روابط میان خطر دارایی ها، خطر محصول، و سرمایه در صنعت بیمه عمر

This paper explores the relation between capital and risk in the life insurance industry in the period after the adoption of life risk-based capital (RBC) regulation. To examine this issue, we use a simultaneous-equation partial-adjustment model. Three equations express the interrelations among capital and two measures of risk: product risk and asset risk. The asset-risk measure used in this paper reflects credit or solvency risk as in RBC. Product risk assessment for life insurance products is rationalized by transaction-cost economics – contractual uncertainty. A significant finding is that for life insurers the relation between capital and asset risk is positive. This agrees with prior studies for the property/casualty insurance industry and some banking studies. But the relation between capital and product risk is negative. This is consistent with the hypothesized impact of guarantee funds in other studies. The contrast between the positive relation of capital to asset risk and the negative relation of capital to product risk underscores the importance of distinguishing these two components of risk.

This is the first study to look at the simultaneous interrelation among capital, asset risk and product risk in the life insurance industry using the framework of Shrieves and Dahl (1992) and Cummins and Sommer (1996). For the life insurance industry, academic research in this area has concentrated mostly on the influence of the life risk-based capital (RBC) regulatory tool. Pottier and Sommer (1997) compared the life RBC results with those of the insurance industry's rating organizations and Ryan and Schellhorn (2000) examined the impact of the life RBC law on life insurer's efficiency. The interrelation between capital and risk for the banking industry received attention from Shrieves and Dahl (1992) and Jacques and Nigro (1997), among others. Berger et al. (1995) provide a survey of capital structure studies in the banking industry. In the insurance industry, Cummins and Sommer (1996) examined the interrelation between capital and risk, but only for the property/casualty insurance industry. In this study, we apply the methods used by Shrieves and Dahl (1992) and Cummins and Sommer (1996) to the life insurance industry in the post-RBC era. The banking literature presents somewhat inconsistent empirical results on the interrelation between the capital-to-asset ratio and asset risk. Shrieves and Dahl (1992) found a positive relation between the capital-to-asset ratio and asset risk for the period of 1983–1986, but Jacques and Nigro (1997) found a negative relation between capital and asset risk for 1991. Berger (1995) found that the level of capital to asset ratio was negatively related to the level of portfolio risk in a study of the relation between capital and earnings in banking for the period of 1983–1989. The asset risk measure of these studies was based on the 1988 Basle Accord RBC guidelines. For the property/casualty industry, Cummins and Sommer (1996) found a positive relation for capital and risk levels in 1979–1990. They aggregated asset and product risk into a single portfolio risk measure using a model based on option pricing. A positive relation between capital, on the one hand, and asset risk or product risk, on the other hand, is consistent with agency theory, transaction-cost economics theory (Williamson, 1988) and insurers' preference to avoid bankruptcy costs (Cummins and Sommer, 1996; Shrieves and Dahl, 1992). Transaction-cost economics (Williamson, 1988) assumes that when the products sold by the firm are riskier, debt financing is harder to obtain because of greater uncertainty that the firm will fulfill its contractual obligation to repay. Thus, firms that sell products with greater risk such as health insurance are expected to hold more capital. Additional theoretical explanation for the positive relation between risk and capital (Shrieves and Dahl, 1992) is that a firm will adopt lower leverage levels because of regulatory costs, unintended effects of minimum capital standards, and bankruptcy cost avoidance considerations. A negative relation between capital and risk is consistent with the hypothesis that deposit insurance for banks and guarantee funds for insurers induce greater risk taking at lower capital levels (Cummins, 1998). Lee et al. (1997) express this idea as the risk-subsidy hypothesis and show its applicability to stock insurers. Downs and Sommer (1999) also show that it holds for publicly traded insurers with some insider ownership. In addition to systematic or macroeconomics risk, insurers face specific risks that arise from the written policy contracts and their invested assets. These are termed product risk and asset risk, respectively. Management of these two risks represents major aspects of an insurer's operations. For that reason, we focus only on the simultaneous influence of asset risk and product risk on the capital ratio of life insurers, although other risk types exist3 (Santomero and Babbel, 1997; Gleason, 2000). For a capital measure, we take the adjusted book value of capital, as defined in the life RBC rules, divided by total firm assets.4 The level of product risk is a measure of exposure to health insurance writings, which we rationalize by transaction-cost economic theory as applied to the life insurance industry. The product risk actually reflects the contractual risk of the relational and incomplete health insurance product as we interpret Williamson (1985) and as reflected in the life RBC.5 We base our measure of asset risk on a modification and approximation of the “regulatory” definition given in the RBC rules for life insurance.6 Thus our asset risk corresponds to the actuarial concept of credit/solvency risk as defined by Shrieves and Dahl (1992), Jacques and Nigro (1997), and Berger (1995) for the banking industry, and by Santomero and Babbel (1997) for the insurance industry. Our asset risk is not a measure of the financial market risk of assets, as defined by Cummins and Sommer (1996). Since this study focuses on the post-RBC era, a regulatory asset risk measure is selected for this study. This selection also allows us to compare our results for asset risk to those in the literature on the banking industry. Using a simultaneous-equations model, we find a positive relation between capital ratio and regulatory asset risk, but a negative relation between the capital ratio and product risk. Caution must be exercised in comparing these results with those of other studies that do not separate the asset risk from the product risk, or that use different definitions of asset risk. Our results appear to be partly consistent with the results of Shrieves and Dahl (1992) for the banking industry. The comparison with the property/casualty study of Cummins and Sommer (1996) is less relevant since they used a different proxy of risk and did not separate the asset from the product risk. The paper is organized as follows: Section 2 reviews hypotheses for relations between risk and capital, describes the life insurance industry's products and the definition of the endogenous variables – capital ratio, asset risk and product risk. Section 3 explains the data and the model, while Section 4 presents the results. The paper concludes with a summary.

This study has explored the interrelations among capital to asset ratio, asset risk and product risk in the life insurance industry in the post-RBC era. As in the banking industry, arguments can be advanced in favor of either a negative or a positive relation between capital and risk. The existence of guarantee funds argues for greater risk taking at lower capital levels. This is articulated in the risk-subsidy hypothesis of Lee et al. (1997). Bankruptcy-cost-avoidance, agency theory, transaction-cost economics theory and other arguments favor a positive relation between capital and risk. Each hypothesis has found empirical support in the banking literature. But only the positive relation argument found support in the property/casualty study by Cummins and Sommer (1996). In this study, we apply methods used both in the banking and property/casualty industries and found support for a positive relation between the regulatory asset risk and capital ratio, but a negative relation between product risk and capital ratio. In our structural equations, we see that at given levels of asset risk, firms with higher product risk are found to have lower capital ratios, ceteris paribus. But at given levels of product risk, firms with higher asset risk are found to have higher capital ratios. This result raises an issue for regulation. Regulators prefer that capital increase with rising levels of each type of risk. Yet the separate equation for capital describes behavior at variance with regulatory preference. It may be that firm behavior with respect to capital allocation is more complex than the simple and essentially additive risk model currently used in Life RBC regulation. It also underscores the importance of distinguishing different risk categories.