Homogeneity and Heterogeneity in How Institutional Investors Perceive Corporate and Securities Regulations

We use an in-depth survey of institutional investors investing in Japan to reveal the homogeneity and heterogeneity of their views on corporate governance regulations. Their opinions exhibited high homogeneity in favoring legislative intervention in corporate management but nuanced heterogeneity with respect to the degree of intervention that they regarded as desirable. A certain cluster of investors, to which investment trusts and advisors were more likely to belong, tended to prefer stronger legal intervention; these investors favored strict tender-offer rules, and they more clearly supported intervention in the composition of boards and the pursuit of executive liability. These reactions may have been motivated by concern for the fact that a certain class of shareholders, particularly banks and insurance companies, was vulnerable to pressure by management, making it hard for corporations functioning autonomously to maximize shareholder values. The survey results suggest that shareholder composition may affect the necessity and effectiveness of legal interventions.


Introduction
Despite decades of research on the role of institutional investors in corporate governance, we know surprisingly little about what they think of the relevant corporate and securities regulations. Although these investors craft their strategies of engagement and investment on the basis of their perceptions of legal rules, most studies have neglected to analyze their perceptions of the legal environment. This paper fills the gap by analyzing an in-depth survey of institutional investors in the Japanese stock market. How investors evaluate corporate and securities laws will inform the work of corporate governance researchers as they examine the social desirability of these laws. Our goal is to investigate the patterns of the views of institutional investors about the relevant regulations, including how their views differ from one another. To this end, the present paper uses both hypothesis testing and exploratory approaches.
First, we develop and test a hypothesis that how investors view corporate governance regulations differs in accordance with whether they are sensitive to pressure from corporate management. Classic and influential studies have shown that investors who have both business-transaction and investment-transaction relationships with investees are sensitive to pressure from corporate management because these investors receive benefits from their business transactions with investees. 1 Banks and insurance companies fall into this category. In contrast, investors such as investment trusts and advisors resist such pressure because they have only an investmenttransaction relationship with the investees, not also a business-transaction relationship. The difference in sensitivity to pressure from management motivates different strategies. On the basis of this finding, we hypothesize that sensitivity to pressure from management, which depends on business type, also affects which legal rules investors prefer.
Second, we use cluster analysis to explore whether the views of institutional investors are homogeneous or heterogeneous enough to be clustered into multiple groups by looking only at their views without considering any other defined or observable categories such as business type and investment amount. If the views of investors may be clustered into multiple groups, we consider what patterns these groups exhibit. To get an accurate picture of the perceptions of investors, it is not enough to look at the average opinions or the opinions with respect to other observable categories. In amending corporate and securities laws, legislators need to know to what extent the opinions of investors are homogeneous and what heterogeneity exists among clusters of investors.
We use a survey of institutional investors in Japan's stock market that, to the best of our knowledge, is the only in-depth survey ever conducted of institutional investors that focuses on their perceptions of the legal aspects of corporate governance. The Institute of Social Science at the University of Tokyo (ISSUT), with which one of the authors of this paper (Tanaka) is affiliated, conducted the survey and reported the aggregate results in 2012. 2 Although the university has provided the survey dataset to researchers, 3 none has yet used this dataset to perform a detailed analysis of the preferences of institutional investors. We take full advantage of this rare and valuable dataset. As we shall explain, although Japan's stock market differs from the stock markets of other countries, it also has many similarities. Analysis of institutional investors in Japan will thus also tell us something about investors in other countries.
The results show that the opinions of respondents were highly homogeneous in favoring a certain degree of legislative intervention in corporate management but exhibited nuanced heterogeneity with respect to the degree of intervention that they regarded as desirable. Their responses were consistent with their observable behavior, such as how they voted at shareholder meetings. When differentiated by business type, the results showed that the views of banks and insurance companies, particularly regarding measures to defend against takeovers, were heavily influenced by their business-transaction relationships with investees. Moreover, the cluster of investors to which investment trusts and advisors were more likely to belong tended to prefer stronger legal intervention-stricter tender-offer rules and stronger intervention in the composition of the board and pursuit of executive liability-than investors in the other cluster did. These reactions may be motivated by the fact that a particular class of shareholders, especially banks and insurance companies, are vulnerable to management pressure, making it harder for an autonomous corporation to maximize shareholder values. The results suggest that shareholder composition may affect whether legal intervention is necessary and effective.
The outline of the survey is as follows. Our sample consists of 88 institutional investors in Japan, including local subsidiaries of foreign investors. Questions consist mainly of four parts: questions about rules regarding (1) corporate controls, (2) takeover defenses, (3) independent directors, and (4) shareholder derivative suits. With respect to (1) corporate controls, focusing on tender offers in Japan, a buyer's duty to launch a tender offer applies only to off-market transactions that acquire more than one third of the total voting rights of a target. In other words, there are basically no restrictions on acquiring controlling shares of a target company through on-market transactions. Partial tender offers are also generally allowed except when the buyer would own two thirds or more of the total voting rights as a result of the tender offer. However, some commentators suggest introducing the European-style rules that are applied to both on-market and off-market transactions according to which a buyer must launch a public tender offer to acquire all shares after having obtained what is considered control of a target (30% of the total voting rights). 4 In light of this institutional background, the respondents were asked to indicate their degree of support for the European-style rules.
Like the United States, Japan has allowed companies to adopt poison pills and other defensive measures. Regarding (2) takeover defenses, the respondents were asked to indicate the degree of importance of specific factors, such as the involvement of independent directors and the relationship with a target, when deciding whether to support a takeover defense. They were also asked to indicate their degree of support for defenses of particular purposes, such as preventing a coercive takeover and protecting the interests of stakeholders. Regarding the appointment of (3) independent directors in public companies, Japan introduced a mandatory rule in March 2021; from 2014 to March 2021, it had used a comply-or-explain rule. The respondents were asked to indicate their degree of support for a mandatory rule and for the comply-or-explain rule. Finally, regarding (4) shareholder derivative suits, ever since the government reformed the derivative suit system in 1993, shareholders have often used derivative suits, and most Japanese listed companies now purchase liability insurance for directors and officers. 5 The respondents were asked to indicate the degree of their agreement or disagreement with statements regarding the merits and demerits of derivative suits.
The data showed that high homogeneity existed in the respondents' views of corporate governance regulations. They preferred some degree of legal intervention but not extreme intervention. Most of the investors did not support European-style strict tender offers, and they seemed to prefer that management respond with defensive measures to resolve the problems of tender offers, such as the problem of coordinating the response of shareholders. They considered both the need for defensive measures and the appropriateness of triggering processes in approving defensive measures, and they supported a wide range of defensive measures. Most respondents also supported legal intervention in the appointment of independent directors. But the most popular option was the comply-or-explain rule, not the mandatory rule. Regarding shareholder derivative suits, most investors highly valued the disciplinary effects on management and thought that there would be few adverse effects, so that they seemed to be satisfied with the existing rule.
Despite the high homogeneity in the views of the investors, the data also showed nuanced heterogeneity. In approving defensive measures, banks and insurance companies tended to regard it as necessary for maintaining the transaction relationships with investees, and these respondents also tended to be generous about defensive measures to protect stakeholders other than shareholders. Domestic investors tended to be wary of abusive takeovers, such as greenmail, and to support defensive measures. On the other hand, foreign investors tended to have neutral or negative attitudes toward defensive measures, and they were less concerned about the rights of minority shareholders.
The cluster analysis further revealed the heterogeneity of the views of investors. By classifying the respondents solely on the basis of their views, without regard to such investor characteristics as business type or capital origin, we were able to classify investors into two groups, a majority group and a minority group. We found that members of the minority group-the group to which investment trusts and investors with large investment amounts tended to belong-preferred stronger legal interventions than other investors did. Investors in the minority group favored the Europeanstyle strict rules for tender offers and more clearly supported the rules for independent directors and shareholder derivative suits. They may have thought that because certain shareholders, particularly banks and insurance companies, are vulnerable to management pressure, maximizing shareholder values is difficult when corporations are autonomous.
An important policy question is whether the survey responses are consistent with the actual behavior of investors. At the time of this survey, the survey responses were consistent with the behavior of institutional investors as manifested by their votes at shareholder meetings. Since then, however, the voting and engagement behavior of shareholders has rapidly changed. What was the minority view in the survey has apparently now become the majority view, perhaps because of changes in the shareholder composition of listed companies as well as changes in the views of investors. The shareholder composition of listed companies has indeed rapidly changed over the past decade: the shareholding ratios of investment trusts and advisors and foreign investors have increased, whereas those of banks and insurance companies have decreased.
This paper contributes to three strands of literature-the preferences of investors, investor heterogeneity, and control transactions-in several ways.
First, although recent studies have used surveys of institutional investors to analyze their preferences regarding various aspects of investment, such as capital structure, 6 private engagement with management, 7 and corporate social responsibility, 8 no study has analyzed the preferences of institutional investors regarding the relevant regulations despite the significance of these regulations. This paper analyzes the first broad-scale survey of institutional investors on the legal aspects of corporate governance.
Second, although researchers have investigated the heterogeneity of investors, 9 no study has examined how investors' perceptions of legal rules vary with respect to investor categories. This paper examines whether one of the influential classifications, i.e., classification by sensitivity to pressure from management, affects how investors perceive legal rules. We also apply cluster analysis to the views of institutional investors without considering investor categories in order to discover what kinds of homogeneity and heterogeneity they exhibit.
Finally, it has been argued with respect to control transactions that the optimal solution to the problem of coordinating the shareholders of target companies 10 will differ depending on the characteristics of jurisdictions. 11 Although Europe has used a mandatory bid rule to address this issue, the United States has empowered the management of target companies, and many commentators have examined the desirability of such measures in these jurisdictions. 12 This study uses many detailed questions about the mandatory bid rule and defensive measures in order to learn what solutions institutional investors regard as desirable in the Japanese market.
Section 2 of this paper describes the survey design. Section 3 reports and analyzes the results in terms of the business type and capital origin of the investors. Section 4 investigates the results of the cluster analysis. Section 5 examines whether the answers of respondents are consistent with their actual behavior. Finally, Section 6 offers conclusions and discusses the implications and limitations of this paper.

Survey Design
The ISSUT conducted the survey, and one of the authors of the present paper (Tanaka) designed it and drafted questionnaire items. From January 2012 to March 2012, the survey company Central Research Services mailed questionnaires to 377 institutional investors who were registered with the supervisory authority in Japan, 13 including local subsidiaries of foreign investors. To focus on institutional investors interested in investing in Japanese companies, the ISSUT asked these investors to respond only if they were investing in the stocks of Japanese companies or were planning to do so. 14 It also asked these investors to give a person with experience in shareholder voting the task of answering the questionnaire, if possible. The ISSUT received 88 responses, a response rate of 23.3%-substantially higher than the response rates for other surveys of institutional investors, such as the surveys by McCahery et al., with a response rate of 4.3%, 15 and Brown et al., with a response rate of 16.1%. 16 Table 1 summarizes the characteristics of the 88 respondents. First, regarding business type, the survey found that 32 respondents (36.4%) worked for banks (excluding trust banks), seven (8%) for life insurance companies, and nine (10.2%) for non-life insurance companies. For the sake of brevity, we call these investors 'banks and insurance companies'; this group accounted for 54.6% of all respondents. Also, 36 respondents (40.9%) worked for investment trusts and advisors, and three (3.4%) worked for trust banks. We call these investors 'investment trusts and advisors'; this group accounted for 44.3% of all respondents.
With regard to country of origin, 87.5% of the organizations were domestic investors, and 11.4% were foreign investors. With regard to the scope of assets under 12 See Bergström et al. (1997), Gilson (2000), Lipton and Rowe (2002), Wachter (2003), and Enriques and Gatti (2015). 13 In our data, investment trusts were limited to members of the Investment Trusts Association, Japan. Investment advisors were limited to members of the former Japan Securities Investment Advisers Association that were also discretionary investment firms. 14 Although this survey includes questionnaire items other than the items about corporate and securities regulations, this paper focuses on items about these regulations because our research interests lie in the views of investors regarding this aspect. 15 See McCahery et al. (2016). 16 See Brown et al. (2019). management, about half of investors had less than 50 billion JPY, and the other half had more than 50 billion JPY. In 2012, 50 billion JPY was about 487 million EUR. 17 Also, 14.8% of the respondents had more than 1 trillion JPY (10 billion EUR). Finally, in the case of 29.5% of the respondents, the same department was in charge of both investment and voting; in the case of 42%, different departments were in charge of these two operations; in the case of 27.3%, it depended or varied. 18 With regard to experience in asset management, 54.5% of respondents had less than 10 years of experience, and 44.3% had more than 10 years of experience. With regard to experience in shareholder voting, 79.5% had experience, and 20.5% had no experience. With regard to the positions of the respondents, a plurality (34.1%) were department heads. The second most common position (23.9%) was section head.

Corporate Control Regulations
In Japan, the takeover market has been active for a long time, and tender offers have been made constantly. Figure 1, giving the number of tender offers in listed companies from 2006 to 2019, shows that the number has been slightly more than 40 cases every year since 2015. As in many other countries, there has always been debate in Japan about what corporate-control regulations, including tender-offer legislation, are desirable. The survey asked respondents the extent of their support for four possible reforms of corporate control. The respondents were asked to use a five-point scale (1: oppose, 2: tend to oppose, 3: neither support nor oppose, 4: tend to support, 5: support) to indicate their degree of support for or opposition to each change in the control regulations of listed companies. 19 Table 2 reports the views of respondents regarding these corporate-control regulations.
The first questionnaire item: 'When a listed company issues a large number of shares that change control of the company, it must obtain the approval of a shareholder meeting (Item 1).' The mean score is approximately 4 ('tend to support'), and the percentage of scores that are 4 or 5 is 83%. These results mean that respondents were generally supportive of the proposal. When the survey was conducted in 2012, listed companies were allowed to issue a large number of shares that would cause 17 The exchange-rate information is from the website of the Mitsubishi UFJ Research and Consulting. See http:// www. murc-kawas esouba. jp/ fx/ past_ 3month. php (accessed 29 September 2022). 18 According to a 2015 survey of U.S. institutional investors, portfolio managers were involved in only 25% or fewer of the voting decisions of most respondents. In both the U.S. and Japan, it does not seem that those in an organization with the most detailed knowledge of portfolio companies are involved in voting decisions. See RR Donnelly, Equilar, and the Rock Center for Corporate Governance at Stanford University (2015). 19 We translated the questionnaire using natural English expressions rather than literal translations, while preserving the meaning of the original Japanese text. Also, in the original questionnaire, 1 corresponded to 'support' and 5 corresponded to 'oppose', and the numbers were in reverse order. In this paper, we have renumbered the numbers so that the higher the number, the more positive the evaluation. The same applies to questions in the other domains. a change of control without the approval of a shareholder meeting if they obtained a positive letter of opinion from an independent third party or if they had an urgent need to do so. 20 Consistent with the opinions of investors, the Japanese Companies Act was reformed in 2014; as a result, companies are now required to obtain the approval of a shareholder meeting in order to issue shares that would change their control under certain conditions. 21 The second questionnaire item: 'When a listed company squeezes out minority shareholders, it must obtain the approval of a shareholder meeting with a greater number of affirmative votes than is required for a special approval (Item 2).' The mean score is 3.5, which differs from 3 ('neither support nor oppose') by a statistically significant amount. The percentage of scores that are 4 or 5 is 51%, and the percentage of scores that are 1 or 2 is 9%. In Japanese law, a listed company can squeeze out minority shareholders if it obtains approval by the special resolution of a shareholder meeting, a law that has not been changed since the time of the survey. 22 In the special resolution of a shareholder meeting, (1) the total votes of the shareholders in attendance must reach a majority of the votes that can be exercised, and (2) two thirds or more of the votes of the shareholders in attendance must be affirmative unless the certificate of incorporation provides otherwise. 23 In addition,

Table 2
Views of investors on corporate control regulations The phrase 'Banks and ins. co.' means 'banks and insurance companies', and the phrase 'Invt. tr. and adv.' means 'investment trusts and advisors'. The asterisks to the right of the mean scores indicate the results of a t test of the null hypothesis that each mean score is equal to 3. The * and ** indicate statistical significance at the 5% and 1% levels, respectively. 'Diff.' indicates the differences in scores between the banks and insurance companies group and the investment trusts and advisors group and the differences in scores between domestic and foreign investors. The asterisks to the right of these differences indicate the results of a t test of the null hypothesis that the underlying population means are the same. Welch's t test statistic was used when equal variances among populations could not be assumed. The * and ** mean the same as above. 'ES' indicates the effect sizes of differences in scores between the two groups. a controlling shareholder of a company can squeeze out minority shareholders even without such a special resolution if the controlling shareholder owns 90% or more of the voting rights and obtains the approval of the board of directors. 24 The results of the survey imply that the respondents tended to support strengthening these squeezeout requirements, presumably because they wanted more protection as minority shareholders.
The third and fourth questionnaire items propose the rules of a European-style tender offer: 'It is prohibited to acquire control of a listed company solely by purchasing shares in the market without making a tender offer (Item 3).' And: 'When purchasing shares that acquire control of a listed company, a tender offer must be made for all shares (Item 4).' The mean score for Item 3 is 3. The percentage of scores that are 4 or 5 and the percentage of scores that are 1 or 2 are both about 30%. The mean score for Item 4 is also 3. The percentage of scores that are 4 or 5 and the percentage of scores that are 1 or 2 are both about 25%.
These results indicate that the views of the respondents on the European-style tender offer were split and, on average, neutral. As explained in the Introduction, Japan imposes almost no restrictions on acquiring controlling shares of a target company through on-market transactions. Although some commentators have proposed European-style rules, only 30% or so of the respondents supported these rules. Quite a few institutional investors may have had the view that, overall, it would be more desirable to have more relaxed regulations in order to encourage as many control transactions as possible 25 than to have stricter regulations in order to resolve potential problems caused by control transactions, such as difficulty coordinating the actions of shareholders of target companies. 26 Alternatively, institutional investors may have thought that these problems could be resolved by allowing target companies to adopt defensive measures. This possibility will be discussed in the next section. Table 2 also reports the answers of respondents in relation to their business type. There were small or no differences in scores between the banks and insurance companies and the investment trusts and advisors. This fact implies that differences in sensitivity to pressure from management do not affect the views of investors on the corporate-control regulations discussed above, particularly the rules governing a European-style tender offer. On the one hand, introducing European-style mandatory tender-offer rules may restrict hostile acquisitions, giving management reason to favor these rules. On the other hand, in light of the fact that the vast majority of control transactions are friendly transactions, management also has reason to support the current, less strict Japanese takeover rules in order to conduct these friendly transactions as easily as possible. On balance, it is unclear whether banks and insurance companies, which are more sensitive to pressure from management, are more supportive or less supportive of European-style rules than investment trusts and advisors, which are less sensitive to such pressure. So it is unsurprising that we find no statistically significant difference in the answers of these business types. Table 2 reports the breakdown of the domestic and foreign institutional investors. We observed a notable difference in their opinions about squeezing out minority shareholders (Item 2). The domestic investors tended to support strengthening squeeze-out requirements, but the attitude of the foreign investors was neutral. The difference is statistically significant, and the effect size is medium. 27

Background Information
Hostile takeovers have occurred often in Japan, albeit in small numbers. In the 2000s, there were several well-known cases, including hostile takeovers of Nippon Broadcasting System by Livedoor in 2005 and of Bull-Dog Sauce by Steel Partners in 2007. After these cases, many Japanese companies adopted defensive measures against the threat of hostile takeovers. In 2008, the number of companies adopting defensive measures peaked at 569, about 24% of listed companies. 28 Since then, however, institutional investors have increasingly opposed introducing and continuing defensive measures. By 2019, the number had gradually decreased to 377, about 10% of listed companies. Cases involving companies with household names have recently been attracting attention. Examples include hostile takeovers of Descente by Itochu in 2019, Tokyo Rope by Nippon Steel in 2020, and Shinsei Bank by SBI Holdings in 2021. In recent years, there have also been a series of judicial decisions regarding the legality of defenses against takeover. 29 The survey was conducted in 2012, when the number of companies adopting defensive measures was declining.
Most of the defensive measures adopted by Japanese companies are pre-warning rights plans, which attempt to make the U.S.-style rights plan or poison pill feasible under the Japanese Companies Act. In a pre-warning rights plan, a potential target company announces the following procedures in advance: (1) an acquirer must provide certain information, including post-acquisition business plans, and must provide time for the board of the target company to consider the acquirer's proposal and, if necessary, to present alternative plans to shareholders. (2) If the acquirer attempts an acquisition without following such procedures, the target company will 27 Cohen's d formula was used. The conventional interpretation is that 0.2 is small, 0.5 is medium, 0.8 is large, and 1.2 is very large. 28  issue share options (stock acquisition rights) to its shareholders with discriminatory conditions. 30 This defensive measure is called a pre-warning rights plan because when it introduces this defense, a target company does not issue share options but merely warns of the possibility that share options will be issued in the future. The conditions of the share options are discriminatory in the sense that only shareholders other than the acquirer may exercise the options or in the sense that the target company may acquire share options in exchange for common shares from shareholders other than the acquirer. The acquirer can receive share options as long as it is a shareholder, but it cannot acquire shares. If other shareholders receive shares under a pre-warning rights plan, the greater number of shares that they now hold reduces the shareholding ratio of the acquirer, making acquisition more difficult.
The rights plans of Japan and the United States differ. When a company introduces a rights plan in the United States, it allots share options to all shareholders, and if shareholders sell their shares, the share options are transferred along with the shares. But in Japan, the Companies Act does not allow an arrangement in which share options are transferred along with shares. Therefore, if a Japanese company allots share options to all shareholders when it introduces a rights plan, shareholders at the time of a hostile takeover may be different from the holders of share options. This means that if the share options are exercised, shareholders other than the acquirer may also suffer economic losses. For this reason, Japanese companies use pre-warning rights plans according to which companies do not allot share options to their shareholders unless and until a hostile takeover actually occurs.
The purpose of pre-warning rights plans is not to allow the board of directors of target companies to act to prevent hostile takeovers at their discretion. The purpose is to compel an acquirer to provide relevant information and time to shareholders to help them decide whether to accept an acquisition. If the acquirer provides sufficient information and time, target companies are not supposed to issue share options, and it will be up to shareholders to decide whether to accept the acquisition. In a pre-warning rights plan, an independent committee consisting of outside directors, auditors or experts is usually involved in the implementation process to prevent arbitral implementation by the board of directors. When introducing the plan, the company usually obtains the approval of shareholders, and the effective period is 1-3 years. When the period expires, the company must again obtain the approval of shareholders.

Factors to Consider When Approving Defensive Measures
Some investors favor defensive measures, and others oppose them, but it has been unclear what factors the investors of either group consider when doing so. So, in this survey, the respondents were asked to indicate the degree of importance of eight specific factors when their institution decides whether to support a takeover defense proposal by an investee company. Degree of importance is measured on a 5-point scale (1: very unimportant, 2: somewhat unimportant, 3: neither important nor unimportant, 4: somewhat important, 5: very important). Table 3 shows the results. For six factors-clarity of trigger conditions, involvement of independent directors in triggering, shareholder involvement in triggering, periodic approval by shareholders, an imminent threat to corporate value, and past corporate performance (Items 1, 2, 3, 5, 6, 7) -the mean scores range from 3.5 to 4.1. Each score differs from 3 ('neither important nor unimportant') by a statistically significant amount, and the percentage of those who answered 4 ('somewhat important') or 5 ('important') exceeded the majority for each item. So the respondents tended to regard these six factors as important. In contrast, for two factorslimited use of a defensive measure depending on acquisition methods (Item 4) and the necessity of maintaining relationships with an investee company (Item 8)-the mean scores are 3.2 and 3, respectively, and the percentage of those who answered 4 or 5 was less than half. That is, respondents had no strong tendency to weigh these factors.
The above results indicate the trends of institutional investors as a whole. Looking at the results by business type, we find that the opinions of respondents were sharply divided with respect to the two items. First, regarding the necessity of maintaining relationships with an investee company (Item 8), the mean score of the banks and insurance companies is 3.8 and that of the investment trusts and advisors is 2.1. The former group tended to weigh this factor; the latter group tended not to weigh it. The effect size is 1.5, which is very large. Clearly, this result is consistent with the management-pressure hypothesis, which predicts that banks and insurance companies are inclined to favor the proposals of management about the companies with which they want to retain business-transaction relationships. Second, regarding the involvement of independent directors in triggering (Item 2), the mean score of the banks and insurance companies is 3.5 and that of the investment trusts and advisors is 4.1. The latter group showed a stronger tendency than the former to weigh the role of independent directors. The effect size is 0.5, which is medium.
There are large but not statistically significant differences between domestic and foreign investors with respect to the threat of an acquisition that damages shareholder value (Item 6) and maintaining relationships with investees (Item 8). The domestic investors may have been concerned about abusive takeover tactics because of a well-known 2007 court case in which the Tokyo High Court regarded the foreign investor Steel Partners as a greenmailer. 31 Also, since foreign investors rarely have business-transaction relationships with investees, it is natural that they treated such relationships as unimportant in their answers (yielding a score of 2.4).

Attitudes Toward Different Purposes of Defensive Measures
In the previous section, we learned what factors institutional investors weigh when approving a defensive measure for any purpose. But what purposes of defense measures do they support? The survey asked respondents to indicate their degree of  support for or opposition to ten possible purposes for the sake of which the board of directors in a listed company takes a defensive measure. A five-point scale was used (1: oppose, 2: tend to oppose, 3: neither support nor oppose, 4: tend to support, 5: support). Table 4 shows the results. For two purposes-preventing an acquisition that exploits profits from the company and preventing a coercive acquisition, such as a two-step tender offer (Items 2 and 3)-the mean score is 4 ('tend to support') or higher, and most respondents answered 4 or 5 ('support'). This result implies that institutional investors do not necessarily oppose defenses for all purposes. For three purposes-preventing the purchase of shares that aims to force the company to buy them back at a high premium, securing information and time for shareholders to make a decision, and preventing an acquisition that would harm the interests of stakeholders like employees and business partners (Items 1, 5 and 10)-the mean score is 3.5 or higher, and most respondents answered 4 or 5. However, with respect to Item 10, there was an enormous disagreement between the banks and insurance companies and the investment trusts and advisors, as we discuss below.
For three purposes-negotiating with an acquirer to improve acquisition terms, preventing an acquisition that is followed by the sale of the company's assets and businesses to achieve high dividends, and preventing the company from incurring excessive debt due to a leveraged buyout (Items 6, 8 and 9)-the mean score is higher than 3. However, the proportion of supportive respondents does not reach a majority. The purpose described in Item 6 is often the purpose of the defensive measures of U.S. companies. The result implies that such a purpose is only moderately popular among investors in Japan. Moreover, although the Tokyo High Court held that defensive measures for the purposes of Items 1, 2, 8 and 9 are permissible, 32 the investment trusts and advisors and the foreign investors were not inclined to support Items 8 and 9.
Finally, for two purposes-preventing acquisition of control by purchasing shares in the market without a tender offer and giving the board time to offer alternative options to shareholders (Items 4 and 7)-the average score is 3. In some situations, acquisitions can be coercive when acquirers buy shares through on-market transactions. 33 In the United States, defensive measures are allowed to some extent against such on-market purchase of shares in order to negate this coercive effect. The result implies that investors in Japan do not necessarily support defensive measures against gaining control through on-market transactions. 34 They may be concerned about how such defensive measures reduce the chances of acquisition.
In Europe, the acquisition of control of listed companies must be carried out by tender offers rather than by on-market transactions of shares. As noted in the previous section, Japanese investors do not necessarily support the European-style tenderoffer rule either. These results therefore suggest that institutional investors in Japan 32 Tokyo High Court order of July 9, 2007, Shoji Homu No. 1806, p 40. 33 See Bebchuk (1988). 34 In Item 3, the respondents supported a defensive measure against a coercive acquisition. This item assumes a coercive takeover by a tender offer, not a coercive takeover by on-market transactions. When we look at the results by business type, we find that the banks and insurance companies were more supportive of most of the defensive measures than the investment trusts and advisors were. Among these items, there are statistically significant differences in Items 7 and 10. Regarding the purpose of giving the board time to provide shareholders with other options (Item 7), the banks and insurance companies have a mean score of 3.3, but the investment trusts and advisors have a mean score of 2.7. The effect size is 0.7, which is medium to large. These results indicate that the investment trusts and advisors had less trust in the discretion of the board of directors than the banks and insurance companies did. With respect to the purpose of protecting the interests of stakeholders (Item 10), the banks and insurance companies have a mean score of 3.9, and the investment trusts and advisors have a mean score of 2.9. The effect size is 0.9, which is large. In general, banks and insurance companies have relationships with investee companies both as shareholders and as creditors, but investment trusts and advisors only have a relationship as shareholders. This difference seems to explain the difference in support for a defensive measure to protect the interests of stakeholders.
The domestic investors tended to support a wide range of defensive measures, whereas the foreign investors tended to be either neutral or opposed. We observed somewhat large and statistically significant differences with respect to whether to give boards discretion to look for better options and prevent abusive takeover methods like greenmail (Items 7, 8 and 9). As mentioned in Sect. 3.2.2, domestic institutional investors may have been especially wary of abusive takeovers since the Steel Partners case. At the time, the decision of the Tokyo High Court had a significant influence.

Independent Director Regulations
In the United States and Europe, directors of listed companies are usually outside directors who are independent of those companies. In contrast, most directors of Japanese listed companies are employees of those companies who have been promoted to the position of director. With the recent changes in the environment of corporate governance, such as an increase in the shareholdings of institutional investors, there have been more and more calls for appointing independent directors in Japan. When this survey was conducted in 2012, the Legislative Council Corporate Law Subcommittee was discussing the pros and cons of mandating the appointment of independent directors. Despite such discussions, the revised Companies Act of 2014 did not require appointment of independent directors. Instead, it required listed companies that lacked an independent director to explain 'why it is not appropriate to have an independent director' (the comply-or-explain rule). 35 However, in 2015, the Corporate Governance Code of the Tokyo Stock Exchange began to be applied to its listed companies; the Code encourages the appointment of two or more independent directors under the comply-or-explain rule. 36 With these rule revisions, the number of listed companies appointing independent directors has continued to increase. In 2014, only 21.5% of 1st Section companies of the Tokyo Stock Exchange had two or more independent directors. In 2020, 98.5% of such companies did so. 37 The latest revised Companies Act of 2021 requires listed companies to have an independent director. 38 What is the rationale for requiring the appointment of independent directors? It may be argued that since independent directors have both good features (independence from management) and bad features (lack of information and incentives), each of which may vary depending on the attributes of particular companies, each company should be free to decide for itself whether and how often to appoint independent directors. However, since an overwhelming majority of Japanese boards of directors have been insiders, the selection of directors may be very much influenced by the interests of managers at the expense of the interests of shareholders. 39 Institutional investors may therefore want some kind of regulation that encourages or requires firms to appoint independent directors.
On this issue, the survey asked respondents to indicate their degree of support for or opposition to three possible regulations of independent directors-mandatory, comply-or-explain, and incentive-based (Items 1, 2 and 3)-on a five-point scale (1: oppose, 2: tend to oppose, 3: neither support nor oppose, 4: tend to support, 5: support). In the survey, the incentive-based rule is a rule that encourages listed companies to appoint independent directors by giving some incentives, such as allowing a wide range of defensive measures use of which is contingent upon the approval of independent directors. Table 5 reports the results. The most popular rule was the comply-or-explain rule, with a mean score of 3.9 and a percentage of scores that are 4 or 5 of 68%. The second most popular rule was the mandatory rule, with a mean score of 3.6 and a percentage of scores that are 4 or 5 of 55%. The difference in mean scores between these two rules is statistically significant with a 5% significance level. The mean score for the incentive-based rule is 3.5, which differs from 3 to a statistically significant degree; but less than half of the respondents supported it. So it seems that at least in 2012, when the survey was conducted, most Japanese institutional investors regarded the ability of firms to freely decide whether to appoint independent 36 See Goto (2018) for details. 37 Data are taken from the website of the Japan Exchange Group. See https:// www. jpx. co. jp/ engli sh/ listi ng/ others/ ind-execu tive/ index. html (accessed 29 September 2022). 38 Companies Act, Article 327-2. 39 Empirical studies of Japanese firms suggest that their boards of directors are indeed structured in ways that are consistent with the interests of managers. See Saito (2011) and Miyajima and Ogawa (2012). In contrast, studies of U.S. firms suggest that boards of directors are generally structured in an optimal way. See Coles et al. (2008) and Linck et al. (2008). directors as being inconsistent with their interests and supported some degree of regulation, such as the comply-or-explain rule or even mandatory appointment.
With respect to differences based on business types, we found that the investment trusts and advisors had a stronger tendency to support the comply-or-explain rule than the banks and insurance companies did. The mean score of the banks and insurance companies is 3.8, and the percentage of 4 or 5 is 58%. In contrast, the mean score of the investment trusts and advisors is 4.2, and the percentage of 4 or 5 is 82%. The absolute value of the effect size is 0.6, which is medium. Both domestic investors and foreign investors tended to support all of the rules promoting the appointment of independent directors. The foreign investors were slightly more supportive, but we did not observe statistically significant differences.

Derivative Suit Regulations
Under the shareholder derivative suit system, if directors and officers of a company are liable for damages to the company, the shareholders can pursue the liability on behalf of the company. 40 The reason for permitting such suits is that if only a company can pursue the liability of directors and officers, there is a risk that the company will not pursue the liability because of corruption among directors and officers. Although the derivative suit system may have a beneficial disciplinary effect on directors and officers, it may also have a chilling effect on management and may encourage frivolous lawsuits. 41 With this background, the survey asked respondents to indicate the intensity of their agreement or disagreement with statements regarding the pros and cons of shareholder derivative suits. A five-point scale was used (1: strongly disagree, 2: somewhat disagree, 3: neither agree nor disagree, 4: somewhat agree, 5: strongly agree). Table 6 shows the results. First, looking at the items about the merits of derivative suits, we find that the mean score for the statement that derivative suits have a disciplinary effect on management (Item 5) is 3.9, and the percentage of respondents who chose scores of 4 (somewhat agree) or 5 (strongly agree) is 72%. That is, most respondents tended to acknowledge the disciplinary effect. In addition, the mean score for the statement that there is almost no risk that directors and officers will be liable if they appropriately follow the internal procedures of their company (Item 4) is 3.3, which differs from 3 by a statistically significant amount. These results suggest that institutional investors value the merits of derivative suits to some extent.
Next, looking at the items regarding the disadvantages of derivative suits, we find that the mean score for the statement that derivative suits are rarely filed and thus are not a particularly meaningful system (Item 6) is 2.3, and the percentage of the respondents who chose scores of 1 (strongly disagree) or 2 (somewhat disagree) is 60%. Regarding the two statements that derivative suits have a chilling effect on management (Item 1) and that derivative suits discourage potential candidates from becoming directors and officers (Item 2), the mean score is 2.7 for both, which differs from 3 by a statistically significant amount. For the statement that frivolous lawsuits can occur (Item 3), the mean score is 3. These results imply that institutional investors tend to disagree with most arguments generally regarded as setting forth the disadvantages of derivative suits. Looking at the results by business type, we find that the investment trusts and advisors had a stronger tendency than the banks and insurance companies to believe that derivative suits are beneficial and not harmful. But the difference between the two groups was small. A statistically significant difference was observed for the statement that frivolous lawsuits can occur (Item 3), and the investment trusts and advisors showed a stronger tendency to disagree with the statement. The effect size is 0.4, which is small to medium. The views of domestic and foreign investors on shareholder derivative suits were almost the same.

Homogeneity and Heterogeneity of Views of Institutional Investors
In this section, we use exploratory cluster analysis to see whether there are any patterns in the views of institutional investors on corporate governance and securities legislation. The number of respondents with no missing values in all questionnaire items was 77, and we use the data of these respondents to perform the analysis. After principal component analysis, we perform cluster analysis using the principal component scores. We also perform a probit regression to see how the characteristics of respondents affect cluster membership.

Principal Component Analysis
Our dataset has five domains: (1) attitudes toward corporate-control regulations, (2) factors to consider when approving takeover defenses, (3) attitudes toward different purposes of takeover defenses, (4) attitudes toward independent director rules, and (5) attitudes toward derivative suits. Since these five domains contain very different numbers of variables, applying cluster analysis means giving each domain a significantly different weight. 42 Therefore, we applied principal component analysis to each domain, 43 extracted a small number of principal components for each domain, and applied cluster analysis to the principal component scores thus obtained. Table 7 reports the results of principal component analysis using a correlation matrix and shows the principal components whose eigenvalues are greater than 1, which means that each component conveys more information than an original variable. 44 To make interpretation of the results more straightforward, we used the varimax rotation. 45 In the domain of (1) corporate-control regulations, we extracted two principal components, and they explained 65% of the variance in the data. We can interpret the first component as support for the European-style mandatory tender-offer rule and the second component as support for tightening the rules of stock issuance affecting corporate control and minority shareholders. Two principal components were extracted for (2) the factors to consider when approving defense measures, which explained 68% of the variance. We can interpret the first component as the appropriateness of triggering conditions and processes and the second principal component as the need for defensive measures.
Two principal components were also extracted for (3) the attitudes towards defenses for different purposes, explaining 62% of the variance. We can regard the first component as support for defensive measures to prevent the acquisition with unjust purposes or means. We can regard the second component as support for defensive measures that aim to protect the interests of companies, shareholders and stakeholders. Only one principal component was extracted for (4) the attitudes toward independent director rules, explaining 65% of the variance, and this component shows supportiveness for enhancing the appointment of independent directors. Finally, two principal components were extracted for (5) the attitudes toward derivative suits, explaining 64% of the variance. The first component shows attitudes toward the advantages and disadvantages of shareholder derivative suits, and the second component shows a tendency to believe that, in reality, shareholders rarely file derivative suits.

Clustering Tendency
Cluster analysis identifies homogeneous groups in data and assumes the existence of a cluster structure. Even if a given dataset lacks a cluster structure, cluster analysis can classify observations as if there were several homogeneous groups, leading to wrong conclusions. It is therefore necessary to test whether data have a clustering tendency before conducting cluster analysis. 46 The clustering tendency problem is the problem of deciding whether data tend to cluster into natural groups without identifying the groups themselves. 47 This paper applies the widely used Hopkins statistic and the recently developed dip-dist method to scrutinize the clustering tendency of the data of principal component scores. 48 46 See Dubes and Zeng (1987) and Adolfsson et al. (2019). 47 See Jain and Dubes (1988). 48 Principal component scores were calculated by linear regression using the principal component coefficients we obtained. 44 We do not standardize the variables because the values of all of the respondents' answers have the same range, and the variances do not differ significantly from each other. 45 Researchers often use this rotation for the sake of ease of interpretation; doing so does not affect the results of the analysis. To compute the Hopkins statistic, we sample actual data points and also generate data points that are uniformly distributed across the data space. 49 For both sets of points, we calculate the distance to the nearest neighbor in the original data set. Then we compare the sum of nearest neighbor distances of the sample of points from the original dataset with the sum of nearest neighbor distances of the uniformly generated points. If the dataset does not have a clustering tendency, both sums of nearest neighbor distances should have almost the same value. The null hypothesis is that the data are no more clustered than uniformly distributed random data.
The dip-dist method uses the dip test, which calculates a statistic called a dip, i.e., the maximum difference between the empirical distribution and the closest uniform distribution. 50 If the dip is sufficiently large, the data are sufficiently different from the closest uniform distribution, contradicting the null hypothesis that the data are generated from a unimodal distribution. The dip-dist method uses the set of pairwise Euclidean distances between datapoints as inputs for the dip test. 51 If the distance distribution has multiple modes, this implies the presence of multiple clusters.
The Hopkins statistic is 0.6323, 52 and the p value is 0.04. At a 5% significance level, we reject the null hypothesis that the data are no more clustered than uniformly distributed random data. As the Hopkins statistic gets closer to 1, data become more highly clustered. In contrast, the dip statistic for our dataset is 0.0044, and the p value is 0.99. We fail to reject the null hypothesis that the distance distribution is unimodal; this suggests that our dataset has no clustering tendency. These results of the two clustering tendency tests imply that although our dataset may have a cluster structure, its clustering tendency is low, if any. This means that we should not overemphasize the heterogeneity of investors' perceptions of corporate and securities regulation. Assuming that their views have a low degree of heterogeneity, we will consider in what aspects we find heterogeneity.

Clustering Methods and the Optimal Number of Clusters
In social science data, clusters are often not well separated, so it would be better to focus on within-cluster homogeneity than on separation. We use the k-means method because it emphasizes homogeneity rather than separation. 53 Also, to check cluster stability, it would be informative to use a supplementary method to see how much the result changes in consequence of the algorithm change. To this end, we use Ward's hierarchical clustering method. The k-means method and Ward's method 50 See Hartigan and Hartigan (1985). 51 See Kalogeratos and Likas (2012). 52 We reported the results of the Hopkins test when 20 samples were used; using other sample sizes, such as 5, 10 and 15, does not affect our conclusions. Too few samples do not represent the characteristics of the data, and too many samples do not meet the beta distribution assumptions. See Adolfsson et al. (2019). 53 See Hennig (2015).
have the same objective function, the sum of squared Euclidean distances between each sample point and its nearest cluster centroid. 54 The next step is to determine the number of clusters in our dataset. Many stopping rules (rules to stop the merging process) or indexes have been developed to determine the optimal number of clusters. Milligan and Cooper compare 30 indexes with simulated data and suggest using one or more of the better-performing indexes. 55 Following their suggestions, we use the Calinski-Harabasz (CH) index and the J-index, the top two performers. The CH index, also known as the variance ratio criterion, uses the ratio between the variance within the cluster and the variance between the clusters. 56 The J-index is a criterion for deciding whether a cluster should be divided into two subclusters that compares the within-cluster sum of squared distances between the objects and the centroid with the sum of the withincluster sum of squared distances when the cluster is optimally partitioned into two subclusters. 57 For both the k-means method and Ward's method, both the CH index and J-index show that the optimal number of clusters is 2. When the number of clusters is 2, in the k-means method, the CH index is 9.94, and the J-index is 1.17; in the Ward method, the CH index is 8.23, and the J-index is 0.81. These results indicate that the optimal number of clusters is likely to be 2, given that the dataset has a clustering tendency.
The k-means method creates two groups of 52 observations and 25 observations, and Ward's method creates two groups of 51 observations and 26 observations. In both clustering methods, the majority group is about twice as large as the minority group. For these two clustering methods, we call the majority group Cluster 1 and the minority group Cluster 2. The k-means method and Ward's method classifications match in 72 observations (93.5%) and do not match in five observations (6.5%). Table 8 shows the respondents' perceptions of corporate and securities regulations by cluster. Because the results of the k-means method and Ward's method are almost the same, we focus below only on the results of the k-means method. Regarding corporate-control regulations, there are notable differences between clusters in two items (Items 1 and 3), which are also statistically significant. With respect to the large issuance of shares that involves the transfer of company control (Item 1), both 54 See MacQueen (1967) and Ward (1963). 55 See Milligan and Cooper (1985). 56 See Calinski and Harabasz (1974). 57 See Duda et al. (2001).

Table 8
Views of investors on corporate and securities regulations by cluster   To prevent an acquisition that is followed by sale of the company's assets or businesses in order to achieve high dividends   The asterisks to the right of the mean scores indicate the results of a t test of the null hypothesis that each mean score is equal to 3. The * and ** indicate statistical significance at the 5% and 1% levels, respectively. 'Diff.' indicates the difference in scores between the two clusters. The asterisks to the right of these differences indicate the results of a t test of the null hypothesis that the underlying population means are the same. Welch's t test statistic was used when equal variances among populations could not be assumed. The * and ** mean the same as above. 'ES' shows the effect sizes of differences in scores between the two groups. Missing values were excluded from the analysis clusters supported requiring the approval of a shareholder meeting. The minority group (Cluster 2) showed a larger degree of support, and its score is close to the full score. With respect to the prohibition of control acquisition without a tender offer (Item 3), the majority group (Cluster 1) tended to oppose it, but the minority group tended to support it. The effect sizes for the differences in these Items 1 and 3 are 1.2 and 1.0, respectively, which are extremely large. With respect to the factors to consider when approving defensive measures, notable differences appear in two items (Items 6 and 8) regarding the need for defensive measures: an imminent threat of acquisition that damages firm value and the necessity of maintaining a relationship with an investee company. For both of these items, the average response of the majority group differs from 3 ('neither important nor unimportant') by a statistically significant amount, but the average response of the minority group does not. In terms of effect size, the difference in the imminent threat of acquisition is medium (0.6), and the difference in the need to maintain a relationship with an investee is large (0.8). For the other six items, there are no notable differences between the two clusters.
Regarding the supportiveness for the different purposes of defensive measures, the majority group was generally more supportive than the minority group. The majority group tended to support defensive measures for all the purposes stated in Items 1 to 10 (i.e., the average scores for all of those items differ from 3 by a statistically significant amount), while the minority group tended to be neutral about defensive measures for many of these purposes (i.e., the average scores for Items 1, 4-6 and 8-10 do not differ from 3 by a statistically significant amount). The minority group tended to be negative even about a defensive measure to give the board time to present shareholders with alternative options (Item 7). We observed statistically significant differences between the tendencies of these groups to support defensive measures for many purposes; and for some of them (Items 1, 7, 9 and 10), the effect sizes range from large to extremely large.
These results suggest that the majority group was positive about using takeover defenses against acquisitions that could harm the interests of shareholders and other stakeholders, while the minority group tended to be more concerned that defensive measures might be abused to protect management. In particular, the minority group did not support defensive measures to prevent a takeover in which the company's shares are purchased in the market without a tender offer (Item 4). But the minority group tended to support takeover regulations to prevent taking control by purchasing shares in the market (Item 3 of the corporate-control regulations). The minority group seemed to think that although acquisition through on-market transactions may cause a problem for coordination of shareholders of target companies and result in value-decreasing takeovers, 58 such a problem should be resolved by government intervention-regulations-not by defensive measures prone to managerial abuse.
Regarding regulations governing independent directors, both the majority group and the minority group tended to support all three forms of regulation: mandatory, complyor-explain and incentive-based. The minority group exhibited a more positive attitude than the majority group. The effect sizes of these differences are between medium and large. Regarding shareholder derivative suits, the majority group was undecisive in evaluating the harmful effects of a shareholder derivative suit, but the minority group tended to think that the harmful effects were small (Items 1 and 3). The effect sizes of the differences are medium, and these differences are statistically significant. Both the majority group and the minority group regarded shareholder derivative suits as having benefits, but the minority group evaluated their usefulness more highly (Items 5 and 6). The effect sizes are between small and medium, and the differences are statistically significant.
In summary, the majority group did not support the European-style tender-offer rules, placed a similar degree of importance on both the need for takeover defenses and the appropriateness of triggering processes in approving them, and was in favor of defenses for various purposes. The majority was also in favor of using independent directors and derivative suits, but to a lesser extent than the minority. In contrast, the minority tended to support European-style tender-offer rules, placed more emphasis on the triggering processes than on the need for takeover defenses in approving them, and favored only limited kinds of defenses. The minority was more strongly in favor of using independent directors and derivative suits than the majority.
On the basis of these results, we may conclude that although both groups acknowledged the need for legal intervention to discipline management and protect the interests of shareholders, the majority tended to support greater managerial discretion. The minority group seemed to regard current Japanese takeover regulation, which generally permits acquisitions by on-market transactions that may cause the coordination problem, as providing insufficient protection to shareholders, and this group wants law reforms to introduce European-style regulation. The minority group did not favor defensive measures for the purpose of solving the coordination problem (and also did not favor defensive measures for the other purposes), probably because they were concerned that defensive measures may weaken the disciplinary effect of hostile acquisitions. But the majority group seemed to think that the coordination problem and other problems of control transactions could be resolved by defensive measures without the need to revise the current takeover rules.
We may regard independent director rules as interventions in the market for independent directors. Both groups seemed to think that the number of independent directors in Japanese companies fell below the optimal number and that some form of intervention was desirable to enhance the independence of boards and increase the disciplinary effect on management. But the minority group had a stronger tendency to support such interventions. We may regard derivative suits as legal devices that complement the market mechanism of disciplining directors. If the disciplining function of markets, such as the capital market and the corporate-control market, are working well, companies should be motivated to hold directors accountable if and when they neglect their duties. But if these market mechanisms are dysfunctional, shareholder derivative suits may be necessary as an additional disciplining mechanism. Although both groups seemed to think that derivative suits are beneficial, the minority had a stronger tendency to think so.

Probit Regression for Cluster Membership
We next perform a probit regression on whether and to what extent specific characteristics of institutional investors and individual respondents increase the probability of belonging to a particular cluster. The dependent variable is a dummy variable that indicates whether an institution belongs to a minority group. If it belongs, the variable takes 1, and if it does not belong, the variable takes 0. Independent variables are organizational characteristics like business type, country of origin and assets under management; and personal characteristics of respondents like years of service in asset management and voting experience. All of these independent variables are dummy variables, each with its own reference category. Table 9 shows the coefficients and marginal effects (partial effects for the average observation) for each of the minority memberships obtained by the k-means method and Ward's method. Because the results of the two methods were almost the same, this section focuses on the results of the k-means method. We observed notable results for the items of business type, investment amount and years of service. The marginal effects in the categories of investment trusts and advisors, the investment amount of 100 billion yen to 1000 billion yen, years of service 10 years to 19 years, and years of service 20 years or more are 0.37, 0.63, 0.63 and 0.53, respectively, which have positive signs and are large. The coefficients of these items are also statistically significant. These results mean that being an investment trust or advisor, having a certain amount of assets and having a long service period greatly increased the probability of belonging to the minority group. A category of the investment amount of over 1 trillion yen had a positive sign but a very small impact. In other words, these results do not imply the monotonic relationship that the more institutional investors invest, the more likely they were to belong to the minority group.
We may interpret these results as follows. First, because investment trusts and advisors are less sensitive to management pressure than banks and insurers, they may support regulatory intervention that could narrow managerial discretion. Second, with respect to the non-monotonic relationship between investment amount and cluster membership, although we cannot be certain of the cause, a good possibility is that the stakes of larger investors strengthen their incentives to correct such market failures as the coordination problem, suboptimal appointment of independent directors and insufficient disciplining of managers. However, whether a significant rule revision will succeed is uncertain, and if a stake becomes extremely large, the downside risk also becomes extremely large. Institutional investors with investment amounts above a certain level may therefore be more cautious than other institutional investors about market intervention. Finally, with respect to the relationship between working experience and support for regulatory intervention, respondents may have recognized as a result of their working experience that current legislation was insufficient to overcome market problems.

Consistency Between Answers and Observed Actions
We now need to investigate whether the responses of institutional investors that we have noted are consistent with their actual behavior. Figure 2 shows the rates of dissenting votes of institutional investors with respect to proposals for introducing takeover defenses at the general meetings of shareholders of listed companies from 2012 to 2018, along with the breakdown of domestic and foreign institutional investors. Figure 2 shows that, among the total votes for the proposals of takeover defenses in listed companies, 80-90% of the votes of foreign investors were dissenting votes every year.
In contrast, only 40% of the votes of domestic investors were dissenting votes in 2012, the year of the survey. This result implies that about a decade ago, most domestic investors supported defensive measures. In the survey, domestic investors accounted for about 90% of the sample, and the majority group had a relatively favorable attitude toward defensive measures. This is consistent with the data on voting behavior. However, the rate of dissenting votes has continued to rise since then. In 2018, 80% of the votes cast by domestic institutional investors were dissenting votes. Correspondingly, as explained in Sect. 3.2.1, although in 2008 the number of companies adopting defensive measures was 569, or 24% of listed companies, by 2019 the number had gradually declined to 377, or 10% of listed companies. These facts suggest that the view in favor of restricting managerial discretion and stronger legal intervention in corporate governance, likely a minority view in 2012, may now have become the majority view.
Statistics on independent directors also suggest this possibility. As Fig. 3 shows, the proportion of listed companies that appointed two or more independent directors was 16.7% in 2012 but increased to 97% in 2021. This change may reflect changes in shareholder preferences, including those of institutional investors. Let us consider the institutional changes regarding independent directors during this period. In 2014, the appointment of at least two independent directors became the comply-or-explain rule under the Corporate Governance Code. In 2021, the amended Companies Act mandated the appointment of at least one independent director. Institutional Shareholder Services (ISS), a voting advisory company that has influenced institutional investors in Japan, now recommends opposing the appointment of top management directors when the ratio of independent directors in the board after the general meeting of shareholders is less than one third, regardless of the type of company. Moreover, the latest Corporate Governance Code, in force since 2021, stipulates the comply-or-explain rule that listed companies should maintain a ratio of independent directors of one third or more. 59 In addition to these institutional changes, the preferences of investors may also have changed. According to Table 8, although the mean score for the minority group regarding the mandatory rule (Item 1) is 4, the mean score for the majority group is 3.3. Most respondents were only weakly in favor of the mandatory rule and did not support it that enthusiastically in 2012. However, when the amended Companies Act of 2021 mandated at least one independent director, shareholders made no major objection. Moreover, some commentators have analyzed the disclosed voting standards of institutional investors in Japan and concluded that most of them now require an increase in the number of independent directors in order to improve corporate governance. 60 That institutional investors are increasingly requiring a larger number of independent directors, along with the growing number of votes against defensive measures, suggests that the preferences of domestic institutional investors may have changed over the decade, so that most of them may now prefer stronger legal intervention.
Finally, we need to consider not only changes in investor preferences but also changes in shareholder composition as factors that can influence the outcome of shareholder meetings. Figure 4 shows changes in the shareholding ratio in listed companies by investor type from 2000 to 2020. From 2012 to 2020, banks and insurance companies reduced their holdings by 3 percentage points. During the same period, investment trusts and advisors increased their holdings by 10 percentage points and foreign investors by 2 percentage points. Miyajima and Saito point out that life insurance companies, having reduced the number of investees in recent years, have begun to engage with investees more aggressively. 61 These changes in Fig. 3 Proportion of listed companies that have appointed two or more independent directors. Source: Website of the Japan Exchange Group, https:// www. jpx. co. jp/ engli sh/ listi ng/ others/ ind-execu tive/ shareholder composition and strategies may have influenced the outcomes of shareholder meetings of listed companies.

Conclusions and Discussion
Policymakers not only restrict the rights of investors but also enhance their rights by such legal devices as regulation of tender offers and derivative suits. If the designs of legal devices do not fit the preferences of their users, those users may want to avoid using such devices, which means that policymakers may not achieve their policy goals. Will investors welcome regulations that require European-style tender offers? Is the system of shareholder derivative suits beneficial from the perspectives of shareholders? To answer such questions, we need to know how investors perceive the relevant regulations. This article has examined their perceptions as manifested in an in-depth survey of institutional investors in Japan.
We found that investors exhibited a high degree of homogeneity in seeking certain legal interventions in such aspects of corporate affairs as acquisitions, director appointments and the pursuit of executive liability. We also found that they exhibited nuanced differences in their views about the degree of optimal intervention. In light of the necessity of maintaining business-transaction relationships with investees, banks and insurance companies tended to support defensive measures. Domestic investors tended to be wary of abusive takeover methods, such as greenmail, and to support defensive measures, while foreign investors tended to be neutral about defensive measures or to oppose them. By classifying respondents solely by reference to their views, without considering investor characteristics like business type or capital origin, we were able to classify investors into a majority and a minority in a two-to-one proportion. Investment trusts and advisors tended to belong to the minority that supported European-style strict tender-offer regulations, generally did not support defensive measures, and more clearly supported legal intervention in the composition of the board and pursuit of executive liability. These investors may have thought that good corporate governance could not be achieved by leaving it to the discretion of Fig. 4 Changes in the shareholding ratio in listed companies by investor type. Source: Website of the Japan Exchange Group, https:// www. jpx. co. jp/ engli sh/ marke ts/ stati stics-equit ies/ exami nation/ 01-archi ves-01. html an autonomous company, since investors such as banks and insurance companies are vulnerable to pressure from management. These facts suggest that shareholder composition may affect whether legal intervention is necessary and effective.
At the time the survey was conducted, its results were consistent with the actual voting behavior of institutional investors at shareholder meetings. However, the voting and engagement behavior of investors has changed over the last decade, and the shareholder composition of listed companies has also changed. What were the views of the minority in our data may have become the views of the majority today. If so, the type and extent of optimal legal interventions may also change. Of course, the optimal legal system is determined by a number of factors, and the views of institutional investors are only one of them. But it is necessary to listen to their opinions, especially regarding rules that directly restrict or strengthen the rights of institutional investors, such as rules affecting tender offers. With respect to the appointment of independent directors, legal intervention may be excessive if the views of investors and shareholder composition change in such a way that the composition of boards adequately reflects the opinions of shareholders; then the composition can be optimal without legal intervention.
Finally, we need to understand the limitations of this study. The first limitation is that the survey was conducted in 2012, so that the results may deviate from the current views of investors. We emphasize, however, that our purpose is not to clarify the current views of investors but to analyze possible patterns of their views. Obviously, the proportion of each pattern in the overall views of investors may change as time passes, and we have demonstrated that the minority pattern in the survey may now be the majority.
Since the introduction of the Stewardship Code in 2014, institutional investors have been disclosing voting standards and results more and more often. On the basis of the revealed preference theory, then, we can predict their current preferences at least to some extent. On the other hand, in 2012 there were fewer disclosure materials, so this survey is valuable insofar as it has clarified the detailed views of institutional investors in such circumstances. The survey was conducted at a time when authorities were revisiting the relationship between companies and institutional investors, and in the next decade, the Stewardship Code and Corporate Governance Code were enforced. In other words, the study was conducted as a new trend of Japanese corporate governance was emerging, one that continues to this day. This study allows us to track investor preferences and behavior before and after a series of reforms.
The second limitation of this survey is that the answers of respondents reflected not only the organizational views of institutional investors but also the views of individual employees. As the probit analysis of cluster membership revealed, the longer the individual respondents had worked, the more likely they were to hold the minority view that was more supportive of legal intervention.
The third limitation is that foreign institutional investors covered by the study were only those investors who had branches in Japan and were properly registered. Since the presence of foreign investors in the Japanese stock market is increasing, the trends of all investors, including investors who have overseas bases and are investing in Japan, is a topic for future research.
Despite these limitations, we believe that this study can inform future theoretical and empirical studies, especially since there is no similar in-depth survey. For example, this study provides valuable information for finance theory. In recent years, theoretical models of financial markets have increasingly assumed heterogeneity of investor preferences to describe different behavioral patterns and derive an equilibrium. 62 Our study provides information on the homogeneity and heterogeneity of investor preferences that will help researchers to examine the assumptions of theoretical models or to employ new assumptions.
The present study can also serve as a benchmark for further empirical research on the behavior of institutional investors. One intriguing possibility is to investigate the views of investors in other jurisdictions and compare them with our own results. Citing Japan as one example, Hansmann and Kraakman argued that the standard shareholder-oriented model (SSM) has influenced ideological, efficient and factual perspectives around the world. 63 As they noted, the model may have nuanced differences in different jurisdictions. By looking at the patterns of the views of investors in other countries, it will be possible to clarify which of these patterns the SSM concept embraces.