Microsoft Word - 10095-37142-1-SM-writer2-new Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 95 The Influence of Directors’ and Officers’ Insurance on Managerial Myopic Behavior Guang-Zheng Chen Department of Accounting, Feng Chia University No.100, Wenhwa Road, Seatwen, Taichung 40724, Taiwan, R.O.C. E-mail: gzchen0120@gmail.com Received: Oct. 1, 2016 Accepted: Oct. 20, 2016 Published: December 1, 2016 doi:10.5296/ajfa.v8i2.10095 URL: http://dx.doi.org/10.5296/ajfa.v8i2.10095 Abstract Literature on the effect of directors’ and officers’ insurance (D&O insurance) on managers’ decision-making supports the contention that D&O insurance encourages managers to engage in opportunistic behaviors that benefit themselves at the expense of shareholders. Managerial myopia is an essential agency issue. The literature suggests that myopic managers have incentives to reduce R&D spending to boost current earnings in order to increase their private benefits. This study examines whether D&O insurance induces myopic R&D cuts. Using a sample of Taiwanese listed firms, the results show that firms with higher levels of D&O insurance coverage are more likely to cut R&D expenditures to avoid earnings declines. This study provide insight into how the incentives arising from D&O insurance play an essential role in determining managerial myopic behavior. Keywords: Corporate Governance, D&O Insurance, R&D, Managerial Myopia, Agency Theory Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 96 1. Introduction Directors’ and officers’ insurance (D&O insurance) alleviates the financial liability of officers and directors stemming from lawsuits filed against them while working for the company. However, empirical studies provide evidence that D&O insurance is likely to motivate managers to engage in opportunistic behavior (Chalmers et al., 2002; Chung and Wynn, 2008; Lin et al., 2011). Managerial myopia is an agency problem that occurs when managers sacrifice long-term value creation projects (e.g., R&D) for the purpose of meeting short-term goals (Porter, 1992). Research (Graham et al., 2005; Roychowdhury, 2006; Osma and Young, 2009) also suggests that managers opportunistically reduce R&D spending to boost current earnings for private benefits. Therefore, I examine whether D&O insurance provide managers with incentives to behave myopically by cutting R&D spending. D&O insurance policy typically provides an essential shield for directors and officers against personal legal liabilities stemming from their business decisions. Nevertheless, D&O insurance can induce moral hazard problems, thus reducing the incentive of managers to act in the best interest of stakeholders (Baker and Griffith, 2010; Lin et al., 2011, 2013). Empirical studies of managerial myopic behavior have focused primarily on R&D expenditure. The evidence is consistent with managers myopically reducing R&D investment to meet short-term earnings goals (Baber et al., 1991; Dechow and Sloan, 1991; Bange and De Bondt, 1998; Bens et al., 2002; Roychowdhury, 2006; Cohen and Zarowin, 2010). Such managerial myopia occurs because R&D investments are immediately expensed under current accounting rules, yet may pay off in the long-term rather than in the short term (Fama and Jensen, 1983; Baysinger et al., 1991). Research indicates that two circumstances are required for myopic investment behavior (Stein, 1988, 1989; Bushee, 1998). First, managers should place greater emphasis on the current market value of their firm. Second, capital markets should misprice current earnings without fully considering their underlying economics, or managers should believe that they do. Since D&O insurance leads managers to act in their own interest rather than the firm’s interest, well-protected managers may have incentives to engage in myopic R&D investment behavior. R&D activities have a long-term relationship with highly uncertain, unpredictable future cash flows, as well as a high risk of failure (Holmstrom, 1989). Since D&O insurance induces managers to take action in pursuit of personal goals, they may be reluctant to engage in R&D activities and more likely to reduce the amount of R&D investment to achieve current-period earnings performance. Research (Fuller and Jensen, 2002; Rappaport, 2005) also provides evidence that capital markets can incorrectly price current earnings, and investors and managers have a mutually reinforcing obsession with short-term performance, with earnings as the most widely accepted metric.1 Since D&O insurance induces managers to pursue opportunistic behavior and damage firm value, well-protected managers may be inclined to reduce R&D expenditures in order to report strong earnings, which in turn, drives up the firm’s stock price in the short- run and increases their compensation. I therefore expect that D&O insurance coverage is associated with myopic R&D investment behavior. 1 It is also consistent with Stein’s (1988, 1989) observation that the capital market is myopic and will induce managers to behave myopically. Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 97 Based on a sample of Taiwanese listed firms over the period 2008-2012, I find that, consistent with my prediction, firms with higher D&O insurance coverage are more likely to cut R&D expenditures when faced with potential earnings declines. This finding suggests that D&O insurance induces managers to make corporate decisions which advance their own interests. The results are robust to several sensitivity checks. This study contributes to the literature in several ways. First, research has investigated the effect of D&O insurance on the outcome of mergers and acquisitions (M&As) (Lin et al., 2011) and financial reporting quality (Chung and Wynn, 2008; Chung et a., 2013; Kim, 2015). This study adds to the literature on D&O insurance by showing that moral hazards related to D&O insurance can affect myopic R&D cuts. Second, this study contributes to the literature on managerial myopia. A large number of studies document that managers are inclined to sacrifice long-term value creation to achieve short-term earnings targets. These studies suggest that managers tend to reduce R&D expenditures in response to concerns about earnings declines or losses (Baber et al., 1991), short managerial horizons (Dechow and Sloan, 1991), catering to the short-term needs of transient institutional investors (Bushee, 1998), and the need to raise capital (Cohen and Zarowin, 2010). This study extends this line of research by provide evidence that D&O insurance coverage appears to be an important determinant of managerial myopia. Finally, this study adds to the stream of real earnings management literature since myopic R&D reduction is a type of real activities manipulation. The reminder of this paper is organized as follows. Section 2 reviews relevant literature and develop the hypothesis. Section 3 outlines the research design. Section 4 details the sample selection and presents the empirical findings. Section 5 concludes. 2. Literature Review and Hypothesis Development 2.1 Research on D&O Insurance D&O insurance is designed to protect directors and officers from liability stemming from actions connected to their corporate position. D&O insurance covers situations in which the director or officer commits fraudulent or illegal activities unintentionally, but does not violate his/her duty to the shareholders and the firm. The insurance policy typically does not cover the total liability, and management should bear part of the cost personally. The items generally covered contain judgments or settlements in actions alleging negligence. The policy does not cover liability including willful misconduct, self-dealing, bad faith, knowing violation of security laws, personal profit, or dishonesty. Holderness (1990) and O’Sullivan (1997) argue that D&O insurance acts as a monitor mechanism for directors and officers because the D&O insurer entirely scrutinizes the insured, and coverage limits and deductibles exist. In addition, Kalelkar and Nwaeze (2015) suggest that abnormally higher D&O insurance coverage can more fully insulate managers against legal penalties stemming from exercising their corporate decision-making authority. Therefore, higher abnormal coverage induces mangers to take actions that will maximize shareholder value. However, a considerable body of research suggests that D&O insurance alleviates the expected litigation risk of managers, thus introducing severe agency problems Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 98 and managerial opportunism. For instance, Zou et al. (2008) indicate that D&O insurance could be opportunistically purchased to protect corporate insiders (managers or controlling shareholders) against potential lawsuit costs stemming from the expropriation of outside shareholders. Chalmers et al. (2002) document a negative association between the amount of D&O insurance purchased at the time of the IPO (initial public offering) and three-year post-IPO stock returns. Their finding suggests that managers are prone to purchase D&O insurance when they are aware of overvalued IPO stocks (and thus the litigation risks caused by subsequent price decreases). In addition, several studies document a significantly positive association between D&O coverage limits and the likelihood that firms restate earnings (Lin et al., 2013; Kim, 2015). Similarly, D&O insurance coverage limits have also been shown to be negatively associated with earnings quality (Chung et al., 2013) and earnings conservatism (Chung and Wynn, 2008). Moreover, higher D&O coverage is pertinent to firms that have lower abnormal-period returns during mergers and acquisitions (Lin et al., 2011), higher costs of debt and equity capital (Lin et al., 2013; Chen et al., 2016), and higher audit fees (Chung et al., 2015). 2.2 Research on Managerial Myopia Myopia refers to underinvestment in long-term value creating activities for the purpose of meeting short-term goals (Porter, 1992). Research provides evidence of managerial myopia, mainly with respect to R&D expenditures. For instance, Baber et al. (1991) demonstrate that when spending on R&D may reduce the ability of managers to report positive income or an increase in income, they decrease R&D expenditures to report stronger earnings. Dechow and Sloan (1991) show that CEOs will reduce R&D expenditures to increase short-term earnings in the final years of their tenure. Bens et al. (2002) find that managers cut R&D expenditure to fund share repurchase programs in order to mitigate earnings per share dilution due to stock option exercise. In addition, Graham et al. (2005) survey and interview over 400 senior executives and report that 80% would cut R&D as well as other discretionary expenditures to meet earnings benchmarks. Roychowdhury (2006) provides evidence consistent with the contention that managers manipulate real activities, such as R&D, to avoid reporting annual losses. Osma and Young (2009) also document that the pressure to report positive levels and changes of earnings in a large sample of R&D-active UK firms results in contemporaneous cuts in R&D spending. More recently, Chen et al. (2015) report evidence that firms with severance pay or fixed employment agreements for managers can reduce their myopic behavior of cutting R&D expenditures to boost short term performance in order to increase their job security. As mentioned above, such managerial myopia can be attributed to the current accounting rules forcing firms to expense R&D in the period incurred. Namely, the increase of R&D expenditures has a negative effect on short-term accounting earnings and stock performance, while benefits from the expenditures occur in the future.2 In the presence of such negative effects, firms have incentives to reduce R&D expenditures to boost current earnings, which results in conflicts of interest between managers and outside shareholders. 2.3 D&O Insurance and Managerial Myopia 2 The literature indicates that R&D projects are typically risky, unpredictable, long-term oriented, multi-stage, labor intensive and idiosyncratic (e.g., Holmstrom, 1989; Baysinger et al., 1991; Kothari et al., 2002). Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 99 Myopic managerial behavior typically occur under two circumstances (Stein, 1988, 1989; Bushee, 1998). First, managers should place greater emphasis on current market value relative to future market value. They are reluctant to wait until the temporary mispricing (if any) based on short-term earnings performance is corrected and thus reluctant to adopt a long-term perspective. Second, capital markets should misprice firms’ current-period earnings without regard to their underlying economics, or managers must believe that they do. Since D&O insurance results in a host of moral hazard issues and reduces managers’ incentives to act in the best interest of shareholders (Baker and Griffith, 2010; Lin et al., 2011), well-protected managers may have incentives to cut R&D to increase short-term earnings rather than engage in long-term value creation. First, R&D investments are long-term, risky, and expensive projects. Since D&O insurance induces managers to act opportunistically, they may focus on short-term earnings performance and thus underinvest in R&D. Second, capital markets can incorrectly price current earnings if investors are short-term oriented (Ellis, 2004) or misinterpret the persistence of earnings components (Sloan, 1996). Research suggests that investors and managers have a mutually reinforcing obsession with short-term performance, and such short-term focus motivates managers to actively participate in the earnings game (Fuller and Jensen, 2002; Rappaport, 2005).3 As well-protected managers are inclined to make decisions that primarily generate private benefits for themselves rather than for shareholders, they may underinvest in R&D to increase current earnings. This boosts their firms’ stock price and maximizes their compensation. Consequently, D&O insurance leads managers to act in their own interest at the expense of shareholders, and well-protected managers may be reluctant to pursue risky R&D activities and inclined to cut R&D to achieve near-term earnings performance. This leads to the following hypothesis: Hypothesis 1: Ceteris paribus, firms with higher D&O insurance coverage are more likely to cut R&D to manage earnings. 3. Research Design To capture managerial myopic behaviors, I follow Bushee (1998) and partition sample firms into three groups: a small earnings decrease (SD) group, a large earnings decrease (LD) group, and an earnings increase (IN) group. The SD group includes firm-years which have a decrease in the pre-tax, pre-R&D earnings from the previous year to the current year and in which the decrease is less than the previous year’s R&D. Baber et al. (1991) and Bushee (1998) suggest that within this group, myopic managers can potentially avoid an earnings decrease by cutting the current year R&D. The LD group includes firm-years in which there is a decrease in the pre-tax, pre-R&D earnings from the previous year to the current year and in which the decrease is greater than the previous year’s R&D. The IN group includes firm-years which have an increase in the pre-tax, pre-R&D earnings from the previous year to the current year. For LD and IN groups, cutting R&D is not helpful or necessary to achieve short-term earnings increase (Chen et al., 2015). 3 Even when the capital markets are efficient, managerial myopic behavior can result so long as managers believe that markets can be fooled (Stein, 1989), which they do (Graham et al., 2005). Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 100 I run the following logit regression to test the effect of D&O insurance coverage on R&D cuts. 0 1 2 3 4 5 6 7 8 9 10 11 12 Prob( 1) ( Ln ) t t t t t t t t t t t t n m t CUT f DOLIMIT PCRD CIRD CGDP TOQ CSALES CCAP MV LEV FCF DIST INSOWN YEAR IND α β β β β β β β β β β β β β β ε = = + + + + + + + + + + + + + + + (1) where CUT is a dummy variable equal to 1 if R&D decreases relative to the previous year, 0 otherwise. DOLIMIT is D&O coverage limit scaled by market value of equity. PCRD is the difference in the natural logarithm of R&D between the previous year and the year before. CIRD is the difference in industry R&D intensity (R&D expenditures scaled by total sales) between the current year and the previous year, where the industry is defined using the two-digit Taiwan Economic Journal (TEJ) code. CGDP is the difference in the natural logarithm of GDP between the current year and the previous year. TOQ is the sum of market value of common equity and book value of debt, divided by the book value of total assets. CSALES is the difference in the natural logarithm of sales between the current year and the previous year. CCAP is the difference in the logarithm of capital expenditures between the current year and the previous year. LnMV is the natural logarithm of market value of equity. LEV is total debts divided by total assets. FCF is operating cash flows minus capital expenditures, scaled by total assets. DIST is the change in pre-tax, pre-R&D earnings divided by previous year’s R&D. INSOWN is the percentage of shares held by institutional investors. In equation (1), the primary variable of interest is the coefficient on DOLIMIT. If well-protected managers are prone to cut R&D expenditures to manage earnings upwards, I would expect a positive and significant coefficient on DOLIMIT in the small earnings decrease (SD) group. By contrast, I expect that the coefficients on DOLIMIT would not be significant for the large earnings decrease (LD) and earnings increase (IN) groups. Following Bushee (1998), I include several control variables that influence R&D investments and the likelihood of cutting R&D. First, I control for a firm’s R&D investment opportunity set by using following proxies: (1) last year’s change in R&D (PCRD) that captures the trend in R&D investments; (2) the change in industry R&D intensity (CIRD) that captures the R&D investment opportunity in the industry; (3) the change in GDP (CGDP) that captures the economy level of investment opportunity; and (4) Tobin’s Q (TOQ), change in sales (CSALES), and change in capital expenditures (CCAP) that capture the firm’s growth opportunities. Each of the above variables is expected to be negatively related to the likelihood of an R&D cut. Second, firm size (SIZE) is controlled for because smaller firms are more likely to suffer cash flow shortages that lead them to cut R&D. Third, leverage (LEV) is included in the model to capture the firm’s incentives to increase earnings to reduce debt contracting costs. Fourth, free cash flows (FCF) is included to control for fund availability because firms have higher incentives to cut R&D when facing a serious cash shortage. Fifth, distance to earnings goal (DIST) is controlled for because larger distance increases the likelihood of an R&D cut to meet earnings goals. Fifth, institutional ownership (INSOWN) captures the monitoring by institutional investors and firms are less likely to engage in Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 101 earnings management through R&D reduction. Lastly, I include year dummies (YEAR) and industry dummies (IND) in the regression to control year and industry fixed effects. 4. Data Sample and Empirical Results 4.1 Sample The initial sample includes nonfinancial firms listed in the Taiwan Stock Exchange (TSE) over the period 2008-2012.4 The data concerning D&O insurance coverage, ownership structure, financial statement and stock price are obtained from Taiwan Economics Journal (TEJ) database.5 Table 1 outlines the sample selection. I start with an initial sample of 7,143 firm-years for the empirical analysis. I delete 969 firm-years due to the lack of D&O insurance data. I drop 2,730 firm-years with missing or insignificant R&D expenditures (R&D expenditures less than 1% of sales). I exclude another 136 firm-years because they do not have sufficient data to calculate the regression variable. The final sample consists of 3,308 firm-year observations; 684 for SD group, 1,067 for LD group, and 1,557 for IN group. Table 1. Sample Selection Firm-years Non-financial firms listed on the Taiwan Stock Exchange from 2008 to 2012 7,143 Less: firms with missing D&O insurance data 969 Firms with missing R&D data in the current year 72 Firms with missing or insignificant R&D in the previous year 2,658 Firms without necessary data to calculate regression variables 136 Final sample 3,308 Small earnings decrease (SD) group 684 Large earnings decrease (LD) group 1,067 Earnings increase (IN) group 1,557 The SD group comprises firm-years for which –RDEXPt-1 < (EBTRDt-EBTRDt-1) < 0. The LD group comprises firm-years for which (EBTRDt-EBTRDt-1) < -RDEXPt-1. The IN group comprises firm-years for which (EBTRDt-EBTRDt-1) > 0. RDEXP = R&D expenditures. EBTRD = pre-tax, pre-R&D earnings. 4.2 Descriptive Statistics Table 2 presents descriptive statistics for the sample (Panel A) and the results of univariate tests that statistically assess the comparisons between SD group, LD group, and IN group (Panel B). To avoid the effect of outliers, all continuous variables are winsorized at the 1% and 99% tails. In Panel A, it is about 47% of the sample firm-years cut R&D expenditures. Overall, the average (median) D&O coverage limit (Coverage_limit) is $170.989 million ($87.374 million), and the mean (median) D&O coverage limit (DOLIMIT) accounts for 6.9% (2.0%) of market value of equity. In Panel B, for almost all of the variables, the mean for the SD group lies between those of the LD and IN groups, consistent with previous research (Bushee, 1998) that is concerned with the performance ranking of this three groups. In addition, the means difference of CUT and DOLIMIT between SD group and IN group are statistically significant at the 1% and 5% levels. This suggests that firms in SD group with 4 I start with the year 2008 because it is the first year with public disclosure of D&O insurance coverage. 5 The data for GDP is retrieved from National Statistics, R.O.C. (Taiwan) website. Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 102 higher likelihood of cutting R&D and higher D&O coverage. However, there is no significant difference between SD group and LD group in CUT and DOLIMIT. Table 2. Descriptive Statistics Panel A: Full sample Mean Standard deviation Q1 Median Q3 CUT 0.468 0.499 0.000 0.000 1.000 Coverage_limit 170.989 362.660 0.000 87.374 194.985 DOLIMIT 0.069 0.140 0.000 0.020 0.072 PCRD 0.056 0.288 -0.093 0.045 0.189 CIRD 0.000 0.002 0.000 0.001 0.001 CGDP 0.019 0.038 -0.014 0.014 0.026 TOQ 1.426 0.767 0.916 1.202 1.687 CSALES -0.015 0.303 -0.167 -0.009 0.142 CCAP -0.078 1.214 -0.760 -0.080 0.585 LnMV 14.841 1.405 13.861 14.713 15.695 LEV 0.323 0.156 0.201 0.309 0.423 FCF 0.037 0.100 -0.014 0.037 0.094 DIST 0.018 5.033 -1.614 -0.121 1.272 INSOWN 0.329 0.207 0.163 0.296 0.468 Panel B: Separately for the small earnings decrease (SD), large earnings decrease (LD), and earnings increase (IN) groups Mean Differences in Means SD group (n = 684) LD group (n = 1,067) IN group (n = 1,557) SD versus LD SD versus IN CUT 0.518 0.540 0.397 -0.022 0.121*** DOLIMIT 0.073 0.078 0.061 -0.004 0.013** PCRD 0.095 0.065 0.033 0.030** 0.062*** CIRD 0.001 0.001 0.000 -0.000*** 0.001*** CGDP 0.015 0.009 0.028 0.007*** -0.013*** TOQ 1.416 1.160 1.613 0.256*** -0.198*** CSALES -0.055 -0.166 0.107 0.111*** -0.162*** CCAP -0.160 -0.180 0.027 0.020 -0.187*** LnMV 14.753 14.628 15.025 0.125* -0.273*** LEV 0.296 0.329 0.331 -0.032*** -0.035*** FCF 0.041 0.026 0.042 0.015*** -0.001 DIST -0.451 -4.292 3.177 3.841*** -3.628*** INSOWN 0.325 0.317 0.338 0.008 -0.012 The SD group comprises firm-years for which –RDEXPt-1 < (EBTRDt-EBTRDt-1) < 0. The LD group comprises firm-years for which (EBTRDt-EBTRDt-1) < -RDEXPt-1. The IN group comprises firm-years for which (EBTRDt-EBTRDt-1) > 0. RDEXP = R&D expenditures. EBTRD = pre-tax, pre-R&D earnings. Variable definitions: CUT = 1 if R&D decreases relative to the previous year, and 0 otherwise; DOLIMIT = D&O coverage limit scaled by market value of equity; Coverage_limit is the D&O insurance coverage limit measured in million dollars. PCRD = the difference in the natural logarithm of R&D between the previous year and the year before; CIRD = the difference in industry R&D intensity (R&D expenditures scaled by total sales) between the current year and the previous year, where the industry is defined using the two-digit Taiwan Economic Journal (TEJ) code; CGDP = the difference in the natural logarithm of GDP between the current year and the previous year; TOQ = the sum of market value of common equity and book value of debt, divided by the book value of total assets; CSALES = the difference in the natural logarithm of sales between the current year and the previous year; CCAP = the difference in the logarithm of capital expenditures between the current year and the previous year; LnMV = the natural logarithm of market value of equity; LEV = total debts divided by total assets; FCF = operating cash flows minus capital expenditures, scaled by total assets; DIST = the change in pre-tax, pre-R&D earnings divided by previous year’s R&D; INSOWN = the percentage of shares held by institutional investors. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively (two-tailed tests). Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 103 4.3 Empirical Results Table 3 presents logit regression results, first for SD group (Model 1) and then for LD group (Model 2) and IN group (Model 3). As shown in Model 1, the coefficient on DOLIMIT is positive and significant at the 5% level, consistent with the hypothesis that firms with higher D&O coverage are more likely to cut R&D to avoid earnings decreases. In addition, the marginal effect for DOLIMIT in Model 1, 0.308, when multiplied by the interquartile range of 0.091, suggests that moving from the first to the third quartile of DOLIMIT increases the probability of cutting R&D by approximately 3%. Among control variables, CIRD and CCAP are significantly associated with the likelihood of cutting R&D in predicted directions. On the other hand, CGDP and DIS are positively and negatively associated with CUT, respectively, which is inconsistent with my predictions. In models 2 and 3, the coefficients on DOLIMIT are not significant at conventional levels. Consistent with my prediction, managers’ incentives to cut R&D to avoid earnings decrease is low or non-existent in the LD and IN groups, and thus D&O coverage is not expected to affect the likelihood of cutting R&D. Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 104 Table 3. D&O Coverage and the Likelihood of Cutting R&D Predicted signs Dependent variable: CUT Model 1 Model 2 Model 3 SD group LD group IN group Coefficient (z-statistic) Marginal effect Coefficient (z-statistic) Marginal effect Coefficient (z-statistic) Marginal effect DOLIMIT + 2.315** (2.43) 0.308 0.141 (0.29) 0.008 0.202 (0.40) 0.014 PCRD – -0.544 (-1.51) -0.137 -0.485** (-2.08) -0.121 -0.891*** (-3.67) -0.227 CIRD – -214.058** (-2.35) -36.881 -59.675 (-0.79) -7.516 -16.985 (-0.37) -21.940 CGDP – 14.451** (2.34) -1.182 5.735 (1.11) 0.305 10.241* (1.78) -1.894 TOQ – -0.234 (-1.53) -0.035 -0.246 (-1.60) -0.038 -0.339*** (-3.29) -0.039 CSALES – -0.830 (-1.43) -0.314 -1.873*** (-6.14) -0.485 -1.189*** (-4.70) -0.333 CCAP – -0.201** (-2.46) -0.053 -0.097* (-1.73) -0.030 -0.232*** (-4.49) -0.055 LnMV – -0.079 (-0.86) -0.010 -0.079 (-1.24) -0.013 -0.252*** (-4.30) -0.048 LEV + 0.289 (0.43) 0.063 0.889** (2.06) 0.196 -0.056 (-0.13) 0.017 FCF – 1.065 (1.03) -0.008 -0.774 (-1.01) -0.260 -2.061*** (-2.76) -0.570 DIST + -0.687** (-2.23) -0.140 -0.001 (-0.05) 0.000 0.006 (0.41) 0.003 INSOWN + -0.419 (-0.85) -0.140 -0.224 (-0.60) -0.020 -0.121 (-0.36) -0.079 Intercept 15.584*** (11.00) 0.986 (0.88) 5.412*** (4.25) Year / Industry dummy variables Included Included Included Pseudo R2 0.108 0.084 0.155 n 684 1,067 1,557 The SD group comprises firm-years for which –RDEXPt-1 < (EBTRDt-EBTRDt-1) < 0. The LD group comprises firm-years for which (EBTRDt-EBTRDt-1) < -RDEXPt-1. The IN group comprises firm-years for which (EBTRDt-EBTRDt-1) > 0. RDEXP = R&D expenditures. EBTRD = pre-tax, pre-R&D earnings. Variable definitions: CUT = 1 if R&D decreases relative to the previous year, and 0 otherwise; DOLIMIT = D&O coverage limit scaled by market value of equity; PCRD = the difference in the natural logarithm of R&D between the previous year and the year before; CIRD = the difference in industry R&D intensity (R&D expenditures scaled by total sales) between the current year and the previous year, where the industry is defined using the two-digit Taiwan Economic Journal (TEJ) code; CGDP = the difference in the natural logarithm of GDP between the current year and the previous year; TOQ = the sum of market value of common equity and book value of debt, divided by the book value of total assets; CSALES = the difference in the natural logarithm of sales between the current year and the previous year; CCAP = the difference in the logarithm of capital expenditures between the current year and the previous year; LnMV = the natural logarithm of market value of equity; LEV = total debts divided by total assets; FCF = operating cash flows minus capital expenditures, scaled by total assets; DIST = the change in pre-tax, pre-R&D earnings divided by previous year’s R&D; INSOWN = the percentage of shares held by institutional investors. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively (two-tailed tests). The z-statistics in parentheses are based on standard errors adjusted for clustering at the firm level. Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 105 4.4 Robustness Checks 4.4.1 Endogeneity To address the endogeneity issue for managerial myopic behaviors and D&O insurance purchase decision, I use a two-stage estimation technique developed by Heckman (1979). In the first stage, I estimate a probit model to obtain the inverse Mills ratio (IMR) by using variables that previous studies (e.g., Core, 1997, O’Sullivan, 2002, Chung and Whynn, 2008) have found to affect D&O purchase decision. In the second stage, I include IMR in equation (1) as an additional control variable to correct for potential endogeneity bias. The first-stage probit model is specified as follows. 0 1 2 3 4 5 6 7 8 9 10 Prob( 1) ( Ln ) t t t t t t t t t t t n m t PURCHASE f MV LEV MTB CEOOWN OUTOWN ACQU DIVE CROSS HITE EXCASH YEAR IND γ γ γ γ γ γ γ γ γ γ γ γ γ ε = = + + + + + + + + + + + + + (2) where PURCHASE is a dummy variable equal to 1 if the firm carries D&O insurance coverage, 0 otherwise. MTB is market-to-book ratio. CEOOWN is the percentage of shares held by CEO. OUTOWN is the percentage of shares held by outside blockholders. ACQU is a dummy equal to 1 if the book value of total assets at the fiscal year-end increases by more than 25% from the beginning of the fiscal year, 0 otherwise. DIVE is a dummy equal to 1 if the book value of total assets at the fiscal year-end decreases by more than 25% from the beginning of the fiscal year, 0 otherwise. CROSS is a dummy variable equal to 1 if the firm is cross-listed in a foreign stock exchange, 0 otherwise. EXCASH is the residual from the regression of cash on determinants of cash holdings.6 Other variables are defined as before. The results of this analysis are shown in Table 4. The coefficient on DOLIMIT remains significant with expected positive sign, whereas the coefficient on IMR is not significant at conventional levels. This suggests that the main results unlikely to be driven by the potential endogeneity problem associated with D&O purchase decision. 6 Following Chung and Wynn (2008), cash holdings is defined as the sum of cash, cash equivalents, and short-term investments. The determinants of cash holdings include firm size, market-to-book ratio, leverage ratio, cash flow (defined as earnings before depreciation and amortization, less interest, taxes, and common dividends), net working capital (excluding cash), percentage of independent directors on the board, outside blockholder ownership, engagement in divestures, cross-listing status, and membership in high-tech industry, where cash holdings, cash flow, and net working capital are all scaled by lagged total assets. Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 106 Table 4. Endogeneity Dependent variable: CUT SD group Predicted signs Coefficient z-statistic DOLIMIT + 2.439** 2.53 PCRD – -0.537 -1.47 CIRD – -205.769** -2.25 CGDP – 17.337** 2.52 TOQ – -0.223 -1.47 CSALES – -0.894 -1.50 CCAP – -0.190** -2.30 LnMV – -0.009 -0.07 LEV + 0.328 0.49 FCF – 1.087 1.05 DIST + -0.666** -2.16 INSOWN + -0.386 -0.78 IMR 0.895 1.10 Intercept 13.323*** 5.05 Year / Industry dummy variables Included Pseudo R2 0.108 n 682 The SD group comprises firm-years for which –RDEXPt-1 < (EBTRDt-EBTRDt-1) < 0. RDEXP = R&D expenditures. EBTRD = pre-tax, pre-R&D earnings. Variable definitions: CUT = 1 if R&D decreases relative to the previous year, and 0 otherwise; DOLIMIT = D&O coverage limit scaled by market value of equity; PCRD = the difference in the natural logarithm of R&D between the previous year and the year before; CIRD = the difference in industry R&D intensity (R&D expenditures scaled by total sales) between the current year and the previous year, where the industry is defined using the two-digit Taiwan Economic Journal (TEJ) code; CGDP = the difference in the natural logarithm of GDP between the current year and the previous year; TOQ = the sum of market value of common equity and book value of debt, divided by the book value of total assets; CSALES = the difference in the natural logarithm of sales between the current year and the previous year; CCAP = the difference in the logarithm of capital expenditures between the current year and the previous year; LnMV = the natural logarithm of market value of equity; LEV = total debts divided by total assets; FCF = operating cash flows minus capital expenditures, scaled by total assets; DIST = the change in pre-tax, pre-R&D earnings divided by previous year’s R&D; INSOWN = the percentage of shares held by institutional investors. IMR = the inverse Mills ratio obtained from equation (2). *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively (two-tailed tests). The z-statistics are based on standard errors adjusted for clustering at the firm level. 4.4.2 Alternative Measure of D&O Coverage This paper adopts the raw D&O coverage levels (coverage limits deflated by market value of equity) as the independent variable of interest in the primary test. To provide more direct and compelling evidence on whether abnormally high D&O coverage increases the likelihood of cutting R&D, I replace the raw coverage variable by an estimated abnormal D&O coverage, ABDOLIMIT, which is defined as the residual from the regression of coverage limits on its determinants7. As shown in Table 5, the coefficient on ABDOLIMIT is still positive and significant. This suggests that the results based on abnormal D&O coverage as the dependent 7 Following Wynn (2008) and Chung et al. (2013), the determinants of D&O insurance coverage limits include firm size, leverage ratio, cash holdings, volatility of stock returns (measured as the natural logarithm of annualized variance of daily return over the current fiscal year), percentage of independent directors on the board, outside blockholder ownership, cross-listed status, and membership in high-tech industry. Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 107 variable are very similar to those reported in the paper using D&O coverage levels. Table 5. Abnormal D&O Coverage Dependent variable: CUT SD group Predicted signs Coefficient z-statistic ABDOLIMIT + 2.225** 2.33 PCRD – -0.540 -1.50 CIRD – -211.079** -2.35 CGDP – 13.707** 2.23 TOQ – -0.219 -1.44 CSALES – -0.836 -1.44 CCAP – -0.201** -2.46 LnMV – -0.148* -1.66 LEV + 0.442 0.66 FCF – 0.985 0.96 DIST + -0.680** -2.21 INSOWN + -0.363 -0.74 Intercept 16.861*** 12.32 Year / Industry dummy variables Included Pseudo R2 0.106 n 684 The SD group comprises firm-years for which –RDEXPt-1 < (EBTRDt-EBTRDt-1) < 0. RDEXP = R&D expenditures. EBTRD = pre-tax, pre-R&D earnings. Variable definitions: CUT = 1 if R&D decreases relative to the previous year, and 0 otherwise; ABDOLIMIT = the residual from the regression of DOLIMIT on the economic determinants of D&O coverage; DOLIMIT = D&O coverage limit scaled by market value of equity; PCRD = the difference in the natural logarithm of R&D between the previous year and the year before; CIRD = the difference in industry R&D intensity (R&D expenditures scaled by total sales) between the current year and the previous year, where the industry is defined using the two-digit Taiwan Economic Journal (TEJ) code; CGDP = the difference in the natural logarithm of GDP between the current year and the previous year; TOQ = the sum of market value of common equity and book value of debt, divided by the book value of total assets; CSALES = the difference in the natural logarithm of sales between the current year and the previous year; CCAP = the difference in the logarithm of capital expenditures between the current year and the previous year; LnMV = the natural logarithm of market value of equity; LEV = total debts divided by total assets; FCF = operating cash flows minus capital expenditures, scaled by total assets; DIST = the change in pre-tax, pre-R&D earnings divided by previous year’s R&D; INSOWN = the percentage of shares held by institutional investors. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively (two-tailed tests). The z-statistics are based on standard errors adjusted for clustering at the firm level. 4.4.3 Alternative Measure of Myopic R&D Cutting Behavior In addition to using the previous year’s R&D expenditures as a benchmark, I also adopt the normal level of R&D expenditures using the model developed by Roychowdhury (2006). To estimate the model, I run the following cross-sectional regression for each industry (defined by two-digit TEJ codes) and year. 1 0 1 2 1 1 1 1t t t t t t RDEXP SALES ASSET ASSET ASSET θ θ θ ε− − − − = + + + (3) where RDEXP is R&D expenditures, ASSET is total assets, and SALES is total sales. Consistent with Roychowdhury (2006), I require at least 15 observations for each Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 108 industry-year grouping. The abnormal level of R&D expenditures is measured as the estimated residual from equation (3). I then create a dummy variable, NEGABRD, which equals 1 if firms with negative abnormal R&D expenditures, 0 otherwise. I further exclude firms that issue seasoned equity offerings as they are inclined to seek external financing due to existing R&D projects. Hence, the exclusion of such firms may produce a cleaner sample that exhibits incentives to manage earnings towards performance targets and thus increases the power of the tests. In Table 6, the regression with NEGABRD as the dependent variable shows that the coefficient on DOLIMIT is positive and significant (marginally significant). Therefore, the results are robust to alternative measures of managerial myopia. Table 6. Abnormal R&D Expenditure Dependent variable: NEGABRD SD group Predicted signs Coefficient z-statistic DOLIMIT + -1.704* -1.79 PCRD – -1.604*** -3.33 CIRD – 14.029 0.12 CGDP – 24.693*** 2.98 TOQ – -0.774*** -3.54 CSALES – 1.212** 2.09 CCAP – -0.187** -2.01 LnMV – -0.363*** -3.40 LEV + 2.107*** 2.87 FCF – -0.469 -0.38 DIST + -1.184*** -3.31 INSOWN + 1.394** 2.40 Intercept -12.075 -0.02 Year / Industry dummy variables Included Pseudo R2 0.174 n 598 The SD group comprises firm-years for which –RDEXPt-1 < (EBTRDt-EBTRDt-1) < 0. RDEXP = R&D expenditures. EBTRD = pre-tax, pre-R&D earnings. Variable definitions: NEGABRD = 1 if firms with negative abnormal R&D expenditures, and 0 otherwise, where abnormal R&D expenditures is calculated using the method of Roychowdhury (2006); DOLIMIT = D&O coverage limit scaled by market value of equity; PCRD = the difference in the natural logarithm of R&D between the previous year and the year before; CIRD = the difference in industry R&D intensity (R&D expenditures scaled by total sales) between the current year and the previous year, where the industry is defined using the two-digit Taiwan Economic Journal (TEJ) code; CGDP = the difference in the natural logarithm of GDP between the current year and the previous year; TOQ = the sum of market value of common equity and book value of debt, divided by the book value of total assets; CSALES = the difference in the natural logarithm of sales between the current year and the previous year; CCAP = the difference in the logarithm of capital expenditures between the current year and the previous year; LnMV = the natural logarithm of market value of equity; LEV = total debts divided by total assets; FCF = operating cash flows minus capital expenditures, scaled by total assets; DIST = the change in pre-tax, pre-R&D earnings divided by previous year’s R&D; INSOWN = the percentage of shares held by institutional investors. IMR = the inverse Mills ratio obtained from equation (2). *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively (two-tailed tests). The z-statistics are based on standard errors adjusted for clustering at the firm level. 5. Conclusion In this paper, I examine whether D&O insurance coverage results in managerial myopia. Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 109 Managers have incentives to boost short term performance to achieve targets for compensation and other contract reasons. D&O insurance could cause moral hazard problems and provide managers with incentives to take actions that attain their personal goals. I thus argue that D&O insurance increases managers’ incentives to cut R&D expenditures to achieve earnings targets. Using a sample of 3,308 firm-years from Taiwanese listed firms over the period 2008-2012 that have D&O insurance information and material R&D expenditures (i.e., greater than 1% of sales). I partition the sample into three groups. The small earnings decrease group comprises firms with declines in pre-tax, pre-R&D earnings less than the amount of previous year’s R&D expenditures. In these firms, managers have incentives to cut R&D to avoid earnings declines. Therefore, D&O insurance coverage is predicted to increase the likelihood of cutting R&D for this group. For firms that have a large decrease or an increase in the pre-tax, pre-R&D earnings, managers do not have incentives to reduce R&D to avoid earnings decreases because it is not feasible or necessary to do so. They are treated as control groups. Consistent with my prediction, I find that firms with higher D&O insurance coverage tend to have a significantly higher incidence of cutting R&D in the small earnings decrease group. Also as expected, the impact of D&O insurance coverage is not significant for the control groups. The results are robust to controlling for endogeneity bias by using a Heckman self-selection model and to alternative measures of D&O insurance coverage and myopic R&D cuts. At the practical level, this study has important implications for regulators and accounting standard-setters in their attempts to protect outside investors in the market place. Regulators may need to oversee the amounts of D&O insurance coverage purchased by firms and require them to strengthen corporate governance effectiveness. Acknowledgement I gratefully acknowledge the financial support provided by the research grant from Ministry of Science and Technology of Taiwan R.O.C. (Grant number: 104-2410-H-035-024). References Baber, W. R., Fairfield, P. M., & Haggard, J. A. (1991). The effect of concern about reported income on discretionary spending decisions: The case of research and development. The Accounting Review, 66(4), 818-829. Baker, T., & Griffith, S. J. (2010). Ensuring corporate misconduct: How liability insurance undermines shareholder litigation. University of Chicago Press. http://dx.doi.org/10.7208/chicago/9780226035079.001.0001 Bange, M. M., & De Bondt, W. F. (1998). R&D budgets and corporate earnings targets. Journal of Corporate Finance, 4(2), 153-184. http://dx.doi.org/10.1016/S0929-1199(98)00006-6 Baysinger, B. D., Kosnik, R. D., & Turk, T. A. (1991). Effects of board and ownership structure on corporate R&D strategy. Academy of Management Journal, 34(1), 205-214. http://dx.doi.org/10.2307/256308 Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 110 Bens, D. A., Nagar, V., & Wong, M. H. (2002). Real investment implications of employee stock option exercises. Journal of Accounting Research, 40(2), 359-393. http://dx.doi.org/10.1111/1475-679X.00052 Boubakri, N., Ghalleb, N., & Boyer, M. M. (2008). Managerial opportunism in accounting choice: Evidence from directors’ and officers’ liability insurance purchases. Working paper. University of Montreal. http://dx.doi.org/10.2139/ssrn.1109254 Bushee, B. J. (1998). The influence of institutional investors on myopic R&D investment behavior. The Accounting Review, 73(3), 305-333. Chalmers, J. M., Dann, L. Y., & Harford, J. (2002). Managerial opportunism? Evidence from directors’ and officers’ insurance purchases. The Journal of Finance, 57(2), 609-636. http://dx.doi.org/10.1111/1540-6261.00436 Chen, X., Cheng, Q., Lo, A. K., & Wang, X. (2015). CEO contractual protection and managerial short-termism. The Accounting Review, 90(5), 1871-1906. http://dx.doi.org/10.2308/accr-51033 Chen, Z., Li, O. Z., & Zou, H. (2016). Directors’ and officers’ liability insurance and the cost of equity capital. Journal of Accounting and Economies, 61(1), 100-120. http://dx.doi.org/10.1016/j.jacceco.2015.04.001 Chung, H. H., & Wynn, J. P. (2008). Managerial legal liability coverage and earnings conservatism. Journal of Accounting and Economics, 46(1), 135-153. http://dx.doi.org/10.1016/j.jacceco.2008.03.002 Chung, H. H., Wynn, J. P., & Yi, H. (2013). Litigation risk, accounting quality, and investment efficiency. Advances in Accounting, 29(2), 180-185. http://dx.doi.org/10.1016/j.adiac.2013.09.009 Chung, H. S. H., Hillegeist, S. A., & Wynn, J. P. (2015). Directors’ and officers’ legal liability insurance and audit pricing. Journal of Accounting and Public Policy, 34(6), 551-577. http://dx.doi.org/10.1016/j.jaccpubpol.2015.05.009 Cohen, D. A., & Zarowin, P. (2010). Accrual-based and real earnings management activities around seasoned equity offerings. Journal of Accounting and Economics, 50(1), 2-19. http://dx.doi.org/10.1016/j.jacceco.2010.01.002 Core, J. E. (1997). On the corporate demand for directors’ and officers’ insurance. Journal of Risk and Insurance, 64(1), 63-87. http://dx.doi.org/10.2307/253912 Dechow, P. M., & Sloan, R. G. (1991). Executive incentives and the horizon problem: An empirical investigation. Journal of Accounting and Economics, 14(1), 51-89. http://dx.doi.org/10.1016/0167-7187(91)90058-S Ellis, C. D. (2004). Capital: The story of long-term investment excellence. Hoboken, NJ: John Wiley & Sons. Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 111 Fama, E. F., & Jensen, M. C. (1983). Separation of ownership and control. Journal of Law and Economics, 26(2), 301-325. http://dx.doi.org/10.1086/467037 Fuller, J., & Jensen, M. C. (2002). Just say no to Wall Street: Putting a stop to the earnings game. Journal of Applied Corporate Finance, 14(4), 41-46. http://dx.doi.org/10.1111/j.1745-6622.2002.tb00447.x Graham, J. R., Harvey, C. R., & Rajgopal, S. (2005). The economic implications of corporate financial reporting. Journal of Accounting and Economics, 40(1), 3-73. http://dx.doi.org/10.1016/j.jacceco.2005.01.002 Heckman, J. J. (1979). Sample selection bias as a specification error. Econometrica: Journal of the Econometric Society, 47(1), 153-161. http://dx.doi.org/10.2307/1912352 Holderness, C. G. (1990). Liability insurers as corporate monitors. International Review of Law and Economics, 10(2), 115-129. http://dx.doi.org/10.1016/0144-8188(90)90018-O Holmstrom, B. (1989). Agency costs and innovation. Journal of Economic Behavior & Organization, 12(3), 305-327. http://dx.doi.org/10.1016/0167-2681(89)90025-5 Kalelkar, R., & Nwaeze, E. T. (2015). Directors and officers liability insurance: Implications of abnormal Coverage. Journal of Accounting, Auditing & Finance, 30(1), 3-34. http://dx.doi.org/10.1177/0148558X14549459 Kim, I. (2015) Directors’ and officers’ insurance and opportunism in accounting choice. Accounting & Taxation, 7(1), 51-65. Kothari, S. P., Laguerre, T. E., & Leone, A. J. (2002). Capitalization versus expensing: Evidence on the uncertainty of future earnings from capital expenditures versus R&D outlays. Review of Accounting Studies, 7(4), 355-382. Lin, C., Officer, M. S., & Zou, H. (2011). Directors’ and officers’ liability insurance and acquisition outcomes. Journal of Financial Economics, 102(3), 507-525. http://dx.doi.org/10.1016/j.jfineco.2011.08.004 Lin, C., Officer, M. S., Wang, R., & Zou, H. (2013). Directors’ and officers’ liability insurance and loan spreads. Journal of Financial Economics, 110(1), 37-60. http://dx.doi.org/10.1016/j.jfineco.2013.04.005 O’Sullivan, N. (2002). The demand for directors’ and officers’ insurance by large UK companies. European Management Journal, 20(5), 574-583. http://dx.doi.org/10.1016/S0263-2373(02)00096-8 Osma, B. G., & Young, S. (2009). R&D expenditure and earnings targets. European Accounting Review, 18(1), 7-32. http://dx.doi.org/10.1080/09638180802016718 O’Sullivan, N. (1997). Insuring the agents: The role of directors’ and officers’ insurance in corporate governance. Journal of Risk and Insurance, 64(3), 545-556. http://dx.doi.org/10.2307/253764 Asian Journal of Finance & Accounting ISSN 1946-052X 2016, Vol. 8, No. 2 ajfa.macrothink.org 112 Porter, M. E. (1992). Capital choices: Changing the way America invests in industry. Journal of Applied Corporate Finance, 5(2), 4-16. http://dx.doi.org/10.1111/j.1745-6622.1992.tb00485.x Rappaport, A. (2005). The economics of short-term performance obsession. Financial Analysts Journal, 61(3), 65-79. http://dx.doi.org/10.2469/faj.v61.n3.2729 Sloan, R. G. (1996). Do stock prices fully reflect information in accruals and cash flows about future earnings?. The Accounting Review, 71(3), 289-315. Stein, J. C. (1988). Takeover threats and managerial myopia. The Journal of Political Economy, 96(1), 61-80. http://dx.doi.org/10.1086/261524 Stein, J. C. (1989). Efficient capital markets, inefficient firms: A model of myopic corporate behavior. The Quarterly Journal of Economics, 104(4), 655-669. http://dx.doi.org/10.2307/2937861 Wynn, J. P. (2008). Legal liability coverage and voluntary disclosure. The Accounting Review, 83(6), 1639-1669. http://dx.doi.org/10.2308/accr.2008.83.6.1639 Zou, H., Wong, S., Shum, C., Xiong, J., & Yan, J. (2008). Controlling-minority shareholder incentive conflicts and directors’ and officers’ liability insurance: Evidence from China. Journal of Banking & Finance, 32(12), 2636-2645. http://dx.doi.org/10.1016/j.jbankfin.2008.05.015