Directors Liability Insurance and the Value of Excess Cash

Directors Liability Insurance and the Value of Excess Cash Chia-wei Chen Associate Professor, Department of Finance, Tunghai University Pei-ying Chen * Ph.D. Student, Department of Finance, National Chung Hsing University Abstract This study determines whether excess cash holdings affect company value between insured and uninsured companies with directors liability insurance. The results show that excess cash holdings are significantly and positively associated with a company s market value, for insured companies in Taiwan. However, recent studies have suggested that the protection of this insurance may magnify the moral hazard problem and thereby reduce the effectiveness of the board, as well as shareholder wealth. The results of this study demonstrate the benefits of this insurance and enrich current studies with empirical evidence from Taiwan, where the development of directors liability insurance is in an early transitional stage, where there are almost equal numbers of insured and uninsured companies. Keywords: Directors liability insurance, Value of excess cash * Corresponding author. Department of Finance, National Chung Hsing University, No.1727, Sec.4, Taiwan Boulevard, Xitun District, Taichung 40704, Taiwan (R.O.C.) Tel:886-4-23590121 ext.35802; Fax: 886-4-23506834; E-mail address: pychen@thu.edu.tw 1

1. Introduction Directors liability insurance is essentially designed to protect shareholder wealth (see, for example, Bhagat, Brickley and Coles (1987), Brook and Rao (1994), and O Sullivan (1997)). It has been purchased by most corporations in recent years, especially in developed countries 1. Intuitively, this insurance is beneficial and in fact, several nations have been encouraging companies to acquire it. However, it is also possible that this insurance may deteriorate the adverse incentive effects or may foster opportunistic managerial behavior, although prior studies show that protecting directors improves the effectiveness of the board or prevents board conservatism. Specifically, directors protected by this insurance are lax in executing their duty, so due to the lack of proper monitoring, there is increased possibility that managers pursue their own interests at the expense of shareholders. Consistent with this view, empirical works that use event studies (see, for example, Chalmers, Dann and Harford (2002), and Zou, Wong, Shum, Xiong and Yan (2008)) suggest that directors liability insurance may damage company value under certain circumstances. Because it is still an open debate as to whether the protection of directors liability insurance is beneficial to companies, this study reexamines the value of this insurance through an unexplored channel. In particular, it is suggested that the value of a company s excess cash holding may provide useful information to understand the 1 Some related studies use the term, directors and officers liability insurance, or D&O insurance. This study uses the term directors liability insurance because the information of officers liability insurance is not disclosed in Taiwan. However, as companies that purchase directors liability insurance in general acquire the coverage for their officers as well, so these terms are interchangeable throughout this paper. 2

relationship between directors liability insurance and shareholder wealth. Excess cash, on one hand, may increase company value because it offers cheap internal funding for potential investment, or it could destroy value by worsening the agency problem (Jensen (1986), Lehn and Poulsen (1989), and Richardson (2006)). As a result, the quality of corporate governance, as addressed by numerous theoretical and empirical works (see, for example, Dittmar, Mahrt-Smith and Servaes (2003), Dittmar and Mahrt- Smith (2007), and Harford, Mansi and Maxwell (2008)), would presumably be crucial in deciding the ultimate effect of excess cash holding on company value. Prior studies of corporate governance have focused on many different internal and external mechanisms, such as board characteristics (Daily and Dalton (1994)), ownership structure (Singh and Davidson (2003)), CEO duality (Rechner and Dalton (1991)), shareholder rights (Gompers, Ishii and Metrick (2003)), and markets for corporate control (Jensen and Ruback (1983)), in order to prevent corporate wrongdoings. However, directors liability insurance, a mechanism designed to protect shareholder wealth in the event of corporate wrongdoings or lawsuits, is rarely discussed. If directors liability insurance indeed encourages directors to perform their duty to maximize shareholder wealth, the quality of corporate governance must definitely be enhanced. Therefore, the value of holding excess cash must be guarded or even increased. However, if the protection provided by this insurance actually provides a disincentive for directors to oversee company executives, holding excess cash could increase the threat of managerial expropriation, which would decrease the value of 3

excess cash, as well as shareholder wealth. Similarly, all else being equal, if the protection provided by directors liability insurance does not affect board behavior and the quality of corporate governance, then there should not be a significant difference in the value of excess cash held by insured and uninsured companies. With a sample of 4,898 firm-year observations from 1,302 listed companies in Taiwan, from 2008 to 2011, the empirical results indicate that directors liability insurance is associated with a higher value of excess cash holding 2. This finding, in contrast to recent studies that highlight the costs or drawbacks of this insurance, gives positive support for directors liability insurance. As the results remain robust, using subsamples and alternative measures, consistent with the view that governance has a substantial impact on value because of its influence on cash, as documented by Dittmar and Mahrt-Smith (2007) and Frésard and Salva (2010), a similar effect is associated with directors liability insurance. This study s finding makes two primary contributions to the existing literature. Firstly, it indicates the possible impact of directors liability insurance on company value, because of excess cash. Prior research has shown that either directors liability insurance or the value of excess cash is correlated to the quality of corporate governance, to some extent. This study shows that directors liability insurance can potentially improve the functioning of the board, so it can better prevent the misuse of corporate resources. As a result, the value of internal funding is expected to be increased. Some 2 The information on directors liability insurance for listed companies in Taiwan has been publicly available since 2008. 4

studies have tried to determine the relationship between directors liability insurance and shareholder wealth, using event studies, such as IPOs (Chalmers, Dann and Harford (2002)), SEOs (Boubakri and Ghalleb (2008)) and M&As (Lin, Officer and Zou (2011) and Chen, Lin and Yi (2013)). Others have analyzed financial or accounting characteristics, such as earnings conservatism (Chung and Wynn (2008)), CEO compensation (Campbell (2010)), and stock liquidity (Chen, Lin and Yi (2011)). This paper highlights excess cash holding as an important factor in determining the relationship between director s liability insurance and shareholder wealth. It is suggested that directors liability insurance better prevents misuse of excess cash, possibly through enhanced board function or external monitoring. Hanson (1992) argues that bidding companies with abundant cash flow pursue low-benefit takeovers. Similarly, Vafeas and Joy (1995) document that abnormal stock market returns are related to a reduction in agency costs, associated with free cash flow. Since shareholder wealth is unambiguously determined by the efficiency of a company s fund allocation, the positive correlation between directors liability insurance and the value of excess cash implies that directors liability insurance reduces corporate wrongdoings and consequently increases shareholder wealth. Secondly, these empirical results are based on a unique dataset, which is very different from the data used by previous studies. Prior studies have mainly used observations from developed countries, in which the majority of companies are already covered by directors liability insurance (see, for example, Core (1997), Chalmers, 5

Dann and Harford (2002) and O Sullivan (2002)). However, the numbers of insured and uninsured companies in Taiwan are almost identical, because the development of directors liability insurance is still in its early transitional stage. Furthermore, this sample provides evidence for Taiwan s market, where the business environment and corporate characteristics, such as business regulations and ownership structure, are significantly different to those of western countries. For example, Claessens, Djankov and Lang (2000) find that East Asian countries tend to have concentrated ownership and their voting rights frequently exceed cash-flow rights, because of pyramid structures and cross-holdings. Claessens and Fan (2002) further indicate that the limited protection of minority rights in Asia allows controlling shareholders to expropriate minority shareholders. Apparently, corporate governance mechanisms work differently in Asian countries, because of the different business environment and corporate characteristics. If the board cannot function effectively, directors liability insurance may not necessarily be enough to force directors to execute their duty effectively, because they may only have limited power to influence a company that has concentrated ownership. Chen and Li (2010) document that directors liability insurance seems not to have a significant impact on the directors care incentive, in Taiwan. Moreover, the protection provided by the insurance may even provide a disincentive to the insured directors. However, if directors liability insurance fails to provide an incentive to directors, the external monitoring associated with it may still motivate the insured companies to 6

act in the best interests of their shareholders. Consistent with this view, Lu and Horng (2007) find supporting evidence for the monitoring role of directors liability insurance. In addition to existing literature, this study shows that the value of excess cash is higher for companies covered by directors liability insurance. This suggests that at least the insurance may prevent the insured companies from misusing internal funding. Yamori (1999), Regan and Hur (2007) and Zou, Wong, Shum, Xiong and Yan (2008) use observations made in Japan, Korea and China, respectively, to offer comparable evidence for a better understanding of the role of directors liability insurance in Asian countries. As the debate on directors liability insurance continues, this study s findings from observations made in Taiwan contribute to an increased understanding of the benefit of the insurance. The remainder of this paper is organized as follows. Section 2 summarizes relevant literature and develops the hypotheses. Section 3 describes the sample. Section 4 explains the variables used. Section 5 presents the empirical methods and results. Section 6 documents the robustness tests and Section 7 concludes this paper. 2. Literature review and hypothesis development Directors liability insurance, as addressed in prior literature, protects the wealth of the insured directors in the event of corporate lawsuits, as long as the directors act honestly and in good faith. Intuitively, this insurance is beneficial from the shareholders point of view. Directors liability insurance not only provides expertise to insured 7

companies in dealing with litigation, but also reduces the associated costs. Core (1997) uses a sample of Canadian companies to show that companies with a greater litigation risk are more likely to acquire this insurance. Similarly, Mayers and Smith (1982) and Romano (1991) document that a corporation s insurance contracts can allocate risk to claimholders, provide an efficient service for claims administration and reduce regulatory costs. Additionally, in order to avoid the potential costs associated with moral hazard and adverse selection problems in the insurance market (Akerlof (1970)), insurance providers can serve as an external monitor of the insured company (Holderness (1990)). Consistent with this view, O Sullivan (1997) uses a sample of 366 companies in the United Kingdom to show that directors liability insurance could be utilized to monitor managers, especially in large companies. Core (2000) provides confirmatory evidence to show that the insurance premium contains valuable information about the quality of the corporate governance of the insured company, implying that insurers indeed investigate and monitor the insured companies. Boyer and Stern (2012) document that insurers have a financial incentive to correctly assess the governance risk of a company that wishes to acquire directors liability insurance, so the assessment information is an unbiased indicator of the company s governance risk for investors. Other than cost savings and external monitoring, as mentioned, protecting directors wealth may encourage directors to more diligently perform their fiduciary duty, which leads to greater board effectiveness. As a result, shareholder wealth is also 8

improved. Brook and Rao (1994) report that liability limitation provisions affect experts expected costs in serving as directors, especially in troubled companies. Bhagat, Brickley and Coles (1987) provide empirical evidence to show the positive effect of directors liability insurance on shareholder wealth. Gutiérrez (2003) and Boyer (2005) both suggest that the insurance protects shareholders wealth more than it does directors. Chen, Lin and Yi (2011) find that companies with directors liability insurance have greater stock market liquidity, which implies that companies with this insurance seem to be more attractive to investors. In summary, the literature shows that directors liability insurance can be beneficial to companies. However, because the data related to this insurance remains limited in most nations, it is extremely difficult for researchers to reach a consensus, based on empirical evidence, so the value of directors liability insurance to companies is still open to debate 3. Baker and Griffith (2006), for example, document that this insurance neither helps monitor corporate governance nor manages litigation costs. Similarly, Chen and Li (2010), using listed companies in Taiwan, find no evidence to show that this insurance has a significant impact on directors care incentives and the company s subsequent performance. Boyer (2003) finds no evidence of managerial opportunism in regard to directors insurance coverage and documents that corporations may not choose optimally their risk management decisions; rather they may rely more on a force of habit than on a clear and concise strategy to manage corporate risk. Romano (1989) 3 Griffith (2006), for example, documents that Corporations lack the incentive to disclose this information on their own initiative, and U.S. securities regulators do not require registrants to do so. 9

argues that affordability is instead the main reason for companies to purchase directors liability insurance. Other than academics who believe that the relationship between directors liability insurance and shareholder wealth has been positive or insignificant, in recent years, there is some research evidence to suggest that the relationship between the two is actually negative. For instance, Chalmers, Dann and Harford (2002) study a sample of companies that had an IPO between 1992 and 1996 and find a significant negative relationship between the three-year post-ipo stock price performance and the insurance coverage purchased in conjunction with the IPO. Potentially, purchasing directors liability insurance may be a signal of opportunistic behavior in company managers. Lin, Officer and Zou (2011) report that acquirers with insurance protection tend to pay higher acquisition premiums, but these acquisitions appear to provide less synergy. Similarly, Chen, Lin and Yi (2013) find that target companies for M&As that have directors liability insurance tend to have lower cumulative abnormal returns around the announcement date. Zou, Wong, Shum, Xiong and Yan (2008) use listed companies in China to show that the announcement to purchase directors liability insurance has a negative effect on the wealth of companies that tend to have stronger incentives to engage in earnings management. Chen, Li and Zou (2012), using data from Canada, conclude that this insurance is associated with incentive problems, so investors penalize insured companies by requiring a higher cost of equity. 10

A review of prior works related to directors liability insurance provides a summary of the pros and cons of having this insurance, but it seems clear that the effect of this insurance on wealth is still an empirical issue and could be sensitive to both research data and methodology. The authors believe that there are some unexplored and interesting aspects that provide more information for a better understanding of the effect of directors liability insurance on wealth. Specifically, this study focuses on an analysis of the potential link between directors liability insurance and the value of excess cash holding. As Frésard and Salva (2010), among others, empirically find that investors valuation of excess cash mirrors how they expect the cash to be used, a positive relationship between the quality of a company s governance and the value of excess cash reserved in the company is expected. Consistent with the concept proposed by Jensen (1986) that agency costs can be increased by increasing free cash flow, if better governed companies can prevent the improper use of excess cash (Dittmar and Mahrt- Smith (2007)), then the effect of directors liability insurance on wealth can be verified by measuring the value of excess cash. Richardson (2006) suggests that certain governance structures, such as the presence of activist shareholders, appear to mitigate the over-investment problem. Following from this idea, if directors liability insurance indeed motivates directors to maximize shareholder wealth (see, for example, Bhagat, Brickley and Coles (1987), Brook and Rao (1994), and O Sullivan (1997)), the free cash is either invested in good projects that have the greatest net present values or it is distributed back to shareholders, 11

rather than used wastefully (Oswald and Young (2008)). Chen, Chen and Wei (2011) indicate that the effect of shareholder rights on reducing the cost of equity is significantly stronger for companies with more severe agency problems due to free cash flows, which supports the notion that excess cash holding deteriorates the problems caused by agency conflicts, so well-governed companies are likely to experience less improper use of internal funding. Ideally, directors liability insurance is supposed to enhance board effectiveness by protecting directors wealth, in order to improve the quality of corporate governance and consequently increase the value of excess cash 4. While the authors believe that directors liability insurance and the value of excess cash may be related to the quality of corporate governance, unlike the straightforward relationship between the value of excess cash and corporate governance, the relationship between directors liability insurance and the quality of corporate governance is complex. The studies that argue that directors liability insurance strengthens board effectiveness and prevents misuse of excess cash are discussed, but other studies assert that the costs of moral hazard associated with this insurance may outweigh its benefits and therefore increase the likelihood of corporate wrongdoings and eventually destroy shareholder wealth (see, for example, Chalmers, Dann and Harford (2002) and Zou, Wong, Shum, Xiong and Yan (2008)). Some studies also suggest that board effectiveness and company value are not correlated to directors 4 There are still several well-cited studies related to the excess cash or free cash flow (see, for example, Stulz (1990), Lang, Stulz and Walkling (1991) and Harford (1999)). However, this study focuses on the linka between the excess cash and the quality of corporate governance, which could possibly be altered by carrying directors liability insurance. Therefore, in the literature review, there is only a selective review of papers that directly investigate the relationship between company governance and the value of excess cash. 12

liability insurance (see, for example, Romano (1989), Boyer (2003), and Chen and Li (2010)). If there is a positive (non-positive) relationship between directors liability insurance and board effectiveness, a positive (non-positive) relationship between this insurance and the value of excess cash is also expected. In other words, if directors liability insurance reduces the incentive for directors to better perform their fiduciary duty or worsens the moral hazard problem, the quality of corporate governance and the value of excess cash are lower. Kalcheva and Lins (2007), using observations from 31 countries, find that when shareholder protection is weak, company values are lower when controlling managers hold more cash. Similarly, if directors liability insurance does not influence directors care incentives, the value of excess cash in insured and uninsured companies should not be significantly different. Because of the lack of consensus on the value of directors liability insurance, it is clear that the effect of this insurance on wealth remains an empirical issue. This study s testable hypotheses are developed as follows: Ha: The value of excess cash in companies with directors liability insurance is greater than that in companies without it. Hb: The value of excess cash in companies with directors liability insurance is lower than that in companies without it. Hc: The value of excess cash in companies with directors liability insurance is not different to that in companies without it. 13

3. Sample The sample used to test these hypotheses is taken from the Taiwan Economic Journal (TEJ) database, which contains listed companies that are traded on either the Taiwan Stock Exchange (TWSE) or the GreTai Securities Market (OTC) in Taiwan, during the period from 2008 to 2011. 2008 is used as the beginning year of the sample because the mandatory disclosure of directors liability insurance began in 2008. Financial companies are also excluded from the sample, because these companies are subject to more distinct regulations than companies in other industries, so the function of the board and the impact of directors' liability insurance on it could be different. Observations in both the top and the bottom 1 percent are also excluded, to reduce bias from outliers. After eliminating observations with missing variables, the final sample consists of 4,898 firm-year observations for 1,302 companies listed in Taiwan, from 2008 to 2011, which represents about 95% of all listed companies in the sample period 5. 4. Variables This study determines whether directors liability insurance influences the value of a company s excess cash holding. Key variables in the empirical tests are measures of directors liability insurance and the value of excess cash. For the directors liability insurance, a dummy variable is used, in line with Core (1997) and O Sullivan (2002), to show whether a company is insured. The dummy variable equals 1 if the observed 5 It is worthy of note that these analyses demand some variables that require information for two years before the sample year, so the size of the sample could be reduced significantly. 14

company is insured with directors liability insurance in a sample year and is 0 otherwise 6. For the calculation of the value of excess cash, the methods of Opler, Pinkowitz, Stulz and Williamson (1999) and Dittmar and Mahrt-Smith (2007) are used to estimate the excess cash, which is the cash not needed for operations or investments. Specifically, it is the difference between a company s actual cash holdings in a given year and the optimal amount of cash estimated by the following regression. ln(cash it ) = α 0 + β 1 ln(total assets it ) + β 2 ln(cash flow it ) + β 3 (net working capital it ) + β 4 (market to book ratio it ) + β 5 (capital expenditure it ) + β 6 (leverage it ) + β 7 (R&D it ) + β 8 (dividend it ) + ε it In this equation, ln(cash) is the natural log of cash and marketable securities over total assets, ln(total assets) is the natural log of total assets, Cash flow is computed as operating income minus interests and taxes divided by total assets, Net working capital is current assets minus current liabilities and cash over total assets and Market to book ratio is the sum of the market value of equity and the book value of total debt scaled by total assets. The market value of equity is defined as the number of shares outstanding multiplied by the stock s closing price at the year-end. Capital expenditure represents capital expenditure divided by total assets. Leverage is the sum of short- and long-term debt over total assets. R&D is the ratio of research and development expenses to total assets. Dividend refers to the common dividend amount over total assets. 6 Other than a dummy indicator to proxy directors' liability insurance, alternative measures are documented in the robustness tests. 15

The value of the estimation of excess cash, using the methods of Frésard and Salva (2010), is determined by the market to book ratio that is used to predict the optimal cash holdings in the previous equation. To test the hypotheses or the impact of directors' liability insurance on the value of excess cash, the equation below is used 7. As previously mentioned, directors' liability insurance is a dummy indicator that signifies whether the observed companies are protected by this insurance within the sample year. The interactive term, directors' liability insurance * excess cash, shows whether the value of excess cash is altered by carrying directors' liability insurance. Value of excess cash it = α 0 + β 1 (directors' liability insurance it ) + β 2 (directors' liability insurance * excess cash it ) + β N (control variables it ) + ε it The control variables used to remove the effect of potential factors on the value of excess cash contain the measure of the profitability calculated as net income, by adding all noncash charges, extraordinary items and interest, the R&D expenditure, the interest expenses and the dividend. All of these measures, while sensitive to the company s size, are all scaled by total assets. In addition, consistent with prior studies, the change in profitability, R&D expenditure, interest expenses, dividend and net assets from year t- 2 to year t is also incorporated into the regressions. Finally, year indicators are also used to determine the potential difference in each individual sample years. Indeed, any 7 In the main tests, lead changes are not included as control variables, to increase the sample size. However, they are incorporated in the robustness analyses. 16

measures and methodologies can always be questioned. However, this arrangement mostly replicated the methods of prior studies to determine the relationship between corporate governance and the value of excess cash, combined with the fixed-effect analysis, to reduce the bias caused by omitted variables. This should still provide reliable evidence. [Insert Table 1 about here] Table 1 shows that the average for the measure of directors' liability insurance indicates that roughly 53% of the observations are protected by this insurance. Not reported in tables, the numbers of insured companies in each year are 563, 628, 661 and 719 in 2008, 2009, 2010 and 2011, respectively. This evidence suggests that the proportion of companies with directors' liability insurance in Taiwan is relatively smaller (about 80%) than in western nations, including the U.K. and Canada (see, for example, O'Sullivan (2002) and Boyer (2005)), but the demand for this insurance has increased in recent years, which implies that carrying directors' liability insurance is a trend and an option to protect shareholder wealth. Table 1 shows that the characteristics of listed companies in Taiwan during the sample period, but the summary statistics of these variables, while lacking comparable data, are within the normal range. The value of excess cash proxied by the market to book ratio has a mean (median) of 1.388 (1.065), which may indicate that the market 17

value of this sample in a period surrounding a financial crisis is only slightly higher than the book value, but some observed companies have relatively high stock prices. The excess cash, measured in natural log form, has a mean (median) of -0.027 (0.162). The divergence between the mean and the median could be again driven by the market instability, which could impact any prediction of optimal cash holdings at the beginning of the sample period. Finally, the other variables suggest that the observed companies, in general, are profitable, even though some of them suffer a negative net income, scaled by total assets. R&D expenditure, interest expenses and dividend payout are on average only about 2.5%, 0.5% and 3% of total assets, respectively, with 95 percentiles around 10% for both R&D expenditure and dividend payout and about 1.5% for interest expenses. The last column of the Table 1 also shows the correlation between the value of excess cash and the other individual variables. The correlation of 0.05 between the value of excess cash and the dummy indicator of directors liability insurance, with a significance level of 1%, shows that the protection provided by directors liability insurance increases the value of excess cash. While this evidence ignores the effect of other factors on the value of excess cash and this relationship possibly means that companies with this insurance tend to have a greater market to book ratio. This correlation gives a visual representation of the potential interaction between directors liability insurance and the value of excess cash. With the exception of the important correlation, the correlation values for the other variables in Table 1, while all below 0.2, 18

show these variables are all significantly related to the value of excess cash (market to book ratio), which not only shows the value of certain company characteristics, but also the necessity that these variables are incorporated as control variables. However, caution must be exercised when interpreting these correlation values, as the hidden interaction between and the impact of omitted variables are neglected. The following sections give more thorough analyses, which provide convincing evidence for the hypotheses. 5. Results In this section, to test whether the link between directors liability insurance and the value of excess cash remains positive, as shown in Table 1, the comparative statistics are first used to numerically show the difference in the value of excess cash in companies with and without the protection of directors liability insurance. As shown in Panel A of Table 2, the value of excess cash in the insured companies on average (median) is 0.290 (0.104) higher than that in uninsured companies, when all observations are used. The differences in both the mean and median are significant at the 1% level. While the average (median) value of excess cash is about 1.388 (1.065), these differences are not trivial and demonstrate the potential benefits of carrying directors liability insurance. [Insert Table 2 about here] 19

In Panels B and C, whether this linkage remains similar in a subsample, in which the levels of excess cash differ is determined. The values in Panel B suggest that, for observed companies with more excess cash than the sample median (0.162), the value of excess cash in the insured companies, all else being equal, is 0.391 (0.142) higher than that in uninsured companies. Likewise, the results in Panel C show that, even in the group with a value of excess cash that is less than the sample mean or median, companies with directors liability insurance continuously exhibit a greater value of excess cash. While the value of directors liability insurance is again confirmed, the values in Panels B and C show that the effect of having directors liability insurance on the value of excess cash is greater in companies with more excess cash, which shows that acquiring this insurance does not necessarily enhance the market to book ratio, but it does increase the value of excess cash holdings 8. [Insert Table 3 about here] Following the univariate tests, multivariate regression analyses are used to find the pure effect of directors liability insurance on the value of excess cash holdings. Specifically, fixed-effect analyses are used to run the regression, because except for the 8 The numbers of observations in each subgroup suggest that the companies with more excess cash are likely to carry directors liability insurance and so the positive effect of this insurance on the value of excess cash could simply be the by-product of the value of holding excess cash. As the comparative statistics results in Panels B and C in Table 1 only partially reject this possibility, in the robustness tests, this concern is addressed by firstly regressing directors liability insurance on excess cash and all other independent variables applied in the main tests, to obtain the residual of directors liability insurance. The regression (3) in the Table 3 is then performed again with the original measure of directors liability insurance replaced by its residual. Consistently, a positive and significant coefficient for the interactive term is observed. 20

characteristics of the panel data used in this paper, it could reduce the bias from omitting firm-specific factors 9. For the variables already mentioned in the previous section, the dependent variable is the value of excess cash proxied by the market to book ratio (Frésard and Salva (2010)), the primary independent variables are the measures of directors liability insurance, excess cash holdings and the interactive term for these variables. While it is expected that carrying directors liability insurance to some extent alters the function of the board or the quality of corporate governance and affects the use or the value of excess cash holdings, the size and significance level of the coefficient of the interactive term is key. For instance, a significant (insignificant) coefficient for the interactive term indicates that the value of excess cash is different (the same) in companies with and without directors liability insurance. As shown in Table 3, three equations are used to demonstrate whether directors liability insurance affects the value of excess cash holdings and whether this link is sustained as additional variables are considered. Particularly, in regression (1), only the dummy indicator of directors liability insurance, the excess cash and year dummies are included as independent variables. This arrangement shows that, all else being equal, the individual effects of the excess cash holdings and directors liability insurance on the value of excess cash. The coefficients for the excess cash and the directors liability 9 Alternatively, Fich and Shivdasani (2006), in their empirical work examining the linkage between busy outside directors and company value, document that Given the high correlation between the marketto-book ratio and corporate governance variables with numerous other company attributes, we view the fixed effects framework as offering significantly more reliable estimates than OLS regressions As results from the LM test and the Hausman test indicate that fixed-effect analysis could be more suitable for the data than either random-effect analysis or OLS, the empirical tests mainly rely on the fixed-effect analysis. 21

insurance are all positive, but only the one for the excess cash reaches the significance level, which implies that the protection provided by directors liability insurance itself is not necessarily associated with a greater market value used to proxy the value of excess cash. While this evidence suggests that this insurance is neither good nor bad for shareholders, this phenomenon should be intuitive, as it is difficult to believe that companies could simply raise their value by acquiring directors liability insurance. However, it is also worthy of note that regression (1) may overlook the interaction between this insurance and the amount of excess cash holdings. In other words, directors liability insurance may cause the companies to use their excess cash, so altering the value of excess cash. Therefore, regression (2) in Table 3 is used to determine the potential interaction between directors liability insurance and the excess cash holdings. The coefficient of the interactive term, directors liability insurance * excess cash, is 0.602, which is not only greater than the coefficients of both directors liability insurance and excess cash in regression (1) but also significant at the 1% level. Combined with the evidence in regression (1) and the summary statistics in Table 1, the excess cash held by companies remains valuable, but companies with directors liability insurance enjoy even greater value from holding excess cash. In particular, as the median of the value of excess cash is 1.065, companies with the median amount of excess cash obtain roughly additional 9.75% (0.162*0.602) of the value of the excess cash if they are insured. In regression (3), except for the key variables, several control variables in line with 22

Frésard and Salva (2010) are incorporated into the regression to further verify whether similar results to those for regression (2) are observed and whether the effect of directors liability insurance on the value of excess cash is driven by potential variables. However, the coefficients of directors liability insurance*excess cash and excess cash are both positive and significant, although the values of the coefficients and their t- values are slightly different. This evidence again confirms the benefits of carrying directors liability insurance. As Chalmers, Dann and Harford (2002), Zou, Wong, Shum, Xiong and Yan (2008), and Lin, Officer and Zou (2011) among others indicate, this insurance may induce opportunistic behavior by managers or increase the moral hazard problem, but the empirical findings suggest that, in Taiwan, the increase in value due to carrying directors liability insurance outweighs its costs, even though carrying this insurance in a market where voting rights in general exceed cash-flow rights is likely to cause severe agency conflicts. Finally, in regression (3) of Table 3, the coefficients of the profitability, R&D expenditure and dividend are significant. The negative link between the value of excess cash and profitability proxied by the sum of net income, all noncash charges, extraordinary items and interest scaled by total assets, and between the value of excess cash and the R&D expenditure, implies, from the shareholders point of view, that it is better for companies to reduce cash holdings and invest them prudently, as doing so will generate profits 10. A positive link between the value of excess cash and the dividend, 10 While it could also be true that making profits or spending in R&D would signal the current or future company value and the market value, which is used to calculate the value of excess cash, is expected to increase. However, as the focus is on the impact of directors liability insurance on the value of excess 23

on the other hand, may indicate that dividend payouts reduce the misuse of free cash and improve the value of excessive cash holdings. Similarly, a positive but insignificant coefficient for the interest expenses gives partial supporting evidence to the notion that increasing financial leverage improves external monitoring and so decreases the likelihood for managerial expropriation. In the end, the negative coefficient for the net assets is caused by the measure of the value of excess cash, in which the total assets is the denominator. While comparable evidence from samples in Taiwan is lacking, the coefficients of the control variables in all regressions, except in subsamples restricted by limited observations, are at least similar in both size and the level of significance. 6. Robustness tests The empirical findings documented in the previous section show that carrying directors liability insurance is associated with a greater value of excess cash, which implies that, in Taiwan, this insurance may play a role in improving the function of the board, offer alternative external monitoring to prevent corporate wrongdoings, or provide expertise to lower the costs suffered by the insured companies during corporate lawsuits. In this section, alternative measures or subsamples are used to confirm the findings that the value of excess cash in companies insured by directors liability insurance is higher than that in uninsured companies. In Table 4, the measure of excess cash is replaced by the natural log of cash and cash and empirical findings could be sensitive to sample selected, this debate is left for future research. 24

marketable securities. This measure may be too naïve, since it ignores company characteristics that influence a company s preference in holding excess cash, but it could represent the information utilized by most investors to evaluate a company. As shown in regression (1), while the coefficient of cash holdings remains positive but loses its significance, the coefficient of directors liability insurance*cash holdings is still positive and significant at the 5% level, which shows that the market does value the cash holdings, particularly in companies with directors liability insurance. [Insert Table 4 about here] In regression (2) of Table 4, keeping all variables, the lead change (from year t to year t+2) in profitability, R&D expenditure, interest expenses, dividend, and net assets is incorporated, to better explain the dependent variable. However, this arrangement significantly reduces the sample size, because observations in 2011 and 2012 must be excluded. With less than 50% of the raw data, either the signs or the levels of significance of some coefficients of the independent variables are changed, but the coefficient of directors liability insurance*excess cash is continuously positive and significant. However, it is worthy of note that the coefficient of directors liability insurance becomes negative and significant, which indicates that the observed companies with the protection of directors liability insurance have a poor market to book ratio and excessive cash in these companies remains more valuable than that in 25

uninsured companies. Indeed, some studies, as shown in the previous section, reveal that carrying directors liability insurance is costly during certain corporate events and the negative coefficient of directors liability insurance in regression (2) may support this notion. As limited observations are used and carrying directors liability insurance may influence the insured companies through different dimensions, the impact of directors liability insurance on the value of excess cash holdings remains the focus in this paper. In Table 5, the abnormal value of directors liability insurance and the lag of directors liability insurance replace the original measure of directors liability insurance. In regression (1), directors liability insurance is predicted, using the equation below, in which total assets and financial leverage, calculated as the total debts scaled by total assets, are used to determine company characteristics. Managerial and director ownership, measured by the percentages of shares held, are used to determine the severity of conflicts of interest between shareholders and agents, which could influence the effect of directors liability insurance on the insured companies. The standard deviation of daily stock returns is primarily used to proxy the company risk Directors liability insurance it = α 0 + β 1 ln(total assets it ) + β 2 (leverage it ) + β 3 (managerial ownerships it ) + β 4 (director ownership it ) + β 5 (standard deviation of daily stock returns it ) + ε it The value of the predicted directors liability insurance is obtained through fixed- 26

effect analysis and the abnormal directors liability insurance is the difference between the actual directors liability insurance and the predicted value. As the use of the abnormal value of directors liability insurance represents the case where the carrying of this insurance is likely to be endogenous, the selection of variables in the process to estimate the abnormal value is highly subjective, due to the scarcity of relevant literature. While the accuracy of the prediction is not guaranteed, prior studies are considered when examining the demand of this insurance, to remove the effect of factors that influence companies decision processes in acquiring directors liability insurance, to determine the pure effect of carrying this insurance on the value of excess cash. Consistently, the coefficient for the interactive term in regression (1) again remains positive, but the level of significance is greater than that in regression (3) in Table 3. In addition, this positively significant coefficient is always seen, even if alternative variables are used to predict the residual of the directors liability insurance, which suggests that concerns about endogeneity may not necessarily be critical, particularly for the sample in Taiwan. In other words, the reasons for companies to purchase directors liability insurance do not alter the positive effect of carrying this insurance on the value of holding excessive cash. [Insert Table 5 about here] In regression (2), the lag of directors liability insurance or the dummy indicator 27

of this insurance in the year t-1 replaces the original variable. Although this approach reduces the number of observations, it is widely used in empirical studies to determine whether the effect of carrying directors liability insurance on the value of holding excessive cash is either slow or postponed. While the size of the coefficient of directors liability insurance*excess cash is greater, the significance of this coefficient is still at the 1% level, which implies that there is a link between the carrying of directors liability insurance and the value of holding excess cash and this effect is constant. Furthermore, it is less likely that higher market valuation induces companies to acquire directors liability insurance. Other than using alternative variables to determine the robustness of these findings, in Table 6, attention is focused on sample selection. Specifically, in regression (1), are the observations for changing from with (without) directors liability insurance to without (with) this insurance. This approach addresses the concern that during the transition period, these observations underestimate the relationship between the directors liability insurance and the value of excess cash. When 141 observations are eliminated, the coefficient of the interactive term consistently remains positive, but greater in both size and the level of significance than that in regression (3) in Table 3 11. [Insert Table 6 about here] 11 Not reported in tables, observations in the year t-1 of these companies changing from with (without) directors liability insurance to without (with) this insurance are dropped. The coefficient of directors liability insurance*excess cash is 0.936 and remains significant at the 1% level. 28

In regression (2) in Table 6, only electronics companies in Taiwan are considered. This restriction limits the sample to observations representing the majority of listed companies either in quantities or in market values 12. As this selection is widely used in study samples in Taiwan, the empirical evidence with 2,807 firm-year observations for 755 companies surprisingly shows an even stronger relationship between the carrying of directors liability insurance and the value of holding excess cash. In the last regression shown in Table 6, in order to reduce the possible bias caused by the events of a financial crisis, firm-year observations in 2008 are eliminated. The coefficient of the interactive term is once again similar to the initial findings 13. Overall, the use of alternative measures or subsamples seems not to distort the results. The benefits, such as external monitoring or improvement of board effectiveness, brought by the protection of directors liability insurance, tend to outweigh the costs, such as moral hazard problems. Finally, not reported in tables, several tests are also used to determine the relationship between directors liability insurance and the value of excess cash. For instance, as the benefits or costs of this insurance to some extent influence the quality of corporate governance, it is possible that the value of directors liability insurance observed in the empirical findings would not be so pronounced if the observed companies had a good quality of corporate governance. To address this issue, the 12 The identification of electronic companies in Taiwan is based on the industry classification codes (from 24 to 31) of the Taiwan Stock Exchange (TWSE). 13 A positive coefficient of the interactive term remains positive if observations in both 2008 and 2009 are excluded. 29

regressions in the subsample, in which the percentages of independent directors are greater than the sample median and in the subsample in which the percentages of shares held by board members are greater than the sample median, are performed again. The positive and significant effect of directors liability insurance on the value of holding excessive cash remains unchanged. Except for the dummy indicator that differentiates between companies with and without directors liability insurance, the coverage of this insurance, measured as a natural log, is used to determine whether more protection is associated with a greater value of excess cash. While some observations lacking the information about the coverage are missing, the coefficient of directors liability insurance*excess cash is significantly positive and in line with the prediction that the degree of protection proxied by the amount of coverage represents the degree of the impact of directors liability insurance on the value of excess cash. However, it is also worthy of note that using the measure of insurance coverage requires caution, since it may overlook the level of risk in each observed company. The insurance premium may be applicable to proxy company risk (Core (2000)) and the consequent behavior in the use of capital funds, but unfortunately insurance premiums in Taiwan are currently not publicly available 14. Finally, the observed companies change from uninsured (insured) to insured (uninsured) in 2009 and 2010 are removed from the sample, to determine whether these 14 The natural log of the amount of total insurance coverage is used as the variable to replace directors liability insurance in regression (3) in Table 3, The coefficient of the interactive term with the alternative measure is 0.048 and is continuously significant at the 1% level. 30