Dickie Whitaker and Dr Gordon Woo explain the importance to reinsurers of risk based capital and dynamic financial analysis (DFA) and how the understanding of catastrophe modelling plays an important role.
Over the last few years reinsurers have been looking at ways to turn a corporate goal of improved risk assessment and diversification into a more quantitative system that integrates this analysis with their corporate assets and quantifies risk based capital requirements. Although techniques of integrated or holistic risk management have been in operation for some time, it is only now that reinsurance companies have been truly able to utilise these on a risk by risk basis. The Bermudian market in particular, with its strong adherence to technical risk assessment, has been at the forefront of these techniques.
Integrated risk management
Operations which once were logically or logistically disjointed can now be integrated into a system which allows for more efficient and effective management. The cost of implementing ever larger integrated systems is defrayed against the greater profitability of business operations and the streamlining of corporate decision-making through the geometrical progression in the price-performance of computer hardware and communication links.
Within the insurance industry, business decision-making has traditionally been devolved to a departmental level: the underwriting department focusing on liabilities, the investment department on assets and the actuarial department on reserving. Furthermore, where it has been sought, specialist expertise in catastrophe modelling has typically been brought in as a specific aid for underwriting and, to a lesser extent, for capital allocation.
There is an increasing appreciation of the value of a more holistic approach which considers the basic coupling between liability and asset analysis. An integrated risk management approach provides a framework for investigating risk/reward equations between a far wider range of business strategies. An example of one of these important trade-offs is that between levels of reinsurance purchase and equity investment. By increasing equity investment while purchasing reinsurance, an insurance company can improve its solvency status, without much detriment to its economic value.
Dynamic financial analysis
The numerical study of risk/reward equations is one of the analytical tasks undertaken within a broad computational framework generally described as dynamic financial analysis (DFA). The description of the financial analysis as dynamic, and hence time-dependent, is the crucial aspect. Risk managers seek to exercise an active degree of control over future business volatility. So far, however, the full integration of risk management with catastrophe management remains to be realised. This is reflected in the comparatively minor level of interaction between catastrophe modellers and actuaries, which is essentially limited to the explanation of the file of hazard scenarios and loss consequences. Within the overall mission of integrated risk and catastrophe management, this interaction would be significantly extended.
Catastrophe management is the business of companies, such as our firm EQECAT, which offer software and consulting services to organisations, primarily within the insurance, reinsurance and banking industries. These services now extend to global coverage of the principal natural hazards of earthquake and windstorm, while providing more limited geographical coverage of flood and volcanic eruption. Our primary aim is to use state-of-the-art scientific and engineering expertise to construct probabilistic computerised models of catastrophic loss. Each model is constructed from an ensemble of a large number of hazard event scenarios, each of which is associated with an occurrence frequency and a regional map of the potential hazard intensity. These scenarios constitute a direct interface for liaison with actuaries, since each represents one specific realisation of a natural hazard which could impact severely on an insurance or reinsurance business.
Risk and catastrophe management
Central to the determination of risk-based capital is a knowledge of the claims distribution. The tail of the claims distribution is particularly important, since any excess of actual claims over premium income involves an exposure of risk capital. An investment which requires more capital because of its higher riskiness should earn a higher return. From a risk-based capital perspective, the riskier the claims potential, the higher the premium that should be charged.
The level of premium which is judged to be commensurate with the risk exposure depends on the particular corporate strategy adopted for the risk-based allocation of capital. This strategy may depend partly on mathematical considerations of ruin probability and loss volatility, and partly on generic ratios such as premium to surplus. (The connection between risk and required capital stems from the limit that management would impose on its tolerance to the probability of ruin.)
The loading factor on risk premiums is crucially dependent on the tail of the claims distribution. The tail uncertainty is complex, arising as it does both from intrinsic stochastic variability in claims data, and also the scientific and engineering uncertainty in model parameterisation. The quality of spatial and structural portfolio information also significantly affects the reliability of tail estimation. Not all of this tail uncertainty is transparent in the documentation of catastrophe models, which is why catastrophe modelling organisations need to provide clarification to assist in the process.
The above illustration of the added value of catastrophe modelling knowledge to actuarial analysis extends to DFA in general. The price of risk may not be simply gauged by variance loading formulae, but is eventually established in a competitive market situation where the forces of supply and demand come into play. If the econometrical tool of game theory is introduced to address the influence of competition, then the loss covariance between portfolio accounts becomes especially significant. The evaluation of this loss covariance is subject to uncertainties which are only fully unravelled through familiarity with the technical details of catastrophe modelling.
Asset and liability correlation
Another example of the need for the full integration of risk and catastrophe management is the correlation between assets and catastrophe liabilities under certain circumstances of economic instability, when a major fall in asset value might be triggered by a great earthquake disaster.
Apart from providing a means for quantifying loss distributions and associated volatility, catastrophe models are adept at calculating the loss covariance between different portfolios, which is a numerical measure of their diversification. These calculations are important both for risk pricing, as well as for risk-based capital allocation. Perhaps less familiar is the capability of catastrophe models to calculate the covariance between losses in different business classes: for example, where because of tsunami action, an earthquake could cause a severe loss to a marine account as well as to a general property account. In the allocation of capital, allowance needs to be made for the contingency of a catastrophic event which could cut across multiple classes. Through covariance calculations of this kind, the integration of catastrophe modelling expertise can add significant value to corporate risk management.
Following the reduction of reinsurance capacity in the early 1990s, significant capital was drawn into the business which fortuitously coincided with a relatively quiescent period of catastrophic losses resulting in good returns for reinsurers. Strong equity markets and good earnings have fuelled capital growth, further adding to capacity in a soft market. To ensure long term stability and profit potential, reinsurers are looking for the strategy that will ensure their company's dominance both when significant losses do occur and in soft market conditions.
With increasing standardisation in the information available to reinsurers and its analysis by catastrophe models, the differential between reinsurers will increasingly be defined by the corporate strategy and its application through DFA analysis. Those companies that succeed in this first should be able both to survive better the next major catastrophe(s), stabilise results and ensure that core business does not transfer to the capital markets.
Bill Riker, president and chief operating officer.
At Renaissance Re, the bulk of our decision support tools are based in a dynamic financial analysis approach. One of the key obstacles we were able to solve was to take a highly computational approach and provide information to our decision-makers within the short time frames required to provide exceptional service in the market.
We believe that using DFA provides significant advantages in evaluating the risk-return characteristics of insurance and/or reinsurance products. The advantage is this: a DFA approach provides information about the real questions managers need answered to properly construct a portfolio of risk: "What is the likely probability of losing or making x amount of money over a time period." Most traditional insurance measures are snapshots or "point in time" views of risk. How the risk interacts with the premiums charged over a time period is much more valuable information.
In catastrophe business, DFA enables a company to combine partially correlated risk to see how this may affect the risk/return characteristics of the portfolio. Having the ability to evaluate these more subtle issues can provide a significant advantage over more simplistic methods. We believe the continued emergence of more and more sophisticated models based on DFA principles will transform the way many organisations evaluate and measure risk.
John Weale, vice president and chief financial officer.
There is no question that as computer power becomes cheaper, more widely available and easier to use, then dynamic financial analysis will have more and more applicability throughout the insurance and reinsurance industry, as well as outside it. The speed with which complex modelling and simulations can now be run has been reduced from weeks to hours, even for companies with multiple lines of business and sophisticated investment portfolios.
However, there are limitations which have to be considered. The broader the range of potential outcomes for each variable, the greater the potential deviation. The greater the number of variables, the likelihood of deviation multiplies. Because of this, we believe that DFA is a useful tool for businesses where the likely range of underwriting results is relatively narrow, for instance general property/casualty insurers and reinsurers.
However, for companies where the majority of business is property catastrophe reinsurance, the range of likely outcomes is from zero to the total aggregate of sums reinsured, which for most companies is hundreds of millions
of dollars, pounds, deutschemarks, etc. Not only is the amount highly unpredictable, but so is the timing and location (and hence currency). Then you have to factor in the possible effects of a major catastrophe on the local money/stock markets and any other markets with which there is a relationship (as was seen in Asia in the latter part of 1997).
The biggest part of the problem is that models are built on historical data and assumptions. For most parts of the world, data on catastrophes has only been gathered for comparatively short periods of time - in most cases less than 50 years; in some for 20 years of less. Even when the data exists, the assumptions used in the models may negate it. For example, the fault-line at Northridge, California, was a known fault, and yet several earthquake models did not include it (until 1994) because there had been no seismic activity from it in over 50 years. As a result, several companies who relied upon those models for exposure management lost a lot of money.
DFA can provide efficiency within modern portfolio theory for many companies, because many simulations can be run with small or large numbers of changes to the variables affecting the outcome, and the optimum asset allocation may still not vary significantly. However, for property catastrophe companies writing business on a global scale, the number of potential outcomes, many significantly different from the others, is so great that the resulting efficient asset mixes are too diverse to be of practical guidance.
Andrew Cook, senior vice president and chief financial officer.
At LaSalle the question of dynamic financial analysis is not how important it will be but how much we can extend its utilisation. We actively employ dynamic financial analysis in our annual and long term financial planning.
Our current model takes into account all of the major inputs that can affect our results and capital base including: premiums written and earned by line of business, investment income, losses by line of business, expenses and cash flows. A key factor in the model is the involvement of all parties at LaSalle including underwriting, actuarial and finance. Given the involvement of the whole company in the process, we are much more comfortable with the results of the analysis and how they impact on our planning processes.
The model produces simulated results by major line of business and in total. This feature allows the company to simulate the effect of increasing or decreasing its participation in any particular line or lines of business while keeping the other lines static. This allows us to focus on the incremental return of each line or combination of lines of business to the company.
The model is also a key part of determining the attachment points and limits for our multi-year reinsurance programme. Further, by incorporating the parameters of our CatEPut programme we can simulate the triggering of the reinsurance and the CatEPut, either individually or in concert with each other.
Capital management is an extremely important area for the company, our shareholders and clients. Having a variety of capital management techniques, such as share repurchases and dividends built into the model, it allows us to see how each method works on its own or in concert. This is particularly crucial when the impact of losses is overlaid on to the capital base.
Overseas Partners Ltd
Bruce Barone, president and ceo.
Peter Komposch Jr, corporate planning.
Insurance and reinsurance executives make strategic decisions on a daily basis and need analytical tools that can help them evaluate competitive alternatives. Our analysis attempts to understand the financial implications of each alternative and assess the probability of its occurrence.
When conducting complex and multi-dimensional transactions, the ability to understand expected results, in a dynamic environment, becomes increasingly difficult. While selecting the best alternative is our continuous goal, we must demonstrate why we make the choices we do. Intuition based decisions or selections based on a gut feeling are no longer acceptable. Our management, our board of directors and even our shareholders want to know why one alternative is chosen over another.
Traditional financial forecasting has been somewhat limited. It produces fixed outcomes (best/expected/worst case scenarios) while revealing little on how expected results may change in an environment where multiple, sometimes irrational variables are at play. Dynamic financial analysis allows us to understand the changes in probability of expectations occurring given several simultaneous events.
DFA forces us to further specify our assumptions and emphasises the implications of variability or uncertainty, improving our potential for better decision making. It has helped us make better projections, conduct more extensive analysis and, hopefully, better understand the implications of our decisions.
Dickie Whitaker is marketing manager - UK and Dr Gordon Woo is principal analyst for EQECAT Inc, London. Tel: +44 (0) 171 357 2312.