The ‘Company with Committees’: Change or Continuity in Japanese Corporate Governance?

abstract

Corporate governance practices are arguably diffusing across the world. This paper examines the adoption of the committee-based governance system (i.e. audit, nomination, and remuneration) in Japanese firms, a practice common in Anglo-American capitalism but potentially contestable in Japan. The study finds that firms that are internationally exposed through cross listing are more likely to adopt the committee system. Moreover, more experienced and highly cross-held firms, with larger proportions of foreign ownership, are more likely to adopt the committee system. On the other hand our study finds partial support for the hypothesis that larger proportions of bank ownership are negatively associated with the adoption of the committee system, suggesting a gradual withdrawal by banks from the traditional monitoring of firms. This paper adds to the longstanding debate on the convergence on or persistent divergence from the Anglo-American corporate governance system. The study thus provides insights into corporate governance changes in non-Anglo/American countries that face a struggle between global capital market forces for change and deep-seated institutional practices of continuity.

INTRODUCTION

Alongside the broad comparisons between shareholder and stakeholder capitalism (Buck and Shahrim, 2005; Dore, 2000), scholars have studied the diffusion of specific shareholder-oriented management practices across the two corporate governance systems (e.g. Fiss and Zajac, 2004). Many of these studies employ institutional theory as an analytical lens, reflecting the diversity of national corporate governance systems (Aguilera and Jackson, 2003) as well as the power and interests of their respective actors (Greenwood and Hinings, 1996). For example, Sanders and Tuschke (2007) examined the adoption of executive stock options by German firms, Ahmadjian and Robbins (2005) considered downsizing by Japanese firms, and more recently Chizema and Kim (2010) studied the appointment of outside directors on Korean boards. As such, these studies have enriched the debate on the possibility of governance systems converging on or diverging from the Anglo-American model (Hansmann and Kraakman, 2001).

In similar vein, this paper examines the adoption of a potentially contestable management practice in Japan: the use of the board committee system (i.e. the creation of nomination, audit, and remuneration committees). Specifically, drawing on institutional theory, we investigate the effect of firm-level characteristics on the adoption or non-adoption of the board committee system in Japanese firms.

Extant studies show that Japanese firms have adopted American-style stock based compensation (Kubo, 2005), and some cases of poor firm performance takeovers, though limited, have been witnessed (Milhaupt, 2005). Despite these observed changes, Japanese corporate governance stands at a crossroads between American-style stock market capitalism and the traditional stakeholder-based capitalism (Buck and Shahrim, 2005), suggesting the presence of both change and continuity (Yoshikawa and McGuire, 2008). Studies, in the domain of Japanese corporate governance, that potentially explain the phenomena of change or continuity, include ownership structure (Yoshikawa and Gedajlovic, 2002), outside directors (Yoshikawa and Phan, 2003), corporate restructuring (Ahmadjian and Robbins, 2005), and shareholder activism (Seki, 2005). Despite the importance of the board system in corporate governance (Dahya and McConnell, 2007) and its attendant committees, however, there is a dearth of work focusing comprehensively on the dynamics or determinants of adopting the committee system, at least in Japan. Gilson and Milhaupt (2005) have come closest to remedying this omission, but, only a year after the recommendation to adopt the committee system was applied. Moreover, their study, rooted in corporate law, fails to fully engage a theoretical framework that takes into account the salience of the institutional environment and lacks a substantial statistical analysis due to the limited observation period available at that time.

Employing the institutional theory lens over an extended period of observation, our study empirically tests a number of hypotheses informed by an understanding of Japan's unique institutional and market environment. The study fits into the large domain of international corporate governance, in general, exploring the potential diffusion of a contested management practice from one governance system to another. Thus, our study provides insights into corporate governance changes in non-Anglo/American countries that face a struggle between global capital market forces of change and deep-seated institutional practices of continuity.

Our study also claims to make theoretical contributions to the field of corporate governance. Broadly, this study contributes to the governance literature on boards of directors in general and the committee system in particular, adding to the longstanding debate on the convergence or persistent divergence of corporate governance systems. In addition, by considering the adoption of the board committee system in Japan, deriving from an Anglo-American governance system, our article contributes to the understanding of the institutional context in the study of governance change. The study argues for the particular relevance of institutional forces and dynamics in Japan, thus proposing the wider application of institutional theory to governance change even in advanced economies during periods of governance reform. Moreover, the fact that the committee system was a welcome governance tool in some firms while rejected by others within the same national environment, provides an explanation for the potential co-existence of change and continuity (Townley, 2002) during transition periods.

This paper is organized as follows. We first discuss theory followed by the research context. Next, we draw on both this theoretical background and the study context in developing our hypotheses. This is followed by a section on methodology after which we present our results. The final section is on discussion and conclusions.

THEORETICAL BACKGROUND

Institutions are defined as both formal and informal rules that constrain human interaction in a society (North, 1991) or shared rules including laws and collective understanding (Fligstein, 1991). These definitions share the notion that institutions provide a framework for social interaction and thereby make social order possible by reducing uncertainty (Scott, 2001).

Since institutions are social structures that are composed of cognitive, normative, and regulative elements (Scott, 2001) and that are embedded in a local context, they are usually highly resistant to change (Sanders and Tuschke, 2007). Hence, institutions, including corporate governance institutions, tend to reinforce the continuity of established systems, behaviour, and practices. Despite this view, governance institutions and their elements do change from time to time (Fiss and Zajac, 2006) due to both external and internal pressures (Oliver, 1992; Scott, 2001). This is evident in research that has extended the domain of institutional theory to understanding the conditions that may lead actors or ‘agents’ to conform to social convention or to challenge the institutional fabric by initiating non-isomorphic action (Dacin et al., 2002). Such work has examined external triggers for, and barriers to, institutional change (Oliver, 1992), organizational responses to these environmental triggers (Lawrence et al., 2002), or organizational features that support change (Kostova and Roth, 2002). These arguments have been extended to studies on corporate governance changes within and across countries (Aguilera and Jackson, 2003). Here, the globalization of capital markets has been seen as an external trigger that contributes to the pressures to change corporate governance systems in many countries (Khanna and Palepu, 2004). In the language of institutional theory, such pressures lead to institutional isomorphism (DiMaggio and Powell, 1983), a tendency for countries and organizations to adopt similar institutions (e.g. corporate governance structures).

However, in the domain of corporate governance, the view that institutional isomorphism or convergence occurs through globalization fails to observe the heterogeneity of firms, evident through ownership structure (Colpan et al., 2011), firm resources, and capabilities. Moreover, such a view ignores the distinct national features and deeply rooted traditions of business practices (Aguilera and Jackson, 2003), including corporate governance elements. Taken together these arguments suggest that globalization-driven changes may not be uniform, even within the same country (Chizema, 2008).

The case of Japan and her firms, in the context of changes in both national and international corporate governance, presents a laboratory for testing and developing a better understanding of institutional theory arguments.

RESEARCH CONTEXT

Following the stock market crash of 1990 and the busting of the property bubble in the subsequent years, Japan suffered a severe economic slump. At a macroeconomic level, the GDP growth rate per capita was 0.6 per cent during the period 1992–2002. In contrast, the USA managed to maintain its historical average of 2.0 per cent for comparative data (World Bank, 2008). At the level of corporate performance, the share prices of Japanese firms have consistently underperformed those in other major economies over the last two decades. Figure 1 shows the poor performance of Japanese shares relative to their US counterparts during Japan's so-called lost decades.

Figure 1. A comparison of Japan and US stock indices between 1985 and 2009

The Japanese banking sector was also hit badly by the property market crash in the early 1990s. In particular, major banks centred in their keiretsu groups, the loose conglomerations of companies, suffered from non-performing loans (Dinç, 2006). The need to dispose of their non-performing loans and to meet the Bank for International Settlements (BIS) capital requirement resulted in the unwinding of cross-shareholdings in keiretsu groups (Dvorak et al., 2001). As a consequence, the main banks lost the centripetal force within the groups and reduced the monitoring function towards their affiliated firms (Gedajlovic et al., 2005). At the same time, foreign institutional investors increased their investments in Japanese shares (Ahmadjian and Robbins, 2005) as their prices fell.

In response to these economic problems and investors' lobbying for change, the Japanese government started reforming its governance system from 2001 through the amendment of its corporate laws (Gilson and Milhaupt, 2005). Notably, the governance reforms were centred on the role and responsibilities of the board of directors (Gilson and Milhaupt, 2005).

First, in 2001 amendments were made to the Japanese Commercial Code (hereafter, the Code) and certain other Acts concerning corporate law. Following these changes, statutory auditors were given an increased role within their corporation and external statutory auditors were made more independent.

Second, in 2002, another amendment was made to the Code. The committee system was introduced as an alternative to the conventional statutory auditor system of corporate governance, discussed below. Third, in 2005, the new Corporation Law was introduced in order to modernize Japanese corporate law. While the amendment to the Code and the relevant Acts in 2001 was aimed at improving the existing system in corporate governance, the amendment in 2002 created an entirely new system, which corporations may opt for as an alternative to the conventional one.

The Conventional Auditor System

A weaker version of the German supervisory board in Japan is the board of statutory auditors. Indeed, the Japanese statutory auditor lacks the power to appoint or remove directors, and does not necessarily represent shareholder or employee interests, since auditors are nominated by the board. The main functions of auditors are to monitor the board's compliance with law and to review the financial statements.

Auditors, who may attend meetings of directors and receive regular reports from directors, tend to be internally promoted employees of the corporation, most commonly having been directors or departmental chiefs. Over the years, several attempts were made to strengthen the auditor system. For example, amendments made in 1993 extended the auditor's term of office and mandated that large companies have at least three auditors, rather than a minimum of one, and that they function as a board of audit. Moreover, one member of the board of audit must be an outside director.

In 2001, amendments sought to further strengthen the auditor regime by extending the term of office and responsibilities of auditors, while increasing the required number and qualifications of outside auditors. Effective from 2005, at least half of the board of audit must be comprised of outside auditors. However, under the conventional auditor system there is no requirement for outside directors.

The Committee System

The 2002 reform allows large firms to adopt an American-style board system for corporate governance beginning April 2003. This suggests the replacement of the board of statutory auditors with three committees, namely audit, nomination, and compensation: the so-called ‘company with committees’ system.

The Code suggests that each committee must have at least three directors, a majority of whom must be ‘outside’ directors. An ‘outside’ director is defined as ‘one who does not have a role in executing the company's business, who has not assumed the position of director, executive officer (shikkou-yaku), manager or any other employee with the role of executing the business of the company or its subsidiaries in the past, and who does not assume the position of director or executive officer of such subsidiaries or manager or any other employee of such company or subsidiaries with the role of executing the business thereof’ (Commercial Code, 2003).

While the Code's recommendation on the appointment of outside directors rules out former executives and employees of the firm, the independence of outside directors is not guaranteed. Even those companies that have opened their boards to outsiders usually invite individuals who are from companies that are corporate partners, creditors, customers, or suppliers, i.e. not independent. The main structural differences between the conventional statutory auditor system and the committee system are shown in Figure 2.

Figure 2. The old and new board system in Japan

Under the committee system, a company will also have executive officers, who are appointed and dismissed by the board of directors and who make the material management decisions for the company. The board of directors may delegate substantial management authority to executive officers under this system. For example, the board may delegate to executive officers the authority to approve the issues of shares of capital stock and bonds. In addition, the Code permits an individual to simultaneously be a director and an executive officer of the company.

Under the committee system, the audit committee generally has powers and duties that are similar to those given to corporate auditors. This includes responsibility for monitoring directors and executive officers in the performance of their duties, as well as the right to propose the appointment or dismissal, or to refuse reappointment, of the company's certified public accountants at the general meeting of shareholders. Accordingly, any proposal for appointment or dismissal of a certified public accountant needs to be submitted to a general meeting of shareholders for approval. The nomination committee is responsible for determining any proposal for the election or removal of directors, which must then be adopted by shareholders at the general meeting. The compensation committee decides on policy regarding the content of compensation and related matters for directors and executive managing directors.

The committee system represents a compromise between the existing auditor system and the Anglo-American system (Yoshikawa and McGuire, 2008). It offers large Japanese firms the choice between the auditor system and a committee system similar to the one adopted by listed US firms. In this system, unlike the auditor model, there is a clear legal separation between monitoring and execution functions, and arguably provides better transparency.

However, this management practice has met with substantial resistance from the business community. Common charges against such a requirement are that outside directors are not well suited to perform a useful role in highly relational Japanese corporate affairs, and that finding suitable outside directors will be extremely difficult given the lack of experience with the practice (Gilson and Milhaupt, 2005). Consequently, the final version of the 2002 reform contains a strong endorsement of the concept of choice, including the option of abandoning traditional Japanese institutions in favour of the US-style committee system.

HYPOTHESES

Japanese firms face increasing pressures to change due to the globalization of financial markets (Khanna and Palepu, 2004). Although such pressures exhibit commonality at national level, individual firms, owing to ownership heterogeneity (Colpan et al., 2011) and other varying characteristics, perceive and deal with changes in different ways. Although the idea of the board committee system in Japan was the product of legal and regulatory initiatives at national level, its adoption or non-adoption is dependent on the individual firm's circumstances.

While we acknowledge the possibility of numerous intervening firm-level characteristics over a course of action, however, we argue parsimoniously that five broad factors determine the choice between adopting or non-adopting the board committee system in Japan. We refer to them as precipitating factors of change or continuity, where change implies that adoption and continuity refers to non-adoption of the committee system, i.e. firms prefer to carry on with the auditor system. First, informed by institutional theory arguments on organizational behaviour imitation (DiMaggio and Powell, 1983), we acknowledge the firm's degree of international exposure as a vehicle for the adoption of management practices from a different governance model (Ahmadjian and Robbins, 2005).

Second, firms have different perceptions of the benefits or costs that are likely to result from the changes that are triggered at the macro environment. The reaction by the firm through the interests of its actors such as shareholders would reflect this perception (Johnson et al., 2010). Third, change or continuity at firm-level is also dependent on the firm's experience and the support that the firm receives from its institutional relationships (Redmond, 2003). Fourth, we argue that firms may be quick to embrace changes if the context of or situation surrounding such changes does not alter the status quo. In other words, the impression on the surface may suggest that change has taken place but a deeper analysis could show otherwise (Westphal and Zajac, 2001). We view this as continuity disguised as change. Fifth, the interests, value commitment, and power of ownership groups potentially determine major corporate decisions (Greenwood and Hinings, 1996). In the context of Japan and based on previous studies (e.g. Ahmadjian and Robbins, 2005; Lincoln and Gerlach, 2004), we identify the salience of foreign and bank ownership as determinants of the decision to adopt or reject the committee system.

In the subsections below, we develop hypotheses that test the determinants of adopting or not adopting the committee system in Japan.

The Firm's Level of International Exposure

International exposure allows firms to question their current practices, comparing them with those of their suppliers, buyers, or competitors who may use different corporate governance systems. Oliver (1992, p. 577) claims that ‘firms that diversify their operations into other sectors or markets, particularly in different countries are likely to be exposed to alternative organizational customs’, a situation that helps firms to learn and adopt new management practices.

Recent trends in globalization show that firms are internationally exposed through financial markets, as they cross list on major stock exchanges, particularly in the USA (Sanders and Tuschke, 2007). These developments have been seen by some scholars as leading to the convergence of corporate governance systems on the American shareholder-oriented model (Hansmann and Kraakman, 2001). Of course, arguments to the contrary abound (see Guillen, 2000, for a detailed review). Notwithstanding the relevance of arguments on governance divergence, the convergence thesis seems to be gaining currency, predicting the gradual erosion of institutional differences among different national economies because of intensified global competition (Deeg and Jackson, 2007). As such, firms that are internationally exposed through product and financial markets have the chance to learn management practices from their customers and competitors drawn from alternative governance models. In institutional theory parlance this represents mimetic isomorphism, a situation where firms imitate practices that are seen to be used by successful firms (DiMaggio and Powell, 1983).

A number of Japanese firms have securities listed in the USA through American Depositary Receipts (ADRs). An ADR is a receipt for the shares of a foreign-based company held by a US bank that entitles the shareholder to all the dividends and capital gains of the underlying stock. ADRs trade similarly to stocks on US exchanges and provide a way for Americans to invest in foreign-based companies. By listing ADRs in the USA, Japanese firms submit themselves to some scrutiny and compliance by the Securities Exchange Commission (SEC). Such compliance, justifiable in institutionary theory as leading to coercive isomorphism (Chizema, 2010; DiMaggio and Powell, 1983), includes the use of committees in company governance. In the context of Japan, we argue that international exposure achieved through cross listing on US stock markets could be seen as a driver of corporate governance reforms. We therefore hypothesize:

Hypothesis 1: Firms cross listed on US markets are more likely to adopt the committee system.

Perception of Benefits or Costs from Changes

Organizational change is a risky decision, implying that managers must have legitimate reasons and compelling incentives to break their existing routines or practices (Greve, 1998). Previous literature suggests that past performance is one such motivator (Tushman and Romanelli, 1985). However, extant studies have provided inconsistent predictions regarding the relationship between past performance and organizational change. Indeed, some researchers suggest that poor performance widens the gap between managerial aspirations and achievements, thereby providing a strong incentive for firms to look for new ways to improve (Miller and Chen, 1994). This argument receives support from the institutional theory view that poor performance results in uncertainty forcing firms to adopt new methods of operating, often imitated from other organizations perceived to be more successful (DiMaggio and Powell, 1983). Thus, poorly performing firms may seek changes with the intention of improving their financial returns.

On the other hand, good performance continuously motivates companies to change (Feldman, 2004; Tsoukas and Chia, 2002). A third possibility is that firms that have a sustained history of good performance are likely to resist any changes to the status quo (Greenwood and Hinings, 1996).

These inconsistent views may be explained, in part, by considering the nature of change. Scholars have broadly distinguished between technical and administrative change. Technical change tends to be associated with previous success as firms continuously reinvent themselves (Tsoukas and Chia, 2002). As Brown and Eisenhardt (1997) observe, many successful firms, such as Intel, 3M, and Hewlett-Packard, have undertaken constant, rapid changes, particularly in their new product development.

Unlike technical change, often seen through R&D and production methods, administrative change relates to new procedures or practices in a firm's key managerial processes. The adoption of the committee system by Japanese firms is thus an example of an administrative change. Such changes are not limited to a specific division but represent a more general departure from organizational routines (Han et al., 1998) and tend to receive strong resistance from inside actors (Hannan and Freeman, 1984). Consequently, administrative changes are difficult to initiate without strong and legitimate reasons.

Poor past performance signifies the ineffectiveness of existing practices, and thus provides strong and legitimate reasons for firms to reform their administrative systems. Thus, where a firm's performance is weak, e.g. share price falls or rates of job creation are low, organizational groups may express their dissatisfaction by tilting the balance of forces in favour of institutional change. Therefore, a firm's performance may cause an erosion of the present commitment, triggering political ‘dissensus’ over existing arrangements (Child and Smith, 1987). This may permit groups less committed to prevailing governance elements to legitimately raise and promote alternative configurations (Oliver, 1992). In the case of poorly performing Japanese firms, the adoption of the committee system may be a justifiable alternative to governance that carries the potential for better performance. We therefore suggest:

Hypothesis 2: Firms with weak financial performance are more likely to adopt the committee system.

Firm Age

Previous studies have shown that older and well-established firms are central or dominant players in the organizational field (Sherer and Lee, 2002), strongly embedded in the old national institutions and thus resistant to change. At the same time, younger firms or the so called ‘born globals’ (Oviatt and McDougall, 1994), with very little investment in past management practices may be keener to embrace changes (Greenwood and Hinings, 1996). This implies that change is likely to take place among younger organizations (Nagaoka, 2005) and that higher resistance is expected among older ones because organizations tend to ossify as they age, possibly inhibiting the adoption of governance reforms.

However, there is a chance that the opposite could be true. First, Greenwood and Hinings (1996) caution that organizations may vary in patterns of commitment to change because of their different locations within their organizational field, a point that reduces the salience of age. A firm's organizational field includes its suppliers, customers, and the rest of its social and economic network. Second, in a study of the Big Five accounting firms, the largest and oldest in this industry, Greenwood and Suddaby (2006) show how actors can occupy socioeconomic positions that make them aware of favourable alternative institutional arrangements, thus becoming motivated to pursue these alternatives. Third, and related to the preceding argument, we posit that the assumption that older firms are more resistant to change than younger ones fails to embrace both the nature of the change in question and the circumstances under which such changes are expected. Where governance reforms are inevitable and have the potential to provide benefits to older and younger firms alike, the chances are that older firms may be in a better position to adapt.

Following the economic stagnation in Japan and general corporate underperformance, all firms, regardless of size or age, needed to change. In this situation, failure to change may not be attributable to actors' resistance but to issues pertaining to their ability to execute change. Here, firm age may be indicative of experience-based capabilities, ability to adapt, reliability, and market credibility (Baum and Shiplov, 2006). This view has also been observed in international business studies where older organizations have been associated with the knowledge and experience necessary for export growth (Reuber and Fischer, 1997).

With age, a firm may develop capabilities to undertake change and adapt to new environments (Kelly and Amburgey, 1991), such as those required for corporate governance. Moreover, Haveman (1993) argues that older organizations have, over the years, built networks and relationships with institutions that may also provide access to important supporting resources. They are thus located in an enabling organizational field. These characteristics may lead older firms, having greater ability and flexibility, to adopt a new form of committee governance in comparison with younger firms, owing to their legitimacy with supporting institutions. With respect to the adoption of the committee system in Japan, supporting institutions could include consultancy firms, the institute of directors, and professional accounting bodies. Integrating these arguments, age reflects experience and the level of the firm's relationships with supporting institutions. Institutional relationships and legitimacy may lead older firms to have greater ability to gain institutional support to change from one form of a governance mechanism to another. We therefore hypothesize:

Hypothesis 3: The more experienced a firm is the more likely the adoption of the committee system.

Opportunity for Continuity under the Disguise of Change

Given that at least two outside directors are needed to serve on each of the three committees, a natural expectation is that adopting firms should have significantly more outside directors than listed companies in general. A study by Gilson and Milhaupt (2005) shows that outside directors comprise between one-third and two-thirds of the boards of Japanese firms that adopted the committee system. By contrast, only one-quarter of all Tokyo Stock Exchange-listed firms have outside directors on their boards.

Notwithstanding the fact that outside directors have been observed to be effective in relation to strategic change (Johnson et al., 1993), restructuring (Pearce and Zahra, 1992), and international diversification (Tihanyi et al., 2003), their appointment to boards is contestable in Japan. Traditionally, outside directors were not known in the keiretsu networks in Japan, with companies connected through cross-holdings, inter-trading, and interlocking directorships (Lincoln and Gerlach, 2004). Reflecting the social cohesion important to Japanese society, keiretsu emphasize unity throughout an organization and non-adversarial relationships, qualities that are likely to be lost by introducing ‘outsiders’. Against this background, a natural expectation is that cross-held firms would resist the idea of the committee system that is highly characterized by the presence of outside directors.

However, the concept of outside directors in Japan is different from that of the American model of corporate governance where they [outside directors] are expected to be independent. As pointed out earlier, outside directors can be affiliated with a major shareholder, parent company or another subsidiary of the company. According to Yamada (2003), for example, virtually all of the outside directors of the Hitachi group companies are affiliated with Hitachi Ltd, and most of the outside directors of the Nomura subsidiaries are affiliated with Nomura Holdings or Nomura Securities. Thus, owing to the legal definition of outside directors in Japan, firms that are in strong networks of cross holdings have a large pool from which potential outside directors can be drawn. For these firms, a great opportunity in the loose legal definition of outside directors is that they can appoint individuals who may not threaten to change the status quo. Moreover, the fact that directors affiliated with parent and sibling firms are eligible to serve on the committees of audit, compensation, and nomination enhances group cohesion and provides a perfect opportunity for parent companies to assert greater control over member firms (Gilson and Milhaupt, 2005).

Effectively, these firms gain the legitimacy and signalling power associated with the adoption of the committee system without actually making practical changes in their management orientation. This observation is consistent with previous studies that suggest that even where inertia appears to have been overcome and governance seems to have been reformed, changes may turn out to be cosmetic, symbolic, or camouflaged (Buck and Shahrim, 2005) with underlying relations unchanged. Following this discussion we suggest:

Hypothesis 4: The higher the level of cross holdings, the more likely a firm's adoption of the committee system.

Ownership Structure

Studies informed by institutional theory (e.g. Greenwood and Hinings, 1996) suggest that within an organization there are different groups with diverse interests and value perception. As Palmer et al. (1993, p. 103) observed, ‘organizations are arenas in which coalitions with different interests and capacities for influence vie for dominance’. This potential intra-organizational conflict suggests that certain groups or actors successfully convert their interests into favourable allocations of organizational resources (Greenwood and Hinings, 1996), evidently at the expense of others. In the same way, such groups could influence the adoption of management practices of their choice. This suggests that pressure for either change or inertia depends upon the extent to which groups have an interest in proposed reforms and how much power they possess to either support or resist them.

In the context of firms, group classifications such as foreign, institutional, or bank shareholders are bound to have differences over a number of major corporate decisions depending on their interests. Moreover, their influence over such decisions could be a function of power geometry. Based on previous studies on Japanese corporate governance (e.g. Ahmadjian and Robbins, 2005), that have emphasized the salience and powerful influence of foreign and bank ownership, we limit our discussion to the two in the subsections below.

Foreign ownership.  In Japanese corporations, foreign owners tend to be predominantly institutional investors from the USA and UK (David et al., 2006). For instance, in 1997, investors from the USA and UK held 32 percent and 39 percent of total foreign shares, respectively (Bank of Japan, 2004). Empirical studies confirm the influence and governance role played by foreign investors (Dahlquist and Robertson, 2001). For example, in the context of Korea, Chizema and Kim (2010) find a positive and significant relationship between foreign ownership and the appointment of outside directors on Korean boards. In the context of Japan, Ahmadjian and Robbins (2005) find a strong relationship between the percentage of shares owned by foreigners and corporate behaviour reminiscent of Anglo-American governance such as downsizing.

Empirical research also provides some evidence that foreign owners do matter and affect corporate outcomes such as firm performance (Yoshikawa and Phan, 2003) and wages (Yoshikawa et al., 2005). The growing popularity of foreign ownership in Japan has been claimed by Yoshikawa and Gedajlovic (2002), as Japanese firms have stepped up investor relations efforts to attract foreign investors. Of course, foreign owners were even there to buy shares from stable domestic owners that needed to exit their holdings during the economic downturn in the 1990s (Dvorak et al., 2001).

Foreign investors, particularly from the USA and UK, have norms and values that emphasize the maximization of shareholder value (Chizema and Kim, 2010), and thus welcome governance elements from the Anglo-American model. To achieve their goals, foreign investors could threaten to sell off their shares, i.e. take the ‘exit’ option (Nooteboom, 1999). The presence of foreign investors indicates confidence that the firm is well managed, while sell-offs signal the possibility of managerial opportunism and poor management (David et al., 2006), potentially reducing stock price and increasing the cost of capital. Hence, firms are likely to be obliged to appease foreign owners by adopting America-style committee system. We therefore suggest:

Hypothesis 5a: The larger the percentage of foreign ownership, the more likely a firm's adoption of the committee system.

Bank ownership.  As in Germany, Japanese banks have traditionally played an important corporate governance role, at least compared with the USA or UK (Dore, 2000). This is because Japanese corporations are organized into keiretsus, with a main bank as both a major lender and owner (David et al., 2006). This web of long-term relationships between stable owners, main banks, and keiretsu partners (Lincoln and Gerlach, 2004) provides monitoring while protecting managers from hostile takeover (Lee and O'Neill, 2003). This suggests that the adoption of the committee system substitutes, in part, the role played by banks in corporate governance, an idea they are likely to resist. First, from a power perspective, banks have ‘voice’ and direct influence through ownership (David et al., 2006) and seats on the board, and thus may not welcome new institutional rules that come along with the adoption of the committee system.

Second, the committee system is a shareholder-oriented governance element, thus consistent with stock market (rather than bank) financing of corporations, a situation that may contrast and ruin the financial and strategic interests of Japanese banks (Gedajlovic and Shapiro, 2002). Through their embedded relationships with firms, banks foster their attitudes and interests that promote continuity (Yoshikawa and McGuire, 2008) towards important stakeholders, including employees and management (David et al., 2010). Moreover, firms that have close ties with banks may resist the adoption of the committee system as a signal of their solidarity with the interests of their finance partners. This view is supported in the empirical literature as bank ties through ownership has been found to be negatively associated with the adoption of executive stock options (another aspect of shareholder-oriented governance system) in German firms (Chizema, 2008). Of course, research suggests that owing to pressures of globalization, banks have started to cut back their conventional ties to their keiretsu firms (Gedajlovic et al., 2005). However, complete loss of bank influence on firms may not be experienced soon. Following these arguments we hypothesize:

Hypothesis 5b: The higher the percentage of bank ownership, the more likely a firm will not adopt the committee system.

METHODOLOGY

Data

Our data sample is drawn from the Topix 500 index that consists of the top 500 Japanese firms listed on the Tokyo Stock Exchange. The index accounts for approximately 85 per cent of the Exchange in terms of market capitalization. Given that the 2002 governance reforms that allow large firms to adopt the committee system were effective from April 2003, our period covers December 2002 to December 2009. Our sample excludes firms that are delisted from the Tokyo Stock Exchange, merged, or acquired by other firms. We also excluded firms from which we could not get complete data, leaving us with a final sample of 235 firms observed over seven years. For each sample firm, financial data is collected from Datastream and corporate governance information is retrieved from the Tokyo Stock Exchange and Kabu Pro websites.

Dependent Variable

In relation to the adoption of the committee system or continued use of the auditor system, the dependent variable, AdoptCom is derived. If a firm adopts the committee system, the variable is coded 1, and 0 otherwise.

Independent Variables

To measure the effect of globalization on the firm, we use a dummy variable Cross listing, coded 1 for firms listed on US stock exchanges and with Level II or III ADRs. Level II and III ADRs compared to Level I or Over the Counter trading (OTC) require foreign firms to comply with SEC regulations in the same way as US firms. Financial performance is taken as an accounting-based ratio: return on equity (ROE). We calculate a three-year lagged moving average for the ROE. For the third hypothesis, Firm age is calculated as the logarithm of the difference between the year of establishment and the observation year. For the fourth hypothesis, Cross holdings is the percentage of a firm X's share ownership owned by the affiliated group companies, whose shares are also partially owned by the same firm. Foreign ownership is measured as the percentage of total shares outstanding held by foreign investors. Bank ownership is taken as the percentage of total shares outstanding held by Japanese banks.

Control Variables

Employee ownership represents resistance to management decisions that may bring disadvantages to employees. In Japan, for example, studies have shown positive association between corporate governance adoptions and downsizing (Ahmadjian and Robbins, 2005). We measure employee ownership as the percentage of total shares outstanding held by a firm's employees. We control for firm size because large firms have better resources and capacity to implement change than smaller ones (Sherer and Lee, 2002). Firm size is taken as the logarithm of the firm's market capitalization. Lastly, we control for industry, based on 2-digit SIC classification. Based on our data we derive nine industry dummy variables.

Econometric Model

We modelled the committee system adoption using discrete-time event history analysis (Allison, 1984). The use of discrete-time analysis allows observation of the same organizations at multiple intervals and pooled time series data through the estimation of logit models of dichotomous outcomes. We estimated a baseline model of the hazard (i.e. likelihood) of the committee system adoption in any of the observed years. This method allows for covariates to vary between time periods. The model has the following equation:

• 

where represents the logarithmic odds that a firm j will adopt the committee system at any point during period t; α(t) implies that the hazard rate (i.e. the likelihood) for adopting the committee system varies across time. To estimate α(t), two year dummies (Pre-2005 and Post-2005) were entered. β represents the set of parameters to be estimated, where independent variables are lagged by one year. This method treats the data as quasi-cross-sectional; if a firm adopts the committee system in year 1, it contributes one firm-year, and at year 2, two firm-years, and so on. Non-adopting firms contribute as many firm-years as there are in the period of observation. In short, each of the censored firms contributes a maximum n firm-years, where n is the longest time interval. This approach has been used by several studies on the adoption of management practices (e.g. Sanders and Tuschke, 2007).

DATA ANALYSIS AND RESULTS

Table I shows the descriptive statistics of the variables and the results of the correlation analysis. Firm performance measured as ROE averages 6.09. The average age of firms in the sample is 62.72 years. The average of cross holdings is 10.78 per cent and the average for foreign ownership is 20.81 per cent. The average for bank ownership is 36.47 per cent. Employee ownership averages 2.33 per cent while the average logarithm of market capitalization is 19.69. Correlations between independent variables are low. As a check we carried out variance inflation factors (VIF) tests and found that all the condition indices are below 10, implying no problems with multicollinearity (Neter et al., 1990).

Table I.  Descriptive statistics and correlation analysis

	Mean	SD	1	2	3	4	5	6	7	8
Notes: * p < 0.05, ** p < 0.01. N = 235.
1. Cross listing	0.04	0.2	1
2. Financial performance	6.08	14.33	0.03	1
3. Firm age	62.72	23.57	−0.11**	−0.04	1
4. Cross holdings	10.78	15.52	−0.02	0.01	−0.36**	1
5. Foreign ownership	20.81	13.07	0.13**	0.15**	−0.07**	−0.17**	1
6. Bank ownership	36.47	13.21	0.02	0.02	0.39**	−0.441	−0.16	1
7. Employee ownership	2.33	6.66	0.03	−0.03	−0.29**	−0.025	−0.07*	−0.27**	1
8. Firm size	19.68	1.11	0.33**	0.06	−0.05*	−0.122	0.46	−0.02	−0.15**	1

Table II shows the cumulative numbers of firms that adopted the committee system as well as the figures of non-adopting firms for the years 2002 to 2009. Almost 9 per cent of firms in our sample adopted the committee system, with the majority of these firms making this decision before 2005.

Table II.  Adoption and non-adoption of the committee system

Year	Adopting firms (cumulative)	Non-adopting firms
2002	00	235
2003	14	221
2004	18	217
2005	18	217
2006	19	216
2007	19	216
2008	20	215
2009	20	215

Table III provides mean differences of firms that adopted the committee system and those that did not. The mean differences for the variables cross listing, cross holding, and foreign and bank ownership are significant.

Table III.  Means t-test for committee-adopting and non-adopting firms

	Mean
	Adopters	Non-adopters	Difference
Notes: * p < 0.05, ** p < 0.01, *** p < 0.001.
Cross listing	0.20	0.05	0.15**
Financial performance	3.19	6.12	−2.93
Firm age	65.40	63.86	1.54
Cross holdings	23.90	10.61	13.29***
Foreign ownership	27.10	20.72	6.38**
Bank ownership	31.75	36.65	−4.90**
Employee ownership	1.09	2.34	−0.44
Firm size	20.25	19.87	0.38

Table IV presents the results of the logistic regression model. Model 1 contains control variables only. Model 2 contains controls plus the first independent variable: cross listing. Model 3 has controls plus the first two independent variables: cross listing and financial performance. Following this pattern we include the six independent variables in Models 2 to 6. Model 7 has the contents of Model 6 plus two time dummies that capture the indirect effects of the recommendation to adopt the committee system made in 2002 and subsequent regulatory changes in 2005. We also include Model 8, showing the interaction of independent variables and the period pre-2005. In each model, we report the coefficient estimate along with its standard error (in parentheses). Using results from Model 7 (the one with all the independent variables), we interpret the likelihood of adopting the committee system based on 1 per cent increase or decrease of a given independent variable.

Table IV.  Event history analysis for the adoption of the committee system

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
Notes: † p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001. N = 235.
Constant	−8.66 (3.59)	−7.33 (3.51)*	−7.37 (3.47)*	−7.50 (3.49)*	−10.17 (3.64)**	−12.40 (3.90)***	−13.23 (4.04)***	−9.31 (3.93)*
Cross listing		1.27 (0.67)†	1.34 (0.67)*	1.39 (0.69)*	1.79 (0.77)**	1.73 (0.78)*	1.68 (0.81)*
Financial performance			−0.02 (0.02)	−0.02 (0.02)	−0.02 (0.02)	−0.02 (0.02)†	−0.02 (0.02)
Firm age				0.01 (0.01)	0.03 (0.01)**	0.03 (0.01)**	0.03 (0.01)*
Cross holdings					0.05 (0.01)***	0.07 (0.02)***	0.05 (0.02)***
Foreign ownership						2.85 (1.42)*	3.94 (1.44)**
Bank ownership						−1.74 (0.82)*	−0.80 (0.79)
Employee ownership	0.01 (0.04)	0.01 (0.04)	−0.01 (0.04)	−0.01 (0.04)	0.02 (0.04)	0.04 (0.04)	0.02 (0.05)	0.01 (0.05)
Firm size	0.22 (0.22)	0.15 (0.17)	0.15 (0.17)	0.15 (0.17)	0.16 (0.18)	0.27 (0.19)	0.26 (0.19)	0.20 (0.19)
Pre-2005							2.45 (0.54)***
Post-2005							0.00 (1.38)
Cross listing ×  Pre-2005								2.11 (0.92)*
Financial performance ×  Pre-2005								−0.02 (0.32)
Firm age × Pre-2005								0.03 (0.01)***
Cross holdings ×  Pre-2005								0.06 (0.02)***
Foreign ownership ×  Pre-2005								−1.20 (0.79)
Bank ownership ×  Pre-2005								−0.49 (0.87)
Chi-square	1.61	2.82†	1.48	0.12	16.29***	7.96***	23.12***	51.74***
−2log likelihood	211.93	209.11	207.63	207.51	191.22	183.26	160.13	160.2
Nagelkeke R-square	0.01	0.02	0.03	0.03	0.11	0.15	0.27	0.26

Starting with control variables, employee ownership and firm size do not have any significant effect on the adoption of the committee system. We also controlled for industry in all the eight models. From the results, industrial classification does not have any effect on the decision to adopt or reject the committee system.

Turning to independent variables, Hypothesis 1 states that firms that are cross listed on US stock exchanges are more likely to adopt the committee system. This hypothesis is supported, as results show that cross listing on US stock exchanges is positively associated with the adoption of the committee system in Model 2 (β = 1.27; p < 0.1), Model 3 (β = 1.34; p < 0.05), Model 4 (β = 1.39; p < 0.05), Model 5 (β = 1.79; p < 0.01), Model 6 (β = 1.73; p < 0.05), and Model 7 (β = 1.68; p < 0.05). Based on the coefficient from the final model, Japanese firms with Level II or III ADRs are 5.4 times likely to adopt the committee system than those that are non-listed.

Hypothesis 2 states that firms with weak financial performance will adopt the committee system. This hypothesis is weakly supported in Model 6 only at (β = −0.02; p < 0.10). We checked for the robustness of our empirical analysis here by using a different performance ratio, i.e. return on assets (ROA) and by using a one-year lag of ROE. There was no significant change in the estimates.

Hypothesis 3 proposes that the more experienced a firm is the more likely it will adopt the committee system. In support of this hypothesis the coefficients for firm age are positive and statistically significant in Model 5 (β = 0.03; p < 0.01), Model 6 (β = 0.03; p < 0.01), and Model 7 (β = 0.03; p < 0.05). An increase in firm age by one year increases the likelihood of a firm adopting the committee system by 2.6 per cent, i.e. 100 [exp (0.03) − 1].

Hypothesis 4 predicts that the higher the cross holdings, the more likely the adoption of the committee system. In keeping with this hypothesis, the coefficients for cross holdings are positive and statistically significant in Model 5 (β = 0.05; p < 0.001), Model 6 (β = 0.07; p < 0.001), and Model 7 (β = 0.05; p < 0.001). Results from Model 7 show that a 1 per cent increase in the firm's cross holdings increases the likelihood of a firm adopting the committee system by 5.5 per cent, i.e. 100 [exp (0.05) − 1].

Hypothesis 5a states that the larger the percentage of foreign ownership the more likely a firm will adopt the committee system. This hypothesis is tested and supported in the last two models, i.e. Model 6 (β = 2.85; p < 0.05) and Model 7 (β = 3.94; p < 0.01). These findings add support to previous research (e.g. Ahmadjian and Robbins, 2005) that points to the salience of foreign ownership in the adoption of shareholder value oriented practices.

Hypothesis 5b states that the larger the percentage of bank ownership the less likely a firm will adopt the committee system. This hypothesis is tested in the last two models, i.e. Models 6 and 7. The hypothesis is supported in Model 6 (β = −1.74; p < 0.05) and not in Model 7. These findings possibly suggest that banks increasingly play a lesser important role, in the corporate governance of firms, than in the past.

Further results in Model 7 show a significant and positive coefficient for the dummy variable Pre-2005 (β = 2.45; p < 0.001) and none for Post-2005. These results suggest that the adoption of the committee system may have been influenced by governance recommendations made in 2002, which were specifically directed at the board system. On the other hand, the broad governance reforms of 2005 had no effect on the adoption of the committee system. Adding time variables contributes significantly to the amount of explained variation (a significant, p < 0.001, chi-square difference of 23.12).

Consequently, the last model focuses on the interaction effects between all the independent variables and the time dummy Pre-2005. Interaction variables Cross listing × Pre-2005, Firm age × Pre-2005, and Cross holdings × Pre-2005 are positively and significantly associated with the adoption of the committee system at (β = 2.11; p < 0.05) (β = 0.03; p < 0.001), and (β = 0.08; p < 0.001), respectively. In additional analyses, we created and tested interaction terms between independent variables and the dummy variable Post-2005. However, none of these interaction terms reached statistical significance.

Given the relatively small number of firms that adopted the committee system, we carried out robustness checks by running some analyses using rare events logistic regression.^[1] The findings are not significantly different from the results shown in Table IV.

DISCUSSION AND CONCLUSIONS

It has been argued by economics-based institutional theorists that economic crisis may produce ‘a sharp break from established procedures. . . . rare windows of opportunity to effect broad reforms are thereby opened’ (Williamson, 2000, p. 598). Consistent with Williamson's view, we have studied, in this paper, at firm-level, the adoption or non-adoption of a management practice, following sharp legal and economic changes in the external environment. Indeed, the governance mechanism of a committee system, already standard practice in American-style governance, was the product of corporate law reforms in Japan following the so-called lost decades.

This study thus provides empirical support for the institutional inertia or change perspective at the national level for the adoption of the committee system, a practice deriving from the Anglo-American system and translated to the Japanese model. Drawing on a framework that is informed by institutional theory arguments, demonstrating the effect of macro-level changes on the behaviour of firms, most of the hypotheses are supported. First, our findings show that Japanese firms that are internationally exposed through cross listing are more likely to adopt the committee system. From an institutional theory perspective, cross listed firms are guided by both coercive and mimetic pressures (DiMaggio and Powell, 1991), complying with SEC regulations and imitating management practices from US firms including the use of the committee system.

Second, we found partial evidence that previous financial performance is negatively associated with the adoption of the committee system. Thus, the notion that poor performance encourages change is supported to a certain extent, possibly suggesting that many firms, in the study, do not perceive the economic benefits of adopting the committee system. Of course, the measurement of performance is open to several interpretations and may vary from context to context. In stakeholder systems, of which Japan is a good example (Dore, 2000), other measures to evaluate firm success (e.g. the number of jobs created or lost) may be preferred to profitability.

Third, we found that more experienced firms are more likely to adopt the committee system. While previous studies found that older firms are resistant to change (Sherer and Lee, 2002), our finding suggest the opposite. In the light of the circumstances that led to corporate governance reforms in Japan, our findings are justifiable. The economic stagnation that Japan experienced prior to governance reforms dictated that all firms needed to adapt to governance changes. Such a situation meant that change became a function of firm capability and resources. Here, more experienced firms are seen to be more capable to change given their experience and enabling longstanding networks.

Fourth, we found that firms with higher proportions of cross holdings are more likely to adopt the committee system. The context of this finding is that firms with higher proportions of cross holdings have the opportunity to appoint outside directors from related firms. Indeed, switching to the committee system requires appointing of outside directors, and firms with cross holdings have a greater pool of outside directors. This arrangement does not only fulfil the legal expectations but provides the parent company with an opportunity to strengthen ties with and coordinate activities in affiliated firms through outside directors, who represent their respective firms.

Fifth, we found that foreign ownership is positively associated with the adoption of the committee system while bank ownership has the opposite effect, however, limitedly. Our findings on foreign ownership are consistent with a number of previous studies. As discussed earlier, foreign ownership in Japan has been found to be positively associated with shareholder value oriented practices such as downsizing (Ahmadjian and Robbins, 2005) and appointment of outside directors to the board (Chizema and Kim, 2010).

Our findings on bank ownership possibly suggest the declining role of banks in corporate governance, consistent with the assertion by Gedajlovic et al. (2005) that due to globalization banks are cutting back their conventional ties to keiretsu firms. The same has been observed in Germany, a country close to Japan in corporate governance terms, where there has been a change in direction by banks towards more investment banking and less interest in corporate monitoring (Sanders and Tuschke, 2007).

The study also shows the moderating effects of a regulatory change that was specifically targeted at reforming the board committee system in 2002. While the Post-2005 reforms were broad, the Code recommendations made in 2002, and meant to take effect from 2003 were effective as suggested by the results of this study. Further analysis shows that the effects of the 2002 recommendations were more pronounced in older and cross listed firms as well as in ones that had higher proportions of cross holdings. It is interesting to note that foreign owners and banks were not influenced by the regulatory changes. This is probably because the two forms of ownership represent very distinct but opposite interests that do not need any form of motivation to manifest.

This study makes theoretical and empirical contributions to research on the adoption governance mechanisms. First, our paper refrains from using the often-criticized agency approach, applying instead a variant of institutional theory to an understanding of governance reform (Aguilera and Jackson, 2003). Second, our paper applies statistical tools of event history analysis to a relatively long period of observation, allowing any effect of legal and regulatory changes to be detected.

Third, through the effects of firm-level characteristics on the adoption or non-adoption of a governance practice, this study enriches an understanding of both change and continuity and the underlying contrast between the two, consistent with the contest between path dependence and radical change (Townley, 2002). The adoption or non-adoption of the committee system in Japan provides a litmus test of the attractiveness of corporate governance practices drawn from the Anglo-American model. Here, the contrasting style of Japanese capitalism vis-à-vis the American variety, the strong embeddedness of her institutions and, on the other hand, the likelihood of adopting the committee system by some firms alongside the potential rejection of the management practice by others, makes Japan, through some of her firms, a valuable laboratory to study institutional changes in corporate governance.

In addition to the theoretical and empirical contributions, this study brings new implications for practice and policy. Accepting the agency theoretical view that the committee system strengthens the supervisory role of the board (Dahya and McConnell, 2007), it is in the interests of shareholders and the economy as a whole that the practice is embraced by as many firms as possible. However, under the current legal and regulatory conditions, the majority of firms are less likely to adopt the committee system in Japan (Gilson and Milhaupt, 2005). While most of the adoptions took place in the early stages, the majority of firms that are not cross listed, have lower levels of foreign ownership and large bank ownership may choose continuity to change. Thus, creating an environment that encourages further foreign ownership and cross listing on American stock exchanges may increase the chances of adopting the committee system.

A further policy implication that can be drawn from this study is that some changes in the Japanese context are highly contestable and that, depending on the nature and importance of such changes, a mandatory style of enforcement may be necessary. At the macro level, this observation partly dismisses any suggestions that Japanese corporate governance is naturally converging on the Anglo-American variety of capitalism, at least for now.

Limitations and Further Research

Notwithstanding the relevance and timeliness of this study, we identify some limitations and suggest avenues for further research. First, while this study provides insight into the current challenges of corporate governance in Japan, the unique circumstances of the study context may not allow the chance to generalize our findings. Indeed, the developments leading to these changes are country-specific, and Japanese corporate governance, shaped by its path-dependent national culture and institutions is in many ways unique. Second, we used indirect measures of institutional effects which may not represent institutional processes. We therefore employ measures that arguably represent evidence of their effects on the adoption of the committee system. Third, our sample is restricted to listed firms since corporate governance reforms in Japan were targeted at this group of firms. Thus the adoption of the committee system is common in listed firms. Although there is still variation among these firms in terms of both size and experience, our understanding of firm behaviour, in this case, does not extend to small firms or the so called ‘born globals’. Consequently, a full and wider investigation of the adoption of the committee system is still awaited.

In the context of the same institutional environment and in the event that more firms adopt the committee system, we identify potential areas for future research. Future work could investigate the characteristics and level of independence of outside directors in Japan. It would therefore be interesting to make comparisons with studies of corporate governance in other Asian countries such as South Korea where outside directors have only been recently introduced in the boardroom (Chizema and Kim, 2010).

Previous research on board committees discusses the determinants of the audit, remuneration, and nomination separately. While the concept of the overall committee system is still at infancy level in Japan, future research could consider the structure and processes of the nomination, remuneration, and audit committees. Such a study would be important as it would be a step towards board committee evaluation and would allow comparison with previous studies that have been carried out in other countries.

ACKNOWLEDGMENTS

We acknowledge Emeritus Professor Trevor Buck, the general editor Professor Andrew Delios, an anonymous editor, and three JMS reviewers, as well as seminar participants at Cass Business School, London, for comments and suggestions on earlier versions of this paper.

NOTE

• [1] 

  Rare events logistic regression is used to estimate and interpret logit results when the sample is unbalanced, i.e. one outcome, in the dependent variable, is rarer than the other (King and Zeng, 1999).