Skip to main content
main-content

Über dieses Buch

The research project leading to this book was initiated in the fall of 1979 when the American Council of Life Insurance (ACLI) contacted Dan McGill, chairman of the Wharton School Insurance Department, about conducting a study on risk classification in life insurance. The ACLI was concerned about legislative and judicial activity in this area and its potential effects on the life insurance industry. A meeting was held at the ACLI offices in Washington, D.C., between several members of the ACLI staff and Dan McGill and David Cummins representing the Wharton School insurance department. An agreement was reached that a study would be conducted at Wharton dealing with issues in risk classification. Although the staff of the ACLI suggested directions the study might take, it was agreed that the design and execution of the study would be solely under the control of the researchers. The researchers also retained unrestricted publication rights in the results of the study. This agreement has been honored by the ACLI during the course of the project.

Inhaltsverzeichnis

Frontmatter

Introduction

1. Introduction

Abstract
Risk classification is the process of separating into groups (classifying) potential insureds (risks). The mechanism of classification is to group the risks by various pieces of information, such as age or occupation. The purposes of classification include, but are not limited to, the determination of acceptability for insurance and the type, amount, and price of that insurance. The justification for classification is that risks are assumed to be placed in relatively homogeneous groups—that is, groups in which the risks have similar probabilities of loss. If classification is accurate, insureds are treated equitably (since similar insureds pay similar premiums and the premiums are related to expected loss) and insurance companies can accurately estimate expected losses.
J. David Cummins, Barry D. Smith, R. Neil Vance, Jack L. VanDerhei

Fundamentals of Risk Classification

Frontmatter

2. The Theory of Insurance Pricing: Loss Distributions and Expected Value

Abstract
Insurance contracts are contracts that promise contingent payments, that is, payments depending on the occurrence of events that may or may not happen. Of course, all contractual obligations are contingent in some sense—for example, a blizzard may close the banks and depositors will not be able to withdraw their demand deposits. The contingent events specified in insurance contracts, however, are neither certain nor unthinkable, yet, at least in the aggregate, have a predictable degree of occurrence.
J. David Cummins, Barry D. Smith, R. Neil Vance, Jack L. VanDerhei

3. The Economic Role of Risk Classification

Abstract
This chapter discusses the economic rationale for risk classification and the effects on the insurance market if insurance companies are unable to classify applicants for coverage.1 The principal result is that the market may become unstable if insurers are unable to classify; that is, there may be no equilibrium set of contracts offerable that will not eventually lose money and have to be withdrawn. This disruption of the market is likely to restrict availability of coverage. Even if an equilibrium set of contracts exists in the absence of classification, it will place some policyholders in a worse position than would be the case in a competitive market with classification. Additional government regulation may be necessary to maintain or improve the operation of the market in the absence of classification.
J. David Cummins, Barry D. Smith, R. Neil Vance, Jack L. VanDerhei

4. Heterogeneity in Risk Classification

Abstract
The traditional method of pricing insurance divides risks into various classes for the purposes of collecting statistics, estimating loss distribution, designing products, underwriting (acceptance or rejection), and pricing. Risks are assigned to classes according to various characteristics, for example, age or medical condition. Each member of a class is charged a premium, which is the expected value of the loss distribution assumed to apply to each member of the class. A critical assumption of the traditional method is that classes are essentially homogeneous; that is, all risks in the class have the same loss distribution. Only if all risks in a class are similar can it be argued that the class expected value is the appropriate premium for each individual in the class.
J. David Cummins, Barry D. Smith, R. Neil Vance, Jack L. VanDerhei

5. Fairness in Risk Classification

Abstract
Earlier chapters have discussed the economic rationale for risk classification, focusing on how this process contributes to the solvency of the insurance company and preserves the availability of insurance by reducing anti-selection. This chapter considers the charge that some aspects of the risk classification process are unfair. Since fairness is often in the eye of the beholder, it is necessary to begin by considering some of the different ways that fairness has been defined. These definitions then are analyzed from the point of view of meaningfulness, desirability, and attainability within the present organization of the insurance industry.
J. David Cummins, Barry D. Smith, R. Neil Vance, Jack L. VanDerhei

6. Conclusions to Part I

Abstract
An efficient and accurate method of risk assessment is necessary to maintain a solvent and competitive private life insurance industry. Economic forces in the unregulated private market usually require that each risk be charged the expected value of its losses; risk assessment refers to the variety of techniques used to estimate this expected value.
J. David Cummins, Barry D. Smith, R. Neil Vance, Jack L. VanDerhei

Current Risk Classification Procedures

Frontmatter

7. Introduction: Overview of Risk Classification

Abstract
Part II describes the underwriting practices of the U.S. life insurance industry. It indicates which factors are examined, to what extent they are scrutinized, and how a rate is eventually developed. This information comes directly from the underwriting manuals of major life insurance and reinsurance companies operating in the United States.
J. David Cummins, Barry D. Smith, R. Neil Vance, Jack L. VanDerhei

8. The Nonmedical Application

Abstract
Most of the material in part II is concerned with medical underwriting, that is, underwriting in which the insurance company pays for a medical examination that becomes part of the information considered in evaluating the risk. Due to rising medical costs, however, a growing proportion of insurance policies are issued without a medical examination. These nonmedical policies rely heavily on the application filled out for the policyholder by the agent.
J. David Cummins, Barry D. Smith, R. Neil Vance, Jack L. VanDerhei

9. Underwriting Medical Impairments

Abstract
This chapter analyzes the underwriting procedures for many of the major medical impairments. Because it was not feasible to include every impairment, a list was created by selecting factors from each of the following groups:
1.
Controversial items specifically included in the list of impairments solicited by the Society of Actuaries for its newest medical impairment study, such as multiple sclerosis, hemophilia, and epilepsy;1
 
2.
Impairments for which statutory restrictions have been developed, such as blindness, deafness, and mental retardation;
 
3.
The leading causes of death among individual life insurance policyholders, such as high blood pressure, diabetes, and cancer.2
 
J. David Cummins, Barry D. Smith, R. Neil Vance, Jack L. VanDerhei

10. Nonphysical Underwriting Factors

Abstract
This chapter explores nonphysical underwriting factors. The first half examines those factors considered controversial, including financial status, use of alcohol and drugs, driving record, and character of the applicant. The second half analyzes occupation, aviation and other avocations, and foreign residence and travel. Twenty-one insurers provided sufficient information on nonphysical factors to be included in the analysis. Three of these, however, rely so heavily on the reinsurers that they could not be analyzed separately. Therefore, 18 underwriting manuals are examined.
J. David Cummins, Barry D. Smith, R. Neil Vance, Jack L. VanDerhei

11. Conclusions to Part II

Abstract
The most prominent characteristics of life insurance underwriting are its flexibility and heterogeneity. Flexibility is needed because average mortality rates never apply to all applicants, even in classes characterized by great homogeneity and ample data. Flexibility also permits underwriters to vary prices to meet competition, countering the apparent rigidity of insurance rate books. Heterogeneity is present primarily because accurate mortality statistics have not been available for most hazards. Thus, in many instances, rates and procedures have evolved through judgments made by underwriters, medical personnel, and actuaries with many years of life insurance experience. These features of underwriting are not necessarily adverse, but the same factors that permit the underwriter to achieve greater equity also provide latitude for the exercise of prejudice and personal whim. We believe that the benefits of a free and competitive life insurance industry outweigh the possible inequities that may arise from the exercise of underwriting judgment. Nevertheless, regulators and public interest groups should be alert to possible problems and injustices.
J. David Cummins, Barry D. Smith, R. Neil Vance, Jack L. VanDerhei

Multivariate Analysis of Underwriting Risk Factors and Mortality

Frontmatter

12. Introduction

Abstract
Life insurance companies in this country have been under increasing pressure recently to justify risk classifications for applicants with certain physical and medical impairments. Although constraints on the use of race, creed, national origin, and religion in life insurance risk classification systems have existed since the 1960s, the traditional method for indicating the mortality associated with specific risk classification factors has remained essentially unchanged for almost 80 years. Several considerations motivate a reevaluation of the traditional method:
1.
Recent surveys conducted by the American Council of Life Insurance have indicated that a significant proportion of Americans disagree with the basic philosophy of the risk classification system currently used in the life insurance industry.1 The results indicated that 42 percent of the public consider unfair the practice of charging higher rates for applicants with heart disease, and 70 percent believe that higher premiums for handicapped applicants are unfair.2
 
2.
The bill H.R. 100, the “Nondiscrimination in Insurance Act of 1979,” would prohibit an insurer from charging a different rate to an applicant because of race, color, religion, sex, or national origin.3
 
3.
Recent regulatory activity has included the approval of “A Model Regulation on Unfair Discrimination in Life and Health Insurance on the Basis of Physical or Mental Impairment” by the National Association of Insurance Commissioners.4 This model bill implements Section 4(7) of the Model Unfair Trade Practices Act, which prohibits “any unfair discrimination between individuals of the same class and equal expectation of life.”5 The rationale for adopting the impairment bill was to prohibit arbitrary classification by the insurer. This bill extends the requirements of an earlier model bill to all impairments.6
 
4.
The U.S. Civil Rights Commission has also generated some interest in this field. During a three-day “Consultation on Discrimination Against Minorities and Women in Pensions and Health, Life and Disability Insurance,” at least two-thirds of the papers dealt with risk classification topics.7
 
5.
In addition to the regulatory and legislative underwriting restrictions that have already appeared,8 several proposals would prohibit rate differentials based on gender or marital status.9
 
6.
Several states have issued regulations that prevent the adoption of new underwriting classifications unless sufficient mortality experience exists to support the level of extra premium charges.10
 
7.
Social pressures that could ultimately transform the current risk classification system have been generated by privacy advocates. Life insurers may soon be forced to search for alternative underwriting factors if the value of information from sources such as investigative reports and the Medical Information Bureau is vitiated.11
 
8.
Recent court decisions on the “unisex” issue have implications concerning the use of sex as a risk factor in employee benefit plans.12
 
J. David Cummins, Barry D. Smith, R. Neil Vance, Jack L. VanDerhei

13. A Multiple logistic Methodology for the Estimation of Risk Classification Models

Abstract
In this chapter a methodology is developed to measure the quantitative relationship between the probability that a policy will terminate by death in the ensuing year and the policyholder’s characteristics at the time of application for life insurance. Due to the dichotomous nature of the response variable (death or survival for the following year). multiple regression analysis is inappropriate.1 Consequently, this study utilizes a maximum likelihood algorithm to estimate the coefficients in a multiple logistic model. The Newton-Raphson technique provides for an iterative solution of the estimated betas after initial estimates are calculated by the application of Fisher’s linear discriminant function.
J. David Cummins, Barry D. Smith, R. Neil Vance, Jack L. VanDerhei

14. The Effect of Physical and Mental Impairments on the Annual Probability of a Policy Terminating by Death

Abstract
With the methodology developed in the preceding chapter serving as a basis, the results of this study can now be presented. The analysis is conducted on two categories of impaired insured lives. The first category consists of insureds with a record of any one of 13 physical impairments currently under regulatory scrutiny. These items were specifically included in the list of impairments prepared by the Liaison Committee of the Society of Actuaries and the Association of Life Insurance Medical Directors of America for use in its most recent impairment study.1 The second category of insureds consists of those with medical impairments on which a reasonably large amount of data was available.
J. David Cummins, Barry D. Smith, R. Neil Vance, Jack L. VanDerhei

15. Comparison of Logistic Model Results with Other Data

Abstract
This chapter provides a detailed analysis of the results obtained for each group of impaired lives. The findings of chapter 14 are compared with the underwriting manual ratings of the two life insurance companies supplying the data and with the results of other risk factor studies on insured populations. Underwriting manuals are used in the life insurance industry to supply the information necessary to rate most of the common impairments. However, no definite rating is given for certain impairments that, due to their complexity, are still treated as the exclusive domain of the company’s medical department.
J. David Cummins, Barry D. Smith, R. Neil Vance, Jack L. VanDerhei

16. Summary and Conclusions for Part III

Abstract
This section of the study was undertaken to determine if the experience generated by certain specific life insurance impairment classifications justified the risk classification models currently in use. Current procedures used to determine the numerical value of debits and credits for applicants have not progressed since the early part of this century. With the increased availability of electronic computational devices, it is now possible to implement a methodology that will determine these numerical values while simultaneously controlling for differences in risk factors between the impaired group and the standard-issue group. This method is an improvement because it will account for the influence of any significant risk factor; the conventional method only controls for the effects of age and duration since the time of issue.
J. David Cummins, Barry D. Smith, R. Neil Vance, Jack L. VanDerhei

Backmatter

Weitere Informationen