EFFICIENCY ANALYSIS
9.2 REGRESSION RESULTS
The regression results using each of the estimation techniques are presented in Tables 16 and 17. Model 1 (or year) indicates one-way fixed effects regression, which controls for time-specific effects that are not otherwise controlled for by other variables, while model 2 (or year+firm) controls for both time and company fixed effects.40 F statistic is presented for all regressions to investigate whether there is company and time fixed effects. The large F statistic rejects the null hypothesis in favor of the fixed effect model (p<.0000).41
We conduct overall regressions for cost and revenue efficiency and analyze the decomposition of cost efficiency by also conducting regressions where the dependent
40
g one-way and two-way random effects model. The results of random We also estimated regressions usin
effect models are consistent with those of the fixed-effect models. 41
The null hypothesis is that parameters of company and time dummies are zero:H0 =µ1= =... µn−1=0 and τ1 = =... τT−1 =0.
variables are pure technical, scale, and allocative efficiency, respectively. Although technical and allocative efficiency provide important indicators, the discussion is mainly focused on the cost and revenue efficiency regressions because they determine the profitability of firms and provide the best measures of overall firm performance.
The regressions, shown in Table 16, reveal significant differences in cost and revenue efficiency of acquiring firms between before and after M & A. The coefficients on MAitin both cost and revenue efficiency are negative and significant, supporting the hypothesis that acquiring firms experience significantly larger losses in cost and revenue efficiency after M & A. This result suggests that expansion of the firm through M & As has the potential to create inefficiencies. As firms become larger and more complex, diversification benefits are offset by the additional costs. Administrating and operating over wider geographical areas and integration of different information systems can lead to higher costs. Bonding different organizations have more potential to create managerial conflict and agency costs since managerial monitoring becomes more difficult.
The signs of explanatory variables are generally consistent with theoretical predictions. The coefficients estimated on firm size are positive and significant in all types of efficiency, with the exception of scale efficiency, indicating that larger firms tend to experience more efficiency than smaller firms. This result is consistent with prior findings (Cummins and Zi, 1998). However, firm size is negatively related to scale efficiency. A possible explanation for this result is that as Cummins and Xie (2005) find that the majority of firms above median size are operating with decreasing returns to scale, firms with DRS may not attain scale efficiency.
We document a significant and positive relationship between the percentage of loss incurred in both personal property and personal liability lines and cost efficiency. Thus, it appears that insurers with a higher proportion of business in personal property
and personal liability lines obtain greater cost efficiency than those with more business in commercial property lines, suggesting that types of business and their combination has an
porta
exposed to catastrophic propert
im nt role in improving insurer’s efficiency.42 This result is consistent with Cummins and Xie (2005). Surprisingly, the percentage of loss incurred in commercial liability lines are inversely related to cost efficiency, indicating that firms with emphasizing commercial liability lines are likely to create cost inefficiency. As shown in Table 17, the decomposition regression shows that the primary source of cost efficiency gains in personal property and personal liability lines is pure technical efficiency, suggesting that automated systems are more advantageous in the personal lines. Because long-tail commercial liability lines such as medical malpractice and workers compensation are more complex and the pattern of loss payment are more uncertain, allocating resources and adopting new technology and marketing systems are relatively difficult in commercial liability lines.
The percentage of loss incurred in personal property lines and personal liability lines is significantly negatively related to revenue efficiency. Thus, firms with a higher proportion of business in personal lines are less advantageous in output-oriented revenue efficiency. The negative relation may be induced by the fact that personal property lines that include homeowners and earthquake insurance are highly
y risks from hurricanes and earthquakes. In addition, because personal liability lines such as primary personal auto liability is written as a compulsory insurance along with auto physical damage, it may be more difficult to choose optimal output combinations to maximize revenue efficiency.
The coefficient on the geographical Herfindahl index is positive and significant in both cost and revenue efficiency, supporting the pro-focus arguments that geographically
regression analysis using the percentage of premiums written in personal property lines, personal liability lines, and commercial liability lines and observe similar results.
42
focused insurers are able to achieve greater cost and revenue efficiency than geographically diversified insurers. This result implies that potential benefits from risk diversification are likely to be offset by the extra costs associated with greater managerial discretion, inefficient allocation of resource, and additional administrative and regulatory issues t
fficiency, implying that insurers with higher premium-to-surplus ratios employ less capital input relative to premium revenues. The liquidity ratio calculated by dividing liquid assets (cash and marketable securities) by total liabilities is significantly negatively related to all types of efficiency. A high degree of liquidity enables an insurer to meet unexpected financial needs without the untimely sale of investments or fixed assets, which may result in substantial realized losses due to temporary market conditions or tax consequences. The negative sign on this variable suggests that firms with higher liquidity hat are required to deal with when operating across different states. The product line Herfindahl index is significant and positively related to cost efficiency. However it is not significant in revenue efficiency. The product line Herfindahl index is negative and significant as related to allocative efficiency, consistent with pro-conglomeration arguments that diversified insurers are more advantageous in choosing cost minimizing combinations of inputs than focused insurers.
The underwriting leverage measured as premium revenues net of reinsurance transactions relative to policyholders’ surplus is significant and positively related to cost and revenue efficiency. The net premium-to-surplus ratio is inversely related to the capacity of an insurer to write additional business because new policies generate liabilities, which must be supported by surplus due to regulatory accounting rules. The positive relationship between cost efficiency and underwriting leverage ratio is primarily attributable to pure technical efficiency, which offsets a negative effect of allocative e
ratio to meet financial obligations to pay off reserves by holding cash and quickly convertible investments have lower cost and revenue efficiency.
As predicted, the coefficient on the direct marketing indicator variable is significantly positive in both cost and revenue efficiency. This results support the hypothesis that direct marketing distributions are more cost and revenue efficient than independent agency distributions. This result is contrary to the finding of Berger, Cummins, and Weiss (1997) who provide evidence that there is no difference in revenue efficiency between direct writing and independent agency. The regression results also show that brokers are more cost efficient th but indicator variable for mixed distribution is not significant in cost efficiency. The advantage of direct marketing and brokerage over independent a ency is mostly attributable to pure technical efficiency which also offset the negative eff e efficiency. We also find that mixed distribution is more revenue efficient than independent agency distribution.
Mutual variable has a positive and significant coefficient in cost efficiency, as predicted if mutual insurers have a comp rative advantage in writing less complex business lines where lower managerial dis retion is required, requiring fewer inputs. There is no significant difference in revenue efficiency between mutual firms and stock firms. The unaffiliated single firms are significantly positively related to all type of efficiency, with the exception of allocative ay be able to iversify risks across member companies, whereas unaffiliated single firm may not have ification opportunities. Thus, managers of an unaffiliated company are likely to be
more risk averse and m maximize revenues
an those of stock firms. The negative sign on allocative efficiency may indicate that less resource allocation is conducted at an unaffiliated company level. The coefficient on
an independent agents,
g
ect of allocativ
a c
efficiency. Group insurers m d
divers
ay have more incentive to minimize costs and th
firms licensed in New York is significant and negative, suggesting that stricter regulation leads to cost and revenue inefficiencies, perhaps due to the imposition of regulatory costs.