MANAGERIAL POWER IN THE DESIGN OF EXECUTIVE COMPENSATION: EVIDENCE FROM JAPAN

Transcription

1 MANAGERIAL POWER IN THE DESIGN OF EXECUTIVE COMPENSATION: EVIDENCE FROM JAPAN Stephen P. Ferris, Kenneth A. Kim, Pattanaporn Kitsabunnarat and Takeshi Nishikawa ABSTRACT Using a sample of 466 grants of stock options to executives of Japanese firms over the years , this study tests the managerial power theory of compensation design developed by Bebchuk, Fried, and Walker (2002) and Bebchuk and Fried (2004). This theory argues that managers of firms with weak corporate governance will use their power to design executive compensation that is manager-advantageous. Using our option grants sample, we test to determine if any of the firm s governance mechanisms are able to limit managerial self-dealing with respect to executive stock options. We find that smaller boards and a higher percentage of independent directors are important governance mechanisms for the control of managerial influences in the design of stock-option compensation. An alternative hypothesis, that firms elect to grant advantageously designed options to encourage risk taking by managers, is not supported by our empirical results. Finally, we determine that the Issues in Corporate Governance and Finance Advances in Financial Economics, Volume 12, 3 26 Copyright r 2007 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: /doi: /S (07)

2 4 STEPHEN P. FERRIS ET AL. market response to the announcements of such grants varies inversely with the extent to which the options are managerially advantageous. Overall, we conclude that managerial power effects are present in the design of executive stock options and that theory of managerial power advanced by Bebchuk et al. holds internationally. 1. INTRODUCTION Recently, Bebchuk et al. (2002) and Bebchuk and Fried (2004) contend that the exercise of managerial power rather than the pursuit of optimal contracting best explains the process by which executive compensation is designed in the United States. Under Bebchuk et al. s managerial power paradigm, executives have the ability to influence their compensation levels by the nature of their positions, their level of equity ownership, and the organizational structure of the board. The managerial power approach views executive compensation as part of the agency problem, with managers using their control over compensation to obtain economic rents. 1,2 This managerial power model of compensation contrasts with the more familiar theory of optimal contracting. In that theory, compensation is seen as a tool to minimize the agency conflict that exists between senior managers and shareholders (e.g., Jensen & Meckling, 1976). The level and composition of compensation are established to provide managers with an optimal set of incentives to undertake those activities which maximize shareholder value. Executive compensation is viewed as a mechanism to help align the interests of managers and shareholders and thereby reduce agency conflict. Bebchuk et al. (2002) and Bebchuk and Fried (2004) further observe that many of the current practices surrounding the granting of executive stock options can not be reconciled with the incentive alignment sought by optimal contracting. Rather, they argue such practices are indicative of a managerial power approach towards compensation. Among the practices that Bebchuk et al. claim are inconsistent with optimal contracting include the granting of options that are already in the money, short waiting periods prior to exercise, and incentive compensation representing only a small percentage of total compensation. In this study, we test for the presence of managerial power effects over compensation in the world marketplace by examining the nature of executive stock options granted in Japan. We elect to examine equity options since they are the compensation component on which Bebchuk et al.

3 The Design of Executive Compensation 5 (2002) and Bebchuk and Fried (2004) focus their theoretical analysis and develop the empirical conjectures that we test. For every Japanese executive stock option granted, we measure its vesting period before it can be exercised, its duration term before expiration, its size relative to fixed compensation, and its moneyness. Like Bebchuk et al. (2002), we consider an executive option to be manager-advantageous if the option can be immediately exercised, has a long duration, is small relative to fixed compensation, and is closer to being in-the-money. Further, following Bebchuk et al. (2002) and Bebchuk and Fried (2004) view of empowered managers, we consider those firms with weak internal governance as having empowered managers. According to Bebchuk et al., firms with weak or ineffectual governance allow empowered managers to emerge who are able to award themselves additional compensation, including advantageously designed options. This study proceeds to test whether or not such empowered managers enrich themselves through the award of options that are designed in a highly manager-advantageous fashion. In addition to being one of the first to test Bebchuk et al. s managerial power view of compensation design, this study makes two other contributions to the compensation literature. First, by studying Japan s experience with executive options, we are able to observe whether the managerial power view of compensation applies only to the U.S. or if it is also globally relevant. This determination is important because it can provide insight into the international usefulness of executive compensation and its design to mitigate the agency conflict inherent in the corporate organizational form. Given the extensive literature that has documented the existence of agency conflict worldwide (e.g., see LaPorta, Lopez-de-Silanes, & Shleifer, 1999; LaPorta, Lopez-de-Silanes, Shleifer, & Vishny, 2000; Denis & McConnell 2003), this study offers valuable insight into how the design of compensation can align the conflicting interests of shareholders and managers. Kato, Lemmon, Luo, and Schallheim (2005) also study Japanese executive stock options and their impact on agency conflict, but do not consider the crosssectional differences in the design of executive stock options. Second, using Japanese executive stock options to test Bebchuk et al. (2002) and Bebchuk and Fried (2004) thesis allows us to consider governance mechanisms that are less common or absent from the U.S. marketplace. For example, Japanese firms often maintain high leverage and close ties with banks (Kester, 1986; Hoshi, Kashyap, & Scharfstein, 1990). These banks can be large creditors and/or significant equityholders. If banks have a significant presence in the firm, then they possess a strong incentive

4 6 STEPHEN P. FERRIS ET AL. to monitor the firm. Diamond (1984) argues that banks can be better monitors of firms activities than other investors. Japanese firms might also belong to a keiretsu, which is a diverse group of firms interlinked through corporate cross-shareholdings and an extensive network of product market and financial relationships (Nakatani, 1984; Prowse, 1990; Berglo f & Perotti, 1994). While the early literature reveals benefits to maintaining bank relations and business group membership, some of the more recent literature suggests the existence of costs resulting from these relations. Weinstein and Yafeh (1998) find that banks might extract rents from their client firms, while Ferris, Kim, and Kitsabunnarat (2003) and Khanna and Yafeh (2005) report little benefits to business group membership. This study thus contributes to the ongoing debate regarding the usefulness of bank relations and group affiliation. More specifically, our analysis examines how the existence of bank relationships and/or keiretsu membership can influence managerial power and compensation design. Our study sample consists of 466 Japanese executive stock options granted during a 5 year sample period extending from 1997 to We test to determine if the presence of various internal governance mechanisms is able to limit the ability of managers to exploit the design of their option compensation. The governance mechanisms that we consider include ownership structure (managerial and institutional), board structure (independence and size), debt (total debt and bank debt), and keiretsu affiliation. We find that firms governed by boards with a higher fraction of independent members, but fewer total members are less likely to grant executive options that are highly advantageous to managers. We also test an alternative hypothesis that firms grant favorably designed options to increase risktaking by managers, but our empirical results are not consistent with such a process. Finally, we conduct tests on the market reaction to announcements of option awards. We find positive abnormal returns to the announcement of executive option grants, which is consistent with most U.S. evidence (e.g., Brickley, Bhagat, & Lease, 1985; DeFusco, Johnson, & Zorn, 1990). We then categorize these options according to the degree to which they are managerially advantageous. We find that the market reacts less favorably to the more managerially advantageous options. Overall, our findings provide empirical support for Bebchuk et al. (2002) and Bebchuk and Fried (2004) managerial power view of compensation design, with powerful managers receiving options that are designed for their benefit. A firm s corporate governance structure, specifically its board of directors, can limit the ability of managers to self-deal in the design of their compensation. We determine that smaller boards and a higher percentage of

5 The Design of Executive Compensation 7 independent directors are important governance mechanisms for the control of managerial influences in the design of stock-option compensation. We also find that the market response to the announcement of such grants varies inversely with the extent to which the options are designed to be managerially advantageous. We organize the remainder of this study into three sections. In the following section we briefly discuss Japanese executive option programs, describe our data, and provide some descriptive summary statistics. In Section 3, we present our major empirical findings. We conclude in Section 4 with a brief summary and a discussion of our findings. 2. JAPANESE EXECUTIVE OPTION PROGRAMS, DATA SOURCES, AND SAMPLE CHARACTERISTICS 2.1. Japanese Executive Stock-Option Programs Historically, Japanese firms were not permitted to use stock options as a component of executive compensation. On May 16, 1997, the Japanese Commercial Code was amended to allow such options for both executives and employees. This amendment became effective on June 1, Japanese stock-option grants must be approved at a general stockholder s meeting. The terms of the option must include the recipient s identity, the number of shares, the share price, time to exercise, the option s duration, and the exercise price. The duration of a stock option is limited to 10 years and the total shares contained in an option plan cannot exceed 10% of the firm s outstanding shares. Although these options are not presently expensed, there is a continuing policy debate about changing this Data Sources Our sample initially includes all executive options granted by non-financial firms listed on the Tokyo Stock Exchange over the period The data used in our study are drawn from four different sources. Information related to stock-option grants, such as announcement dates, grant dates, exercise price, vesting and expiration periods, are obtained from Daiwa Securities. Data on the firms block equityholders, number of board members, fraction of independent board members, and bank debt data are collected from various issues of Kigyo Keiretsu Soran. We should mention

6 8 STEPHEN P. FERRIS ET AL. that although Kigyo Keiretsu Soran does not identify which directors are independent, it does provide information on directors employment history and directors current affiliation. Following Kang and Shivdasani (1999) and Miwa and Ramseyer (2005), we identify directors not affiliated with the firm as an independent director. The executive compensation data are taken from the Nikkei Economic Electronic Databank System (NEEDS) database developed by Nihon Keizai Shimbun America, Inc. Total compensation for executives is at the firm level. Finally, the financial statement data and stock return data are obtained from the PACAP Databases Japan. We initially identify 1,259 employee stock-option grants. We eliminate 274 grants approved for non-executives. Another 230 option grants are eliminated due to insufficient data regarding option characteristics. Of the remaining grants, we retain only those for which we have complete governance data for the granting firm. Application of these screens results in a final sample of 466 options that were granted by 255 different firms Firm Characteristics The time series of our sample of option grants is presented in Panel A of Table 1. Our 466 observations are distributed over 5 years, with the majority of them occurring in the years 2000 and No firm, however, grant more than one option per year. This reduces concerns over repeated observations of a given firm within our sample. This clustering of observations over the last 2 years of our sample is also noted by Kato et al. (2005). The lack of observations early in the sample period probably reflects slowness by Japanese firms to exploit the regulatory change allowing the use of options in executive compensation. Our analysis of the industry distribution of executive options shows that they are employed across most industry types, but with a heavy concentration in manufacturing and wholesaling/retail. Rather than representing any particular industry trend, this pattern corresponds to the distribution of Japanese business activity across these industries. In Panel C, we provide a comparative analysis of the option-granting and non-granting firms. We observe a number of important differences between these two groups of firms. We find that the granting firms are significantly larger than those that do not grant options. This difference is robust to both market and book measures of size. The granting firms are also more profitable, use less debt, are less risky, and enjoy higher market valuations than their non-granting counterparts.

7 The Design of Executive Compensation 9 Table 1. Executive Option Grants: Sample Characteristics. Panel A: Number of executive options granted by year Year Number of Observations % of Sample Total Panel B: Number of executive options granted by industry Industry Number of Observations % of Sample Agriculture, forestry, fishery, and mining Construction Manufacturing Wholesale and retail Real estate Transportation and communication Services Total Panel C: Characteristics of firms that grant executive options Firm Characteristics Granting Non-Granting Difference Total book assets (in millions of yen) ( ) ( ) Market value of equity (in thousands of yen) ( ) ( ) Total debt/total assets (0.467) (0.582) EBIT/total assets (0.037) (0.026) EPS = net income/shares outstanding (19.039) (8.427) Beta (0.756) (0.993) Market-to-book value of assets (1.157) (0.989) Number of observations Notes: This table provides sample size information for a sample of 466 executive options that were granted by non-financial Japanese firms during the period from 1997 to The number of grants is presented by year (Panel A) and by industry (Panel B). The characteristics of firms that grant options and that do not grant options are also presented (Panel C). For all option granting firms, the firm-specific variables are measured at the fiscal year-end immediately preceding the option grant announcement date. Means, standard deviations (in parentheses), and difference in means, are reported. Significance at the 1% level.

8 10 STEPHEN P. FERRIS ET AL. Overall, the findings contained in these three panels suggest that the adoption of executive stock options began slowly in Japan after their legalization, but has accelerated in more recent years. Executive stock options are being used by firms in all industry groups in approximate proportion to their presence in the Japanese economy. The firms that elect to use them are large and profitable. Their low leverage ratios and high market valuations further suggest that these granting firms are also generally well managed Option and Governance Characteristics In Table 2, we describe the characteristics of the options that serve as the focus of our analysis as well as the governance structure for our optiongranting sample firms. We see that the number of options issued by these granting firms is small relative to the total number of shares outstanding. This is likely to be reflective of the relative newness of options and the desire of firms to be conservative in their initial use of them in Japanese compensation packages. The waiting period to exercise is approximately 20 months, while most options in our sample have a life span of about 4 years (46.5 months). In addition to a stock option, an executive might receive a bonus or an increase in their base salary. From Table 2 findings, we see that the value of the option grants relative to other forms of additional pay is fairly high at 70%. This suggests the existence of a strong pay-forperformance relation for those executives awarded these options. Finally, we observe that most options are awarded out-of-the-money. This contrasts with the U.S. experience where Bebchuk et al. (2002) note that most options granted are at-the-money. In Panel B of Table 2, we provide an overview of the governance structure for our sample firms. These variables are important to our later analysis of the ability of the firm s internal corporate governance mechanisms to control the design of options awarded to executives. We find that, on average, corporate managers own B7% of the firm s equity while financial institutions (34.4%) and other corporations (21.8%) are the other large equity investors. These ownership statistics are consistent with Prowse (1992) results concerning the equity ownership structure of Japanese firms. The boards of directors for our sample firms are large by U.S. standards and average over 19 individuals. A little over 14% of these directors can be classified as independent. These board characteristics are similar to those reported by Kang and Shivdasani (1999). Our sample firms finance about

9 The Design of Executive Compensation 11 Table 2. Option Characteristics and Governance Characteristics. Variable Mean Median Q1 Q3 Std. deviation Panel A: Executive option characteristics Number of option shares/total shares outstanding Waiting period to exercise (in months) Length of time before option expires (in months) Option value/(option bonus + raise in salary) (Current stock price option exercise price)/current stock price Panel B: Governance characteristics and risk-taking of firms that grant executive options % ownership by managers % ownership by financial institutions % ownership by other corporations % independent directors to total directors Board size (number of directors) Total debt/total assets % of firms with bank loans % of firms that belong to a keiretsu Standard error from market model Notes: Panel A provides summary statistics for a sample of 466 executive options that were granted by non-financial Japanese firms during the period from 1997 to In Panel B, the governance characteristics of firms that grant options are presented. All governance variables are measured at the fiscal year-end immediately preceding the option grant announcement date. The standard error is estimated from the market model in which the firm s monthly returns are regressed on the monthly returns of an equally weighted market portfolio for a 60-month period prior to the option grant.

10 12 STEPHEN P. FERRIS ET AL. half of their assets with debt (46.8%) while over 80% utilize some kind of bank debt. Approximately 20% of our sample firms are members of a keiretsu. 4 Finally, the panel also reports summary statistics on firm-specific standard errors from a market model. We use this variable to test the possibility that options are granted in an effort to increase managerial risk taking. 3. EMPIRICAL FINDINGS 3.1. The Relation between Internal Governance and Options In this section, we conduct a preliminary examination of the relation between the characteristics of the option granted and the firm s corresponding governance structure. Bebchuk et al. (2002) and Bebchuk and Fried (2004) argue that weak or ineffectual governance allows managers to obtain more power or to become entrenched. Specifically, firms with disperse ownership or fewer institutional shareholders, weak or ineffectual boards, and anti-takeover provisions, are posited to have empowered managers. In this study, the governance characteristics that we consider are ownership structure (managerial and institutional), board structure (independence and size), financial structure (total debt and bank debt), and keiretsu membership. We ignore anti-takeover provisions because they are practically nonexistent in Japan. Firms with high levels of equity owned by managers might grant fewer manager-advantageous options because these managers are less likely to demand options that are harmful to share value. But it is equally likely that managers as large shareholders are more empowered to grant themselves advantageous options. Assessing the ultimate effect of managerial ownership on the nature of executive stock options remains an empirical issue that is addressed later in this study. Concentrated ownership of equity by external investors is established in the literature as a monitor of firm activity (e.g., Demsetz & Lehn, 1985; LaPorta et al., 1999). In our sample, the large external shareholders are all institutions. Therefore, firms with high levels of equity ownership by financial institutions and other corporations might grant fewer manageradvantageous options. Firms with boards that have more independent directors and fewer directors are usually described in the governance literature as more effective. Weisbach (1988) finds that independent board members are

11 The Design of Executive Compensation 13 more effective monitors than insiders. Yermack (1996) determines that smaller boards are more active monitors, with board processes losing effectiveness as board size increases. Consequently, we anticipate that firms with a higher fraction of independent directors or smaller boards are better able to limit managerial self-dealing in the design of their executive stock options. Jensen (1986) argues that debt service can offer a partial solution to the agency problem between managers and shareholders. Further, Hoshi et al. (1990) note that Japanese banks are active monitors of their client firms. Prowse (1990) contends that firms belonging to a keiretsu have multiple monitors through the many implicit contracts that exist among group members. Therefore, firms with more debt and/or a greater bank presence, and firms that belong to a keiretsu, are likely to be better monitored and thus grant fewer manager-advantageous options. Previous studies examine the relation between a firm s corporate governance and its executive compensation. Hartzell and Starks (2003) find that CEO compensation is higher in firms with low institutional ownership. Core, Holthausen, and Larcker (1999) report that CEO compensation is greater when firms boards are larger and less independent. Their studies are broadly consistent with Bebchuk et al. (2002) and Bebchuk and Fried (2004) managerial power view of executive compensation. Our study contributes to this literature by focusing exclusively on executive stock options as a device for managerial exploitation of entrenchment. To conduct our empirical tests, we first create an options characteristics index which measures the extent to which an option grant possesses characteristics that are advantageous to managers. Our option characteristics index is the sum of four different dummy variables. The first dummy variable is equal to one if the option s time to exercise is below the sample median and zero otherwise. The second dummy variable has a value of one if the option s duration to expiration is above the sample median and zero otherwise. A third dummy variable is equal to one if the option value relative to fixed compensation is below the sample median and zero otherwise. 5 A fourth and final dummy variable assumes a value of one if the option s moneyness is above the sample median and zero otherwise. For each executive option, the value of these four dummy variables is summed to create an aggregate score. Index values range from 0 to 4, with higher values of the index being consistent with options that are more advantageous to the manager. 6 Of the 466 option grants in our sample, 19 have an option index score of 0, 121 have an index score of 1, 161 have a score of 2, 119 have a score of 3, and 46 have a score of 4. 7

12 14 STEPHEN P. FERRIS ET AL. To determine whether the firm s governance structure affects the design of their executive option grants, we assign firms into subsamples based on ownership distribution (equity ownership by managers, financial institutions, and other corporations), board structure (size and independence), and the total debt ratio. Firms in the high and low subsamples are those firms whose governance measures are above and below the corresponding sample median, respectively. The subsamples for bank debt or keiretsu membership are constructed on the presence or absence of their characteristics rather than reference to a median value. For each subsample, we report the mean (median) value of the options characteristics index. Table 3 contains our corresponding statistical tests of differences between the various subsamples. In Table 3, we observe statistically significant differences in the option characteristics index for three classifying variables. Firms with higher managerial-ownership grant executive options that are less manageradvantageous than compared to firms with low managerial ownership (1.61 mean index vs mean index). This finding suggests that the existence of significant managerial ownership can mitigate managerial power effects. Managers appear not to grant themselves self-serving options at the expense of their share value. We further see that firms with more independent board members and smaller boards grant options that are less manager advantageous. The other governance characteristics do not seem to influence option design. Therefore, it appears that board structure is one of the most-important governance-related factors regarding the ability of managers to empower themselves and thereby extract favorably written stock options. This result is consistent with Bebchuk et al. (2002) and Bebchuk and Fried (2004) thesis that firms with relatively weak or ineffectual boards empower managers who can subsequently influence their compensation in self-serving ways. This result also contributes to the board literature which suggests that boards with a higher fraction of independent directors and fewer directors are better boards. The greater explanatory power of boards regarding compensation design relative to other governance mechanisms reflects its explicit responsibility for the design of executive compensation. Therefore, it is not surprising that better boards grant fewer manager-advantageous options. In this study, we also test an alternative hypothesis that firms grant options to managers to encourage risk-taking. Consistent with Demsetz and Lehn (1985), Prowse (1992), Himmelberg, Hubbard, and Palia (1999), and Holderness, Kroszner, and Sheehan (1999), we use the standard error from

13 Table 3. Option Characteristic Index Values and Measures of Firm Governance. High Low High Low Mean Median Mean Median Dif. mean t-test Dif. median Z-test % ownership by managers % ownership by financial institutions % ownership by other corporations % independent directors to total directors Board size (number of directors) Total debt/total assets Bank loan dummy Keiretsu dummy Standard error of market model Notes: The sample consists of 466 option grant announcements made by non-financial Japanese firms from 1997 to We create a manager-advantageous option index as follows. We create four dummy variables, where the first dummy variable is equal to one if the option s time to exercise is below the sample median, a second dummy variable is equal to one if the option s duration to expiration is above the sample median, a third dummy variable is equal to one if the option value relative to fixed compensation is below the sample median, and a fourth dummy variable is equal to one if the option s moneyness is above the sample median. For each executive option, the values of these four dummy variables are added up to create a manager-advantageous option index from 0 to 4, where 4 denotes those options that are most manager-advantageous. The sample is divided into subsamples based on the firm s ownership distribution (equity ownership by managers, financial institutions, and other corporations), board structure (size and independence), total debt ratio, and standard error from a market model. Firms in the high and low subsamples are those firms whose governance measures are above and below the sample median, respectively. Firms that have bank debt and that belong to a keiretsu are also put into a high subsample. Firms with standard errors above the median are also put into the high subsample. All governance characteristics are measured at the fiscal year-end immediately preceding the option grant date. Standard error is estimated from the market model in which the firm s monthly returns are regressed on the monthly returns of an equally weighted market portfolio for a 60-month period prior to the option grant. For each subsample, we report the mean and median manager-advantageous option index. We conduct t-tests for the mean differences in the index between the high and low subsamples and Wilcoxon Z-tests for the median differences in the index between the high and low subsamples. * Significance at the 10% level. Significance at the 5% level. Significance at the 1% level. The Design of Executive Compensation 15

14 16 STEPHEN P. FERRIS ET AL. the market model as a proxy for firm-specific risk. More specifically, our standard error is estimated from the market model in which the firm s monthly returns are regressed on the monthly returns of an equally weighted market portfolio for a 60-month period prior to the option grant. If firms are rewarding managers for risk taking, then we expect those firms with manager-advantageous options to exhibit more risk (i.e., higher standard errors). From the last row in Table 3, however, we see that such a relation does not hold Logit Model of the Relation between Internal Governance and Options To better understand the cross-sectional variation of our option characteristics index, we present estimates from a multivariate ordered logit regression model using our option index as the dependent variable. The explanatory variables include all of our governance variables, including ownership structure (managerial and institutional), board structure (independence and size), capital structure (total debt and bank debt), and a keiretsu membership dummy variable. The summary statistics of these variables are reported in Table 2. The ordered logit regression results are reported in Table 4. In Table 4, we again see that firms with boards having a higher fraction of independent directors grant fewer manager-advantageous options. The parameter coefficient on the independent director variable is negative and statistically significant at conventional levels. Firms with fewer board members also grant fewer manager-advantageous options. The parameter coefficient on the board size variable is positive and statistically significant. These board variable results confirm our findings from Table 3; better boards are able to limit managerial power effects in compensation design. Unlike Table 3, however, we see that the managerial-ownership variable is not significant when controlling for other governance variables. Interestingly, we see that when a large fraction of equity is held by other corporations, those firms grant more manager-advantageous options. This result suggests that other corporations serve as only weak monitors, allowing managerial power effects to influence compensation design. Our finding that corporate shareholders might be weak monitors is consistent with the conclusions of previous researchers. Lichtenberg and Pushner (1994) and Pushner (1995) find a statistically significant negative relation between firm performance and equity ownership by corporate shareholders. They argue that Japanese firms with large corporate shareholders are

15 The Design of Executive Compensation 17 Table 4. Ordered Logit Analysis of Equity Stock Option Grants. Explanatory Variables Parameter Coefficients Intercept Intercept Intercept Intercept % ownership by managers % ownership by financial institutions % ownership by other corporations % independent directors to total directors Board size (number of directors) Total debt/total assets Bank loan dummy Keiretsu dummy Standard error of market model Pseudo R Number of observations Notes: This table presents parameter coefficient estimates using an ordered logit regression model. The sample consists of 466 option grant announcements made by non-financial Japanese firms from 1997 to The dependent variable is the manager-advantageous option index, which is calculated as follows. We create four dummy variables, where the first dummy variable is equal to one if the option s time to exercise is below the sample median, a second dummy variable is equal to one if the option s duration to expiration is above the sample median, a third dummy variable is equal to one if the option value relative to fixed compensation is below the sample median, and a fourth dummy variable is equal to one if the option s moneyness is above the sample median. For each executive option, the values of these four dummy variables are added up to create a manager-advantageous option index from 0 to 4, where 4 denotes those options that are most manager-advantageous. The explanatory variables are all measured at the fiscal year-end immediately preceding the option grant date. The bank loan dummy is equal to one if the firm has bank debt, otherwise the dummy is equal to zero. The keiretsu dummy is equal to one if the firm belongs to a keiretsu, otherwise the dummy is equal to zero. The standard error is estimated from the market model in which the firm s monthly returns are regressed on the monthly returns of an equally weighted market portfolio for a 60-month period prior to the option grant. Significance at the 10% level. Significance at the 5% level. Significance at the 1% level. insulated from external discipline, allowing managers to consume perquisites rather than pursuing value maximization. Finally, we see again that firms with higher levels of idiosyncratic risk (i.e., firms with higher standard errors) do not grant more manageradvantageous options. The statistical insignificance of the idiosyncratic risk

16 18 STEPHEN P. FERRIS ET AL. proxy is inconsistent with the view that options are granted to encourage managerial risk taking. An issue that raises concerns about the component of the option design, its vesting period, its duration term, its relative size, or its moneyness, is the primary explanatory factor in the logit models presented in Table 4. In unreported tables, where each option characteristic is an indicator dependent variable in a logit model or a continuous dependent variable in an OLS model, we find that the board variables have the most explanatory power, and with the hypothesized signs, when the option s relative size and its moneyness are dependent variables. In other models, the board variables have the hypothesized signs, but are statistically insignificant. Unlike Table 4, however, the total debt ratio is statistically significant and possesses the hypothesized sign in almost all of the models. 8 We choose not to report these results given the limited additional information they provide Announcement Period Returns For each announcement of a stock-option grant, we calculate daily abnormal returns using a standard event-study methodology. We estimate market model parameters over a 200 trading day interval, beginning day 220 relative to the announcement day and ending on day 21. Our benchmark market return is the PACAP equally weighted market return. 9 Daily abnormal returns are used to obtain a cumulative abnormal return (CAR) from day t prior to the option grant announcement to day +t following the announcement. Because of missing return data for our sample firms, CARs can be calculated for only 411 of our sample option grants. Table 5 presents CARs for various event windows. In Panel A of Table 5, we observe that the cumulative abnormal returns are uniformly positive and significant across all estimation windows. For example, the mean CAR( 1,1) and CAR( 10,10) are 1.02% and 1.76%, respectively; both are significant at the 1% level. This result is consistent with previous findings for U.S. (e.g., Brickley et al., 1985; DeFusco et al., 1990) and Japanese (Kato et al., 2005) announcements of executive stock options. Our results indicate that the market favorably capitalizes the anticipated changes resulting from an increased use of executive equity incentive compensation. We next conduct an initial analysis of how cross-sectional differences in option characteristics affect the magnitude of the announcement period returns. Using the options characteristics index, we use an index value of 2 as

17 The Design of Executive Compensation 19 Table 5. Stock Price Reaction Around Executive Option Grant Dates. Panel A: CARs for all option grant announcements N Mean Median CAR ( 1,0) CAR (0,1) CAR ( 1,1) CAR ( 5,5) CAR ( 10,10) Panel B: CARs by manager-advantageous option index Index = 0, 1, or 2 Index = 3 or 4 N Mean Median N Mean Median CAR ( 1,0) CAR (0,1) CAR ( 1,1) CAR ( 5,5) CAR ( 10,10) Notes: The sample consists of 411 option grant announcements made by non-financial Japanese firms from 1997 to We compute abnormal daily returns using the market model. We estimate the market model by using 200 trading days of return data ending 21 days before the repurchase announcement. We use PACAP equally weighted market returns as the benchmark. CAR( t, t) is the cumulative abnormal return from day t to day t where day 0 is the option grant announcement date. In Panel B, we split the sample into a high manager-advantageous option index and a low index. The index is created as follows. We create four dummy variables, where the first dummy variable is equal to one if the option s time to exercise is below the sample median, a second dummy variable is equal to one if the option s duration to expiration is above the sample median, a third dummy variable is equal to one if the option value relative to fixed compensation is below the sample median, and a fourth dummy variable is equal to one if the option s moneyness is above the sample median. For each executive option, the values of these four dummy variables are added up to create a manager-advantageous option index from 0 to 4, where 4 denotes those options that are most manager-advantageous. Panel B reports CARs for a subsample of firms whose option index is equal to 0, 1, or 2, and a subsample of firms whose option index is equal to 3 or 4. * Significance at the 10% level. Significance at the 5% level. Significance at the 1% level. our threshold value to divide the options into subsamples of more or less managerially advantageous options. Those options with scores of 2 or less are viewed as less managerially advantageous while those with scores of 3 or more are classified as more managerially advantageous.

18 20 STEPHEN P. FERRIS ET AL. In Panel B of Table 5, we separately estimate the announcement period returns across the two subsamples. For those options least favorable to managers, we observe statistically significant positive announcement period returns. This is consistent with a positive market response to options perceived to be designed from an optimal contracting perspective. The lack of statistical significance of the announcement period returns for those options that are more managerially advantageous probably reflects the market s uncertainty regarding their ability to adequately incent management and consequently impact firm performance. These options have characteristics that are suggestive of managerial influence over compensation design and thus are less likely to stimulate increased performance by managers. These results provide an initial suggestion that the market does not uniformly view the granting of executive stock options as a favorable development. Rather, the market s response is conditional upon the characteristics and terms of the option grant itself Multivariate Analysis of Announcement Period Returns In this section, we use multivariate regression analysis to further examine the ability of the option characteristics index to contribute to an explanation of the cross-sectional variability in the daily returns surrounding executive stock-option grant announcements. The dependent variable in our regression analysis is the announcement period return. Our primary independent variable of interest is the manager-advantageous executive option dummy that equals one if the options characteristics index has a value of 3 or 4, and zero otherwise. We also include a number of control variables in our model specification. Greater utilization of incentive compensation through option grants will likely have a more immediate impact on firms that are currently unprofitable or lack attractive internal investment opportunities because these firms require additional managerial effort. Hence, we include EBIT standardized by total assets and the firm s market-to-book ratio as regressors. Because larger firms are more likely to adopt and benefit from the incentive effects of stock-option linked compensation, we also include the log of total assets as a control value. Financial leverage controls for possible managerial monitoring by creditors and consequent substitution for the use of incentive compensation to reduce agency conflicts. Finally, we control for possible year and industry effects with appropriate dummies. Table 6 contains our regression results from this analysis.

19 The Design of Executive Compensation 21 Table 6. Cross Sectional Regression Analysis of the Stock Price Reaction. Explanatory Variables Dependent Variable CAR ( 1,0) CAR (0,1) CAR ( 1,1) CAR ( 5,5) CAR ( 10,10) Intercept High mgr-adv option index Log(total assets) EBIT/total assets Total debt/total assets Market-to-book value of assets Year dummies Yes Yes Yes Yes Yes Industry dummies Yes Yes Yes Yes Yes Adjusted R F-value Number of observations Notes: The sample consists of 411 option grant announcements made by non-financial Japanese firms from 1997 to The dependent variable, CAR( t, t), is a cumulative abnormal return surrounding option grant announcement dates. We compute abnormal daily returns using the market model. We estimate the market model by using 200 trading days of return data ending 21 days before the option grant announcement. We use PACAP equally weighted market returns as the benchmark. The manager-advantageous option dummy is equal to one if the manager-advantageous option index is equal to 3 or 4, otherwise the dummy is equal to zero. The option index is created as follows. We create four dummy variables, where the first dummy variable is equal to one if the option s time to exercise is below the sample median, a second dummy variable is equal to one if the option s duration to expiration is above the sample median, a third dummy variable is equal to one if the option value relative to fixed compensation is below the sample median, and a fourth dummy variable is equal to one if the option s moneyness is above the sample median. For each executive option, the values of these four dummy variables are added up to create a manager-advantageous option index from 0 to 4, where 4 denotes those options that are most manager-advantageous. All other explanatory variables are measured at the fiscal year-end immediately preceding the option grantannouncement date. Parameter estimates using the OLS regression method are presented. Significance at the 10% level. Significance at the 5% level. Significance at the 1% level. We observe in Table 6 that the manager-advantageous option dummy is inversely related to the announcement period return. This result confirms our earlier results that the market reacts differently to executive stock options, with the response depending on the option s characteristics. Executive options that are highly advantageous to managers have a negative effect on announcement period returns. This suggests that the market doubts the ability of such options to incent mangers to undertake

20 22 STEPHEN P. FERRIS ET AL. value maximization activities. We find that this inverse relation is robust to a variety of announcement period return estimation windows. 10 Finally, we observe that profitable firms and low market-to-book firms enjoy larger announcement returns. Although such a result might initially appear surprising for profitable firms, it is consistent with firms adopting executive options to mitigate the agency costs associated with the free cash resulting from their high profitability. Firms with few growth opportunities might benefit from adopting executive options to motivate their managers to more aggressively seek value-enhancing opportunities. 4. CONCLUSION Bebchuk et al. (2002) and Bebchuk and Fried (2004) contend that the exercise of managerial power, rather than optimal contracting, explains executive compensation design. They argue that firms with weak internal governance have unchecked, and thus powerful, managers who are able to grant themselves compensation that serve themselves more than their shareholders. Therefore, executive compensation is likely to be part of the agency problem rather than a solution to it. In this study, we test the managerial power thesis of Bebchuk et al. (2002) and Bebchuk and Fried (2004). Specifically, we test to see if firms with weak internal governance enable managers to grant themselves executive stock options that are manager-advantageous. In our study, we use Japan s experience with executive stock options. Little is known about executive stock options outside the U.S., so a study of Japanese stock options contributes to the growing literature on international agency costs. Also, Japanese data offer us the ability to consider governance devices that are not pervasive in the U.S. In particular, Japan is a bank-centered financial system, with an economy where business groups (i.e., the keiretsu) are prevalent. Therefore, we are able to test the potential roles of banks and business group affiliation in influencing compensation design. Using a sample of 466 executive stock-option grants, we first identify executive stock options that are manager-advantageous. We consider an executive option to be manager-advantageous if the option can be immediately exercised, has a long duration, is small relative to fixed compensation, and is closer to being in-the-money. We then identify various internal governance characteristics to see if firms with better governance are able to mitigate managers ability to grant themselves self-serving