Because the unique characteristics that may influence a firm’s risk taking differ across industries, we further separate our regression analyses of life insurers and property-liability insurers and include specific control variables that may affect each industry. Prior studies on corporate governance (e.g., Bromiley 1991; Wright et al. 1996) argue that corporate governors may make decisions today, but the impact of those decisions on risk taking may be reflected in the future. Thus, our regression analysis includes the time lag between corporate governance and risk.
The regression equations for life insurers are specified as follows:
(1)
and those for property-liability insurers are:
, (2)
where j = 1, 2, and 3. Risk1 equals business risk measured by the standard deviation of the
20
log of underwriting income; Risk2 equals investment risk, which is measured by the standard deviations of the proportion of stock investment in the total invested assets; and Risk3 equals the standard deviations of ROE. Following Gorton and Rosen (1995) and Wright et al. (1996), we examine whether the relationship between risk taking and the share of insider stock ownership is inversely U-shaped by including INSIDER and INSIDERSQ in regression analysis, where INSIDER is the percentage of shares owned by the director and officers and INSIDERSQ is the square of INSIDER. We also include cash-flow rights (CR) variable in the regressions, such that CR is the cash-flow rights held by controlling shareholders. As we stated previously, the deviation between voting rights and cash-flow rights (DEV) may influence a firm’s risk taking.
Following La Porta et al. (2002), Claessens, Djankov, and Lang (2000), and Yeh (2005), we adopt two measures, including the difference between voting rights and cash-flow rights and the ratio of voting rights to cash-flow rights, to represent the magnitude of deviation between voting rights and cash-flow rights. We also distinguish the cash-flow rights held by the controlling shareholders from insider ownership. Because insiders mainly receive salary and/or bonus compensation, whereas the controlling shareholders generally receive benefits from their cash-flow rights, we must examine the impact of the cash-flow rights of controling shareholder and the ownership of director and officers on risk taking separately.
To determine whether the board plays an important role, we include four board composition variables—ownership concentration (OWNCON), board size (BSIZE), board independence (BIND), and CEO duality (DUAL)—that we deem important for effective board monitoring in our regression. The OWNCON variable equals the percentage of shares owned by the first 10 largest stockholders; BSIZE is the number of members on the board; BIND is the board independence, which equals the percentage of outside directors among the board members; and DUAL is a CEO duality dummy, which equals 1 if the CEO also serves as the chairperson of the board and 0 otherwise.
We include the following control variables: Herfindahl concentration index (HI), firm size (SIZE), and capital-to-assets ratio (CAP). For the property-liability insurance industry, we also include the reinsurance ratio (REINS). HI is used to control for a possible influence from line of business on risk taking by firms; SIZE is the natural log of total assets; CAP is the ratio of equity capital to total assets; and REINS is the ratio of reinsurance premiums ceded to the sum of the direct premiums written and reinsurance premiums. For each risk variable, we also employ a return control variable (RETURN). We use the ROE standard deviation and business risk and the book value of rate of return on assets for investment risk to control for a possible influence from return on risk.2 Finally, we include Year 2000 and Year 2001 as year dummies.
III. Research Results and Conclusion
This paper investigates the impacts of corporate governance structures on risk taking in the property-liability and life insurance industry. Our study thus extends existing literature on corporate governance on two main fronts. First, the corporate governance system of the insurance
2 We do not use the market value of ROE and ROA, because most insurers in Taiwan are not publicly traded firms and data on market value are not available for most life and property-liability insurers.
industry in Taiwan, which holds board members and managers fully responsible for bankruptcy, offers an interesting environment in which to explore its unique regulatory impact on insurers’
risk taking behavior. Second, Taiwanese insurance firms are primarily family controlled, with a high degree of ownership concentration. In such an ownership background, corporate governance mechanisms are more important as a means to prohibit possible misconduct by controlling shareholders, especially in the insurance industry, where insurers use their great financial leverage to conduct business. To our knowledge, ours is the first study to examine the effects of voting rights and cash-flow rights ownership on insurers’ risk taking behavior in the insurance industry.
In general, our results provide evidence in support of the claim that corporate governance structure plays an important role in influencing insurers’ risk taking behavior even under sticker regulatory rules. Consistent with prior literature, we find that deviations between voting rights from cash-flow rights have positive impacts on profit risk for life insurers and investment risk for property-liability insurers. However, the evidences should be interpreted with caution because some coefficients on these variables which were statistically significant in the prior studies are not statistically significant. Thus, the results implies that stricter liability constraint may discourage insurers’ risk taking behavior or may reduce/replace the effect of some corporate governance factors on insurers’ risk taking behavior. This result is consistent with the results of Saunders, Strock, and Travols (1990) and Esty (1998). It is an interesting finding of this paper and provides us with additional insight concerning corporate governance research filed.
The relationship between investment risk and insider ownership is inversely U-shaped for both segments of insurance industry. Furthermore, for life insurance, board independence and board size is negatively related to business risk; CEO duality relates negatively to investment risk;
higher cash flow right and board independence may discourage profit risk. Among property-liability insurance firms, CEO duality, board independence, and board size also have negative impacts on insurers’ different risks. In particular, CEO duality relates negatively to business risk; the relationship between investment risk and board independence is negative;
board independence and board size have negative impact on profit risk.
Additional findings suggest that compared with property-liability insurance companies, corporate governance variables have greater impact on the risk taking behavior for life insurance industry. The empirical results seem to further imply that the difference of business practice between these two segments of insurance industry may lead to different impact on the risk taking.
Comparing to the property-liability insurers, life insurers have higher leveraged equity and exposure position. Thus, life insurers bear higher liabilities and greater solvency risk. As a results, the corporate governance variables may have greater impacts on discouraging the risk taking behavior of life insurance companies to avoid insolvency problem.
The implication of our findings is that some of the corporate governance mechanisms have real and predictable effect on monitoring risk taking by insurance firms. In addition, our evidence shows that stricter liability rule, which holds board members and managers fully responsible for cases of bankruptcy, does not effectively discourage risk taking as a whole. Insurance regulator may minimize excessive risk taking by alternative means, such as higher minimum capital
22
requirement, stricter risk-based capital rule, or prompt disclosure about improper transaction.
The analysis of this paper suggests some future avenues for future research. First, our findings imply that stricter liability constraint may reduce/replace the effect of some corporate governance factors on insurers’ risk taking behavior. Whether unlimited liability rule has negative effects on insurers’ risk taking deserves further study. The second potential avenue for future research is to examine whether different insurance regulation has different effects on insurers’
risk taking behavior. For example, the risk-based capital regulation and the investment restriction in Taiwan may discourage the risk taking by insurance firms. Thus, a comparison of between insurers’ risk taking among different countries would provide more insights to this issue. We also encourage additional studies to examine whether poor corporate mechanisms increase the probabilities of financial distress or bankruptcy by using different proxies to measure risk in the insurance industry.