Universal Truths.

Insurers with lower lapse ratios, lower expenses and higher investment yields generally provide universal life policies with higher surrender values.

Life insurance studies indicate wide price dispersion among insurers for any particular type (e.g., whole life or universal life). To explain such variation, some analysts have suggested that cash-value life insurance should be viewed as a package of options. That is, the higher price of a particular policy might be due to that policy's superior guarantees, enhanced flexibility and higher-rated insurer, as compared with another policy. This explanation is attractive, given its intuitive basis and apparent support for the existence of a highly competitive life insurance market.

However, data from A.M. Best Co.'s Best's Policy Reports, 1995, indicate that 10-year surrender values for a sample of homogeneous universal life insurance policies ranged from $9,105 to $19,807. At the individual policy level, it is clear that policy expenses, mortality charges and surrender charges explain the majority of variation across policies. In addition, it seems reasonable that characteristics of individual insurers (company expenses, lapse ratios, organizational form, yield on investments, etc.) also are related to the level of surrender values that an insurer provides.

A greater understanding of factors external to the policy itself and those related to policy performance will provide important information to insurers, financial-services professionals and others.

A Look at Universal Life

To provide insight on the relationship between insurance company-level characteristics and life insurance performance, we examined universal life insurance data from 1985 to 1995 for a sample of policies from Best's Policy Reports, 1995, with annual premiums of $1,500. Policy data represent each policy's actual (as opposed to projected) 10-year surrender value and are based on $100,000, type-A (level) death benefit policies for a 45-year-old male nonsmoker. The sample of universal life policies consists of all policies for which there exist complete policy data and company-level data (73 out of 90) for the sample period. This study examines results only for issue-age 45 (male), since data across various initial ages (other than age 45) are not available. Insurance company-level data are from the database maintained by the National Association of Insurance Commissioners. The sample of companies includes stock insurers (41), mutual insurers (22) and mutual-owned stock insurers (10).

The variable of focus here is the 10-year end-of-year surrender value and its relationship to certain insurer characteristics. Several firm characteristics and their expected relationship to the surrender value are discussed below.

Expectations

Insurers with higher lapse ratios generally face higher costs, and thus have less ability to provide superior products, suggesting a negative relationship to the surrender value. Insurers with higher general expenses would have less ability to provide superior products. Insurers with greater yields on investments have relatively greater funds with which to provide superior products. Insurers with higher premium income may have greater premium volume by providing policies with superior performance (higher surrender values). This measure also proxies for size, and economies of scale suggest that larger insurers may be more likely and/or able to offer superior products. Insurers with higher ratings tend to be the financially sound insurers most able to offer superior policies. Insurers with greater overall gains relative to income have relatively more funds with which to provide superior products.

An insurer's organizational form may be related to policy performance, in that stock insurers may be more efficient and able to offer superior products. But a commonly stated goal of mutual insurers is to provide coverage at minimum cost to their policyowners--thus, mutual insurers may provide superior products, compared with stock insurers. Insurers that frequently change product mix are likely to be those that take advantage of market opportunities and may be the insurers that aggressively pursue new business by providing superior products. Conversely, insurers that frequently change product mix incur significant expenses and may not be able to provide superior-performing products.

Wide Variations

The table below, "Insurer Statistics in Relation to Policy Surrender Values," shows the correlation coefficients between the surrender value and the variables for several insurer characteristics.

The maximum surrender value in the study ($19,807) was more than two times the minimum surrender value ($9,105). Significant variation also was apparent for most of the other variables of interest. The one notable exception was the A.M. Best rating variable for which all insurers were rated at least A-in this sample.

Correlation coefficients provide a simple view of how one variable moves in relation to another variable, without controlling for the possible effect of other confounding variables. For example, if lapses and surrender values increase commensurately, the correlation coefficient would be 1. Conversely, if lapses increase and surrender values decrease commensurately, the correlation coefficient would be negative 1. If lapses and surrender values are not related at all, the correlation coefficient would be zero. The correlations in the table indicate that policy surrender values are negatively related to lapse ratios (-.47) and positively related to investment yields (.23). The correlations also suggest that stock companies are more likely to offer policies with lower surrender values (-.22). Although the table does not show other correlations, it should be noted that organizational form and lapse ratios are strongly positively correlated, indicating that mutual insurers tend to have lower lapse ratios than sto ck insurers.

Results from further (regression) analysis generally are consistent with the correlations above. Regression results show that the lapse ratio is negatively and significantly related to life insurance surrender values. Thus, empirical evidence indicates that higher lapse ratios impose significant costs on insurers that are passed on to consumers via policies having lower surrender values. The regression analysis also suggests that insurers with higher general expenses pass along these costs via policies with lower surrender values. In addition, results from the regression analysis indicate that insurer investment yields are positively and significantly related to universal life 10-year surrender values. None of the other variables in the table was significant in the regression model.

Superior Surrender Values

The findings indicate that insurers with lower lapse ratios, lower general expenses and higher investment yields generally provide universal life products with superior 10-year surrender values. Because the data are for policies providing $100,000 of universal life coverage for 45-year-old nonsmoking males, the disparity in policy performance likely is not due to differences in underwriting across insurers. Similarly, because of the homogeneity of the sample of policies, the disparity in policy performance likely is not due to any significant difference in the package of options provided by the various contracts.

It also is noteworthy that the regression results do not indicate a significant difference relative to universal life policy performance based on premium income, insurer rating or between stock and mutual insurers, In the debate on insurer demutualization, one argument against conversion from the mutual form has been that mutual insurers offer lowercost. life insurance compared to stock insurers. Results here suggest that the primary indicator of relatively high surrender values is the insurer's lapse ratio. While the regression results do not indicate that organizational form is significantly related to 10-year universal-life surrender values, simple correlation analysis indicates that mutual insurers tend to have lower lapse ratios than stock insurers.

Findings also should provide additional motivation for insurers, financial-service professionals and regulators to continue to focus on methods to reduce policy lapses. Reduced lapse ratios will benefit insurers in an increasingly competitive market in which companies must find ways to reduce costs. In turn, policyowners should benefit from reduced lapse ratios via policies with superior cash values.

The study's findings indicate that insurers with lower lapse ratios, lower expenses and higher investment yields generally provide universal life policies with higher surrender values. For this study's sample of insurers, the average lapse ratio over the 10-year period from 1985 to 1995 was 12%. Insurers that are able to reduce costs, especially costs associated with policy lapses, will be able to offer products with superior performance.

James M. Carson is interim director and Katie research professor at Illinois State University In Normal, Ill., and editor of the Journal of Insurance Issues. Randy E. Dumm is assistant professor of insurance at Florida State University in Tallabassee, Fla.

Insurer Statistics in Relation To Policy Surrender Values

A look at a sample of universal life policies from 1985 to 1995 shows that certain characteristics of an insurance company have direct or inverse relationships to a policy's surrender value. The correlation coefficients below range from-1 to 1, representing the degree to which each of the variables relates to the surrender value.

                           Expected
Insurer                  Relationship
Variable                  to S.V. [1]  Correlation [2]
Lapse Ratio                   -             -.47 [***]
General Expenses              -             -.08
Yields on Investments         +              .23 [**]
Premium Income                +              .01
A.M. Best Rating              +              .13
Overall Gains to Income       +             -.16
Organizational Form          +/-            -.22 [*]
Change Product Mix           +/-            -.11

(1.)S.V.: Surrender Value

(2.)(***.)(**.)and(*.)are significant at the .01, .05 and .10 level, respectively. Correlation coefficients range from 1 (two variables increase commensurately) to -1 (one variable increases and the other decreases commensurately).