1.057 17 4.041 1.I01 l.0% 0.006 1.006 1.000 I.000 1.000 100.0% 60.041 1.m 1.002 1.004 100.400 1987 1988 1989 1990 1991 1992 1993 1.056 4.400 1.057 0 .000p1.114 1.001 1.002 1.057 4.ooo l.041 I .ooo --1.O06 1.041 1.057 4.000 1.114 1.400 1.5% 0.000 f 1.000 J 1.000 1.400 2-3 Loss Development Factors Based on Payments Converted to 1994 $Dollars 4-5 3-4 5-6 6-7 7-8 8-9 9-10 lo-11 11-12 1.000 1.057 4.400 1.KKl 100.114 1.114 1.190 93.400 1.016 1.016 1.4w 1.114 p.400 1.ooo 100.ooO 1 .114 1.@96 1.000 1.6% 99.ooo 1.6% 1.!?QO l.001 1.002 1..667 1.mJu 100.006 1.0% 0.434 4.O% l.041 1.001 lM).0% 24.041 1.ooo l.1% 0.057 104 4.4ooJiXT 1.114 1.ooo 1.u41 1.M?O 1.0% I1.006 1.OOo] 1 .000 pl.4&I 1.001 l.0% l.400 1.m 1.000 l.OOO 1 .026 99.ooo 1.0% 1.006 1.002 1.’ 1.057 0 1983 4.114 1.ooo l.SU Paid-to-date 1994 1993 1992 Calculation of Indicated 1991 1990 1989 1988 4.040 4.057 4.Ooa 1.057 3 (in 1994 Ddlars) 1986 1985 4.016 1.000 1.000 l.016 1.M)O 1.000 mail 1./U4 1 3.797 4.004 l.016 1.404 1.4% 1.I00 1.001 1.0% l.041 1.m __-1.000 l.0% 0.Oca l.041 J 1.000 1.041 1.002 1.002 1.004 I.0% Y.953 4.114 I.0% 0.408 4.001 1.056 4.000 1.6% 0.016 1.002 1.006 1.impact of Medical Inflation Illustration based on a Fifteen-Year Triangle where the Dollar amounts and the payment patterns are the Fixed Basis for Further Simulations Appendix G Exhibit 5 Year 1-2 1.068 97.114 1.057 1 4.0% 84.m 1.057 0 1984 4.000 r 1BOl~ 1.ooOJ 1.000 l.015 4.

623 649 260 104 42 17 7 3 Reinflated 51 61 67 Payments 21 9 26 11 29 13 1 0 0 0 0 0 4% InfIation 8% InfIation 10% Inflation 2.688 1.752 1.187 1.890 3.izM14zap5zMM~~ 2.704 1.Qhl199514e6mlspB19sszMM~zLMzzMz.Impact of Medical Inflation Illustration based on a Fifteen-Year Triangle where the Dollar amounts and the payment patterns are the Fixed Basis for Further Simulations Appendix G Exhibit 6 f XkaE 1994 Dollars Expected Payment of Reserves in r.765 702 757 7135 292 327 346 121 141 152 4 5 6 2 2 3 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 .085 3.

089 1.117 1. ralherthan(olhe ag~~~agelaclors correb#on among 6% &vwiopmentsi .96 1.075 1.167132 1.015 1.246 2.245 1.60 60-72 72.473 I.014 Wemselves.977 oec 1978 DE 1979Dec 1980 oec 1981 oec 1982 oec 1983 oec 1984 Dee 1985 Dee 1986 0% 1987 Dee 1966DSC .033 1.068 1.062 l.520 1.027 I.iOd 1970 Dot 1971 Dot 1972 oec 1973Dcc .304 bated kr.055 1. 132 1.i54326 1.025 1.074 1. 120 1.032 I.041 1.03 0.017 1.059 1.025 1.019 1.386 1.020 1.496 1.025 1.612 1.231 1.026 1.395 1.116 1.031 1.192 1.obB 1.098 1.262 2.059 7 -3.g”omlal PalameferS:+ ll7" Ok sigma 0.048 Is!48 1.400 1.and arederived by invrd.193 1.044 1.025 1.88 0.043 1.093 1.16 0.016 1.074 1.169 2.174078 1.1he disidbulion is citto(ATA.057 1.21.144 1012 1.027 1.182388 1.033 I.027 1.33 0.004 1. We USedin B Mo~~wZMIO simulalion.063 6 -3.022 1.464 2.019 9 -3.64 1.101934 Simulated ATA" 2.109 1.031 1.21 0.038 1.026 1.228 1.502 2.36 36-48 1.62 O.195 1.013 1.096 1.044 1.021 1.051 1.016 1. II -4.397 1.113 1.364 3 -1.136 5 -2.496 2.115 1.020 1.889Dec 1990 O%C 1991 oec 1992 Ilee 1993Dec i994 Dee 1 Lc.388 1.409 I.046 1.19.024 1.018 1.034 1.016 1.017 1.529 2.666 2.191 1.017 1.12574l I.013 f.046 1.012 1.Q32 1.013 1.055 1.167 4 .437 1.021 10 -4.041 1.022 1.126 1.047 2.263 2.227 1.010 1.974 ooc .060 1.026 1.017 * ~ognormal parametersare a* the ~mukledage-lo-age on lilting B logno~aldisltibulionlo the column olage.106 1.422 2.345 2.452 2.038 1.020 1.207 1.056 1.068 I.016 1.310 2.023 1.023 1.185 1.164 1.69 0.019 1.017 1.019 1.r lndcmnlty an* Audi 48.470 84.094 1. To get a better 111.016 1.037 1. DeYe.027 1.021 1.030 1.066 1.132060 1.426 2 1.018 1.041 1.043 1.023 1.068 1.031 1.070 1.020 1.131 1.179 2.037 1.56 0.065 1.113606 I.016 1.046 1.226 1.065 t.190 1.607 1.ow 1.055 I.Ot6 12-a 24.140 1.124266 1.82 0.407 1.039 6 .127 1.168 1.040 1.109 1.133 1.334 2.ng the C~m~latiw Density of ~h~lognormal.398 1. dillerentmalurities.046 1.053 1.a~umingno I).496 1.049 I.209 I.026 1.064 1.034 96.159 1.203765 l.026 106.500 1.opment Faclors 1c.385 1.976 0% .3.022 1.011 1.oao 1.btS 2 9.093 1.018 132.024 1.019 1.020 1.030 1.076 1.237 1234 1.121 1.199 2.167487 1.377 2.2.140 1.134 1.976 ooc .400 I.473 1.232 1.069 1.020 120.111 1. 2.192 2.036 1.016 1016 1.107 1.00 0.to-age fackxs.019 1.Appendix H Exhibit 1 Page 1 PC?.

007 1.006 1.W? 1.010 t .23 0..1 I.01.021 1..007 I.w9 1.w5 12 -4.014 1.013 1.013 14 4. 13 -4.013 . I.64 0. I.264 1.004 1.coa 204.268 1.01.006 1.965613 1.007 1.007 17 -5.19 0.Appendix H Exhibit 1 Page 2 (44.012 1 .012 I.156 1.I07 ma 276.01.190354 .95 0 226962 1.013 1..309 ~.011 1.o.013 1.010 1.4.279641 l-006 IS -4.005 286 .045 252.4 1.008 1.011 1.01.044 1.004 1 .013 1.013 1. 1.Wl 1.012 1.oo7 23 24 . 1 ma 1.wa l.012 1.011 156-166 1.005 I.W6 21 -5.MM 216-228 1.009 I .007 240-252 1 .006 1.004 1.xX 1.008 ~..007 .009 168-160 1.011 1.017 1.ao.011 I.010 I.008 1.wa 1.009 1.204 1.010 I.006 1.006 1 .52 0.007 1.009 1.176673 1.012 t.010 IS .015 1.163310 1.60767 I.010 I.005 22 eJ.016 1.216 i.011 16 4.011 .170093 1.010 1.016 16 wx 0.w9 228.m 1 .240 1.xX 1 .c‘% 1.010 1.w9 1009 i.o 1.01.192 1.012 1.007 1.012 I.291440 1.61 0.34 0.0.cca 1.w5 20 4.004 1.x6 1.12 0.012 192.2.65 0.007 I.280225 .009 I.006 264-276 1.009 1..012 1.226015 1.0. 0.