Book versus Market Underwriting Betas
首次直接比较基于账面收益和市场收益的承保贝塔，发现两种方法在相同样本上结果不同，且账面贝塔对市场因素更敏感，对保险业资本定价研究有重要参考价值。
Book Versus Market Underwriting Betas Abstract Previous researchers have estimated underwriting betas for application within the Capital Asset Pricing Model to determine the equilibrium rate of return to insurers. Both book and market returns have been used to estimate these risk measures, but previous studies have not been directly comparable and results have been inconsistent. This study provides the first direct, pairwise comparison of underwriting-specific betas derived from both book and market returns and considers the implications for researchers. The underwriting-specific Capital Asset Pricing Model (CAPM) has been proposed to derive equilibrium rates of return to property-liability (P-L) insurers for purposes of both external regulation and internal capital budgeting (Fairley, 1979; Hill and Modigliani (HM), 1987; Venezian, 1983). The focal point of the empirical research is estimation of the single risk parameter, the underwriting beta. While these often are extracted from market data, some researchers propose direct estimation using book returns (Biger and Kahane, 1978; Cummins and Harrington (CH), 1985). CH supply particularly counter-intuitive evidence that such book-derived underwriting betas are more sensitive to market factors than are market-derived estimators. Researchers examining a book or market procedure have tended either to ignore the alternative estimation method or to compare their results with previous research using different samples and time periods. Consequently, two remaining research questions include whether the two methods produce different results over identical samples and whether either method is more sensitive to market factors over time. This study addresses these issues by supplying the first pairwise comparison of book and market underwriting betas over identical time periods. The previous literature and the rationale for this study are discussed in the next section. Previous Research Consistent with contemporary financial theory, researchers exploring underwriting-specific pricing models generally assume that market prices best impound investors' views of future cash flows to insurers (Kahane, 1979). Estimation of underwriting-specific betas is particularly difficult, however, because insurer equities reflect returns on both underwriting operations and invested reserves. Fairley (1979) developed a procedure, later refined by HM, whereby a weighted proxy for an insurer's investment portfolio beta is deducted from the observable equity beta to produce a residual beta purported to be the underwriting beta. CH, however, noted inherent problems with sample size, accurate estimation of investment portfolio betas, and disaggregation by underwriting line. An alternative, direct procedure was introduced by others, who initially regressed annual book returns to insurers on market index returns (Biger and Kahane (BK), 1978; Hill, 1979). Both BK and Hill found these betas do not differ significantly from zero, the latter suggesting that lag time between market adjustments and book profits creates a bias toward zero. A further confounding factor is provided by evidence that insurers smooth underwriting returns via manipulation of loss reserves (Weiss, 1985). CH addressed the problem of bias by applying maximum likelihood (ML) regression models to quarterly book returns for 14 insurers during the periods 1970.1 through 1975.3 and 1975.4 through 1981.2. CH confirmed that simple regression betas closely approach zero, but five-period, lagged betas shifted significantly from positive to negative over the two periods. They noted that market underwriting betas generated by HM for the period 1976 through 1980 are not substantially negative and further asserted that the apparent differences may be attributable to errors in the Hill-Modigliani parameter estimates (p. …