Loss Models, 5th Edition

19Simulation

19.1 Basics of Simulation

Simulation has had an on-again, off-again history in actuarial practice. For example, in the 1970s, aggregate loss calculations were commonly done by simulation, because the analytic methods available at the time were not adequate. However, the typical simulation often took a full day on the company's mainframe computer, a serious drag on resources. In the 1980s, analytic methods such as the recursive formula discussed in Chapter 9 and others were developed and found to be significantly faster and more accurate. Today, desktop computers have sufficient power to run complex simulations that allow for the analysis of models not amenable to analytic approaches.

In a similar vein, as investment vehicles become more complex, insurance contracts may include interest-sensitive components and financial guarantees related to stock market performance. These products must be analyzed on a stochastic basis. To accommodate these and other complexities, simulation has become the technique of choice.

In this chapter, we provide some illustrations of how simulation can be used to address some complex modeling problems in insurance, or as an alternative to other methods. It is not our intention to cover the subject in great detail but, rather, to give you an idea of how simulation can help. Study of simulation texts such as Ripley [105] and Ross [108] provides many important additional insights. In addition, simulation can also be an aid in evaluating ...