All material subject to strictly enforced copyright laws. © 2022 Euromoney Institutional Investor PLC group
Derivatives

The Use Of Economic Scenarios & Surveys In Risk Management

The economy and financial markets have a large impact on the performance of insurance companies. In the long-term, rising per capita income and property values have an impact on demand for insurance, increasing premium growth. In the short-term, however, the economy and markets primarily affect insurance companies through the valuation of assets and liabilities. On the liabilities side, lower inflation reduces the cost of future property/casualty insurance claims. For life insurers, low inflation generally lowers interest rates so insurers with interest rate guarantees in their life/savings policies may suffer financial stress if rates decline sufficiently. On the asset side, insurance companies hold mostly bonds and equities to pay future claims. These assets rise and fall along with interest rates, credit spreads, corporate default rates, equity markets and--if the company owns foreign assets--exchange rates. In managing these risks, insurers may shift asset allocations, alter their allocation of risk capital, change the terms and conditions on their policies and hedge their interest, market and exchange rate risks with derivatives. Stress Test

Prudent risk management dictates that an insurance company's assets and liabilities be stress tested. The degree and sophistication of stress testing appears to vary widely from company to company, depending on size and the needs of the insurance company. Some companies have a complex statistical methodology for stress testing their assets and liabilities. Others rely more on brainstorming and asking "what if" questions: What if interest rates fell sharply? What if equity markets collapsed? Whatever the methodology, two things are necessary for economic stress testing. First, a scenario with projections of economic variables, such as interest rates, inflation, equity markets, etc. Second, probabilities attached to the scenarios.

The scenarios must cover important economic risks, span the appropriate future period of concern, encompass all major geographic regions, and include relevant variables. For efficiency, the number of scenarios, time horizon, regions and variable set are limited. The scenarios might deal with, for example, short- and long-term recessions coupled with stock market crashes; long-term economic depression; long-term deflation; short- and long-term stagflation; severe exchange rate movements, etc. As indicated, the time horizon can vary. Most scenarios are quite ugly, involving large declines in equity markets and sharp increases in default rates and credit spreads. Without the major moves, it is not a stress test. Occasionally, companies may prefer to include a couple of scenarios which are not severe, as a contrast to more stressful ones. Of course, a baseline or benchmark forecast is also required, for the standard comparison. All regions where the company has a significant portion of assets and liabilities should be included.

Variables

The variables should include such factors as growth in real gross domestic product (GDP), inflation, interest rates, corporate bond default rates, growth in equity market valuations and exchange rates. Many of these variables are highly correlated. For example, a recession with a sharp decline in real GDP also entails a widening of credit spreads and an increase in bond defaults. Most, but not all, of the previous recessions have been preceded by rising inflation and interest rates and include a decline in equity markets. As the recession proceeds, interest rates and inflation fall, the yield curve steepens and growth resumes. The key point in designing the scenarios is to keep them economically consistent and coherent. They should be plausible and reflect past relationships, if still appropriate. Of course, some past relationships are not as valid today as previously. For example, the world is different from the 1930s, when the U.S. and much of the rest of the world experienced a depression. In particular, monetary authorities do not make as many policy mistakes and inflation is now well under control. Nevertheless, a depression--with output below its previous peak for a sustained period of time--is still a possibility, though remote.

Another decision to make is whether the scenarios are independent. If independent, then the probabilities of the individual scenarios are additive. This feature, an assumption, simplifies the use of the scenarios and how they can be interpreted and utilized in assessing their importance to the company's risk profile.

Each of the scenarios must be assigned a probability. For economic scenarios, this is not a trivial exercise. Choosing the period of time over which the model should be estimated is not straightforward, but depends on the dynamics of the economy. A small, dynamic model of the economy would need to be built--ideally based on simultaneously estimated equations--and distributions for each of the variables estimated. With the covariance matrix of the model, the probability of the scenarios could potentially be evaluated. This method faces a few problems. First, it would take a lot of time and effort. Also, the probability of a recession rises and falls given recent historical events--the model would provide the average probability of a recession over its estimation period. Finally, it would yield a dubious amount of information--a small specification error (something that is easy to do in building models of economies) could provide a misleading interaction term and inappropriate probabilities.

Expert Surveys

An alternative method for estimating probabilities is the expert survey technique. Under this methodology a carefully crafted questionnaire is distributed to knowledgeable experts who can provide a consensus view of the probabilities of the scenarios. Completing questionnaires is time-consuming, so the questions must be clear-cut. It is not easy to answer the question: "What is the probability of the following scenario?" and then display a table of real GDP growth, inflation, interest rates, etc. Also, the scenario may be confidential to your company. Instead, it is better to ask such questions as: "What is the probability that the U.S. economy averages negative growth for the next five years?" or "What is the probability that the Standard & Poor's 500 stock index will decline 25% or more in 2004?" Since, for example, interest rates are largely determined by growth and inflation, not all economic variables need to be surveyed. The scenarios can be fully characterized by growth, inflation, exchange rates and equity market movements. Finally, to induce repeat responses, a small corporate gift may be appropriate for those who complete the questionnaires on schedule.

Once the questionnaires are tabulated, the answers must be mapped over to the scenarios. All scenarios with the characteristics of deflation and five years of negative growth must be assigned probabilities consistent with the responses on the survey. A matrix that assigns probabilities to the scenarios and maps these over to the median probabilities of the survey questions will do the trick.

The scenarios and their probabilities can be used to support decisions on asset allocation of an insurance company's proprietary investment portfolio and to hedge market risks. In addition, for insurance companies there may be a need to change the terms and conditions of the client's policy. This would be particularly true for life insurance companies with equity and interest rate guarantees imbedded in their savings products. If, for example, the risk rises of a sharp decline in interest rates, the company may prefer to go long in duration for their proprietary assets, lower the level of interest rate guarantees, increase their exposure to credit risk and hedge their interest rate exposure. Exchange rate, interest rate, and equity market hedges may also be appropriate, given the responses to the survey and the implications for the company's capital.

This week's Learning Curve was written by Kurt Karl, head of economic research and consulting at Swiss Rein New York.

We use cookies to provide a personalized site experience.
By continuing to use & browse the site you agree to our Privacy Policy.
I agree