1. Solvency Testing Model SD-CNSF México October, 2002

2. Contents 1.Background 2.Solvency Testing Model 3.Dynamic Solvency System 4.Perspectives

3. Background

4. In Mexico, the insurance operations are carry out under a dynamic behavior of financial and risk variables. This situation occurs mainly because a business line high rotation, competition and inflation effects. At the same time, traditional insurance companies are not specialized and they can manage both, life insurance and non-life insurance, increasing the administration complexity.

5. Background For this reason it is important to have efficient risk analysis tools, for identifying business line risk factors and analyzing capital sufficiency in the medium and short term.

6. Background The National Insurance and Surety Commission of Mexico (CNSF) has been developing a Dynamic Solvency Testing Model to encourage self-regulatory practices by insurance enterprises, and at the same time strengthen preventive supervision. Based upon the mathematical analysis solvency model, a dynamic solvency testing computing system (SD-CNSF) was developed, which allows a prospective analysis of both solvency and risk exposure factors.

7. Solvency Testing Model

8. The dynamic solvency model of the CNSF incorporates aspects of the Mexican regulation, as well as the laws of the behavior of the risk variables of each insurance line of business in Mexico. Probability density functions have been fitted using the last five years statistical information from the Mexican market.

9. The statistical data correspond to each company’s claim amount for a specific business line i, MR i (t). The claims amount is expressed as a percentage of the written premium PE i (t), loss ratio. The loss ratio t for business line i at year t is expressed as: Solvency Testing Model

10. For each insurance business line, it was proved through statistical analysis, that the loss ratio random variable, has the typical characteristics of a gamma probability function, whose mathematical expression is: Solvency Testing Model

11. Consequently, gamma density functions were fitted, for each business line, using Mexican insurance companies statistical data. Solvency Testing Model Parámetros  9.0000  0.0450

12. A Kolmogorov-Smirnov goodness-of-fit test was performed to prove the probability distribution functions’ adequacy on each business line. The Kolmogorov-Smirnov goodness-of-fit test is based on the absolute value of the maximum difference between the sample cumulative distribution values and the hypothetical cumulative distribution: Solvency Testing Model

13. The probability density functions for each business line are as shown: Solvency Testing Model

14. The company’s capital position at time t (CAP t ), can be expressed as the company’s capital position at time t-1 (CAP t-1 ) plus the capital contributions at time t (AC t ) plus the operation flow (profit or loss) at time t (R t ) : Solvency Testing Model

15. The insurance company’s solvency margin at time t, is calculated as a portion of company’s assets at time t (  ), minus the solvency requirement at time t : Solvency Testing Model

16. The company’s solvency requirement ( RS(t)) is obtained by adding the solvency requirements of every business line : Solvency Testing Model

17. The operational flow of the insurance company at time t, is calculated as the difference between inflow (premium, investment earnings) and outflow (expenses, premium reserves, ceded premium, claims) at time : Solvency Testing Model

18. Written premium (PE), is simulated based on the historic premium growth rate of the company in the last five years, for each business line, making the growth rate fluctuate, within a small interval of values around the historic   by using a uniform distribution function: Solvency Testing Model

19. Retention Premium (PR) is calculated by multiplying a historic rate of retention premium, by the written premium, making the rate of retention premium fluctuate, within a small interval of values around the historic value by using a uniform distribution function. Solvency Testing Model

20. Acquisition Cost (CA) is calculated as a percentaje of the retention premium (historical percentage), making the percentage fluctuate, within a small interval of values around the historical value by using a uniform distribution function : Solvency Testing Model Costos de Adquisisión

21. Administrative Cost (CO) is calculated with a formula that involves a part as fixed cost, and another part as variable cost that depends on the premium: Solvency Testing Model

22. Investement Earnings (PF), is calculated as the amount of assets multyplied by their asset yield rates, making each rate fluctuate, within a small interval of random values in accordance with the expected trend : Solvency Testing Model

23. Premium reserve is calculated, in the case of short term insurance, as a percentage of the retention premium. The portion of unearned premium reserve is calculated by a formula which involves the average retention premium of last two years: Solvency Testing Model

24. The simulation process, is based on the so-called “inversion method”, which consists in generating random numbers with a continuous uniform distribution on (0,1), and then applying the inverse of the cumulative distribution function of the loss ratio random variable: Solvency Testing Model

25. Finally, with the model it is possible to generate several scenarios and calculate a ruin probability as well as the expected value of future capital necessities. Solvency Testing Model

26. Dynamic Solvency System

27. Based on the dynamic solvency model, the CNSF has developed a dynamic solvency computing system (SD-CNSF), which carries out simulations of stochastic processes based on each of the insurance business lines’ probability functions, as well as on the company’s business plan scenarios. The input information of the system is a data base, which contains all financial information of the company in the last five year.

28. Financial Statementes Business Plan Macroecono mical Expectations Invested Assets Estimated Risk Factors Projected Financial Statements Future Capital Necessities Sensitivity Analysis Results Processes Stochastic Proceeses Simulations Scenarios Index Calculatios Grafhs SD-CNSF The system functioning is based on the information, processes and the next results: Risk Probability Functions Dynamic Solvency System

29. Next, we are going to show the Dynamic Solvency Computing System (SD-CNSF). SD-CNSF Dynamic Solvency System

30. Perspectives

31. The development of the Dynamic Solvency Computing System (SD-CNSF), is in an initial phase. We hope to improve and to incorporate new routines in order to increase its efficiency. The Dynamic Solvency Computing System (SD-CNSF), will allow to implement preventive regulation. The Dynamic Solvency Computing System (SD-CNSF), is a quite flexible tool that could be updated to the normative changes.

32. October, 2002 Solvency Testing Model SD-CNSF