Richard Shaw examines stress and scenario testing for non-life insurers.
Stress testing and scenario analysis is part of best practice in the overall management of a non-
life insurance company. Such analyses, being
based on a study of the impact of unlikely but not impossible events, allow companies to understand the risks they face under extreme conditions.
Further impetus to perform this type of analysis has come from the Financial Services Authority (FSA) in Consultation Paper (CP) 190, 'Enhanced capital requirements and individual capital assessments for non-life insurers'. Companies will be required at all times to maintain overall financial resources, including capital and liquidity resources, which are adequate to ensure there is no significant risk of liabilities not being met as they fall due.
Companies need to identify the major sources of risk under the following categories:
- Insurance risk
- Market risk
- Credit risk
- Operational risk
- Liquidity risk
For each of these major sources of risk, the company must carry out stress and scenario tests appropriate to the nature of the risk.
Stress testing and scenario analysis
Scenario analysis is the process of evaluating the impact of specified scenarios on the company's financial position. The emphasis is on specifying the scenarios and following through their implications. Scenario analysis typically refers to a study where a wide range of parameters are varied simultaneously. The scenarios could be chosen as events intended to have a defined probability of occurrence, for example, a 'one in a hundred years' event.
Stress testing is the process where we evaluate a number of statistically defined possibilities to determine the most damaging combination of events, and the loss they would produce. The likelihood of such an event is then assessed.
There are basically two types of events: historical and hypothetical. Historical events have actually occurred, which reduces the arbitrariness of their inclusion, and provides clarity as they are more readily understood. Hypothetical events, on the other hand, are more thorough and systematic, but anticipate risk with no historical parallel.
Insurance risk concerns the inherent uncertainties as to the occurrence, amount, and timing of insurance liabilities. Insurance risk consists of two aspects: the risks associated with the writing of new business (underwriting risk) and the risks inherent from business already written (reserving risk). Underwriting risk is the risk associated with the uncertainty of business written in the future, both new business and the renewals of existing policies. This would include catastrophe risks. Reserving risk is the risk associated with the potential inadequacy of claims reserves and provisions for unearned premiums and unexpired risks.
Examples of factors to consider for underwriting risk would be:
- the uncertainty of future claims experience;
- incorrect pricing owing to poor data or an inappropriate method;
- the effects of rapid growth in business volumes due to underpricing, or conversely a decline in premium volume due to overpricing;
- a lack of underwriting controls, such as inappropriate underwriting strategy or a failure to apply underwriting guidelines and policy wordings;
- a potentially catastrophic aggregation of claims;
- the geographical mix of business, including concentrations of risk and lack of diversification;
- inappropriate reinsurance programmes, lack of availability of suitable reinsurance, or large reinsurance price rises.
For reserving risk, examples of risk factors are:
- the adequacy of claims reserves;
- the adequacy of provisions for unearned premiums and unexpired risks;
- the frequency and severity of large claims and latent claims;
- changes in the legal system, such as increased court awards or changes in policy wording interpretation;
- the effects of inflation on claims reserves and expenses;
- social changes resulting in an increase in the propensity to claim or to pursue litigation.
Credit risk covers the risk of loss if another party fails to perform its obligations, or fails to perform them in a timely manner. Allowance should be made for the financial effects of non-payment of reinsurance and of the non-payment of premium debtors such as intermediaries. (See box for factors to consider.)
Market risk is the risk that market movements in (for example) interest rates, foreign exchange rates, or asset prices lead to an adverse movement in asset values which is not matched by a corresponding movement in the value of liabilities. (See box for factors to consider.)
Liquidity risk is the risk that a firm has insufficient financial resources to meet its obligations as they fall due, or can only secure the resources at excessive cost. (See box for factors to consider.)
Operational risk is the risk of direct or indirect loss resulting from inadequate or failed internal processes, people and systems, or from external events. Some operational risks cannot be quantified accurately and so a more qualitative approach to assessing operational risks is appropriate. A scoring mechanism of high, medium, and low for the purposes of assessing the capital required for each risk might be used. (See box for factors to consider.)
It is important to consider consequential effects and not just to analyse events in isolation. Examples of some of these knock-on effects are:
- Ripple effects The direct and indirect effects that are assumed to follow the event, often with some time delay, eg a large catastrophe loss could lead to an increased probability of reinsurer insolvency and hence higher credit risk.
- Management action Events do not happen in isolation and here management action is assumed to occur in response to an adverse scenario, eg bad loss experience leads management to increase premium rates in subsequent time periods.
- Regulatory action Some adverse scenarios will cause some form of regulatory response, eg a failure to meet the minimum regulatory capital requirement.
- Rating agency action Some adverse scenarios may lead to rating agency action with consequential follow on effects, eg a significantly bad underwriting loss results in a large capital reduction, the extent of which leads to a rating downgrade with negative implications for future premium volume.
There are various modelling considerations:
- Frequency The analysis must be conducted regularly to reflect the changing characteristics of a portfolio. At a minimum, this should be annually in conjunction with the business-planning exercise.
- Forecast period The time horizon needs to be long enough for the effects of the stress and scenario tests to be captured, which for a non-life insurance company is a minimum of two years. In most cases the time horizon will be up to five years.
- Dependency Any analysis should take account of the complexity of a company's business operations and the correlation between risk factors.
- Modelling techniques A deterministic approach, using the probability distributions to determine the extremes of events, enables greater control over the ripple effects and management actions that may follow events. However, a stochastic approach based on economic capital models is more advanced, allows for dependencies, and will provide an overall probability distribution of the company result. A combination of deterministic and stochastic modelling is needed, keeping in mind the limitations of each approach.
- Model output This should be sufficient to produce the income statement, balance sheet, and solvency calculation for each year-end in the period under consideration.
Designing stress tests
Designing the stress tests requires a mix of skills to provide an understanding of the business, identify the risks, and perform the modelling and analysis. Expertise will be needed from actuarial, underwriting, finance, risk management, and other functions within a company. Companies may need to consider the external views of supervisors, consultants, and rating agencies. The nature and extent of the tests will, among other things, need to take into account the company's solvency position, market position, lines of business, investment policy, business plan, and general economic conditions.
Implementation of scenario analysis and stress testing should form part of a company's overall risk management framework. The overall approach should be kept simple, so as not to lose sight of the overall objective. Some of the key activities will be:
- deciding on the risk measure;
- development of a risk language and common definitions;
- detailed investigation of the company's operations;
- data capture and analysis;
- setting of assumptions;
- scenario and stress modelling;
- investigation of how risks can be mitigated;
- recording and reporting of results;
- discussions with internal and external parties, such as the regulators.