Research Paper Family Business Performance in the Japanese Stock Markets

Kerry Hammond* Undergraduate Student, Harvard University

Abstract In this article, I discuss the results of my analysis of Japanese family firms. I investigated the relationship between family equity ownership and earnings performance among the member companies of the Tokyo Stock Exchange’s Nikkei 225. Japan provides a compelling landscape for this analysis because of the maturity of its financial markets where western corporate models and local business practices have been intertwined since the Meiji Restoration of the 19th century. The results of my statistical analysis demonstrated a positive relationship between family firm ownership and earnings performance and support that family firms outperform in the Japanese market. These findings demonstrate that family businesses do perform differently from non-­family firms in both markets and are relevant to shareholder value creation through earnings performance. These findings were contrary to my hypotheses that family firms would underperform non-family firms among Nikkei 225 members because of perceived or actual corruption, nepotistic behaviors, and poor talent development programs that are associated with the supposed less objective and strategic management of family businesses. However, family firm outperformance is supported by several strategic benefits that the family organization confers on businesses. For example, family firms could outperform by creating shareholder value through long-term strategic management and aligning family manager and family owner interests in a way that non-family firms do not.

Keywords: family business, family firm, family ownership, firm performance, stock performance, shareholder value, Japan.

* Corresponding author e-mail: [email protected]

91 วารสารญี่ปุ่นศึกษา Japanese Studies Journal

1. Introduction tential to outperform non-­family businesses. It was the goal of this investigation The significance of family firms may be to examine the relationship between family partially explained by the strategic alignment business ownership and firm performance and of management and ownership interests within contribute to the existing body of research the family as well as the long-term strategic on family businesses. Family business firm business and succession planning within these performance is relevant to how companies aligned interests (Sharma, 2004). create shareholder value. Whether or not This investigation considers the family firms are able to perform differently relevance of family businesses in dynamic from non-family firms is relevant to company economies, where globalization and efficiency stakeholders and investors, as this information are raising the standard of performance for could reshape the way stakeholders evaluate companies worldwide. Family businesses have the value propositions of their investments. a storied history in East Asia. If the family is The study uses data from publicly-traded where the modern, diversified corporation came firms listed on the Japanese stock markets to be (Allouche et al., 2008; Suehiro, 2001; from 2007 to 2013, exploring new market Suehiro and Wailerdsak, 2004), then does the conditions before, during, and after the 2008 future of the modern conglomerate still have Global Financial Crisis (GFC). I use the a place for families in business? If so, how percentage of equity share ownership by do these family businesses perform in the free founding family members and their heirs to market? Companies continue to be challenged define family businesses in this investigation. by the pressure to remain competitive in Prior research (Allouche et al., 2008; Saito, developing financial markets under constant 2008; Anderson and Reeb, 2003) also regulator and corporate governance reform. considers family businesses with family Better understanding of family businesses in managers. In my study design, research question, the context of dynamic economies can inform and hypotheses I focus only on family researcher and investor understandings of the ownership to explore whether ownership alone role families play in global markets. is a sufficient condition for firm outperformance This research specifically questions or underperformance, regardless of manager how family businesses perform relative to status. I found, contrary to negative non-family businesses. Several studies have perceptions of family firms like nepotism found that there is a relationship between and concentrated ownership (Suehiro, 2001; family businesses and performance in the market; Suehiro and Wailerdsak, 2004; Anderson and however, there is debate as to whether the Reeb, 2003), family businesses have the po- legacy and organization of family businesses

92 Family Business Performance in the Japanese Stock Markets Kerry Hammond leads to a positive or negative difference in organization (Anderson and Reeb, 2003). earnings performance of these businesses. Concentrated shareholders have the potential This investigation contributes to the debate to pursue private benefit at the expense of on family firm performance in the specific other shareholders and firm performance geographic context of Japan by using earnings (Shleifer and Vishny, 1997). Their findings data from large, multinational corporations suggest that shareholder concentration can also listed on the stock markets. In Japan, the be understood as a proxy for control over the family business narrative is uniquely paired firm, even if shareholders are not managers. with a cultural and ideological context that was Anderson and Reeb (2003) found that conductive to securing profits and succession family firms in the S&P 500 perform better through the family business organization, than non-family firms when defining family several of which still exist in evolved forms firm through family equity stake or presence across the publicly traded market in the of family managers and firm performance country. through return-on-assets, or the share price 1.1 Literature Review of Family return per unit of company asset. In the Business case of Japan, Allouche et al. (2008) found This investigation was motivated by that family businesses performed better than two primary perceptions of family business. non-family businesses using measures including Firstly, I am interested in whether or not return on equity, return on invested capital, family businesses are less efficient than and return on assets. They also note that the non-family businesses because of their con- very public difficulties that well-­known family centrated ownership. Secondly, family firms businesses like Daiei experienced during the remain a pervasive and visible type of business Asian Financial Crisis (AFC) made an entire around the world; by improving our understand- subgenre of businesses – all family businesses ing of family businesses, market participants – appear old-fashioned, rigid and ineffective can make more informed investment decisions. in their control and strategy due to family The concentration of equity ownership presence. Suehiro (2001) found that family sometimes exhibited by family businesses firms were not a significant contributor to the contributes to a perception that family businesses Asian Financial Crisis in Thailand, which is are a less efficient organization than dilute contrary to the perception that family busi- ownership of public companies. It has been nesses are a less efficient organizational struc- argued that family businesses are “unique” in ture. A later study by Suehiro and Wailerdsak their concentrated control (Saito, 2008) and (2004) demonstrated that certain types of are perceived as a less efficient economic diversified family firms in Thailand survived

93 วารสารญี่ปุ่นศึกษา Japanese Studies Journal the crisis better than other family firms. Based allows us to improve our understanding of on examples such as this, I explore whether the characteristics of different economies. The Japanese family firms may still outperform potential differences between the perception of despite the financial crisis and reputational family firms and their empirical performance harm they experienced following the AFC. and how these perceptions and performance Family businesses continue to be a can vary in the Japanese context justifies visible type of business around the world. further inquiry into this field. Anderson and Reeb (2003) found that 1.2 The Legacy of Family Businesses families are present in one-third of the S&P in Japan 500 companies, and Faccio and Lang (2002) The historical narrative of economic that 44% of Western European firms are development in Japan allows observers to controlled by families. According to a 2004 understand Japanese family businesses across review of current research on family firms, the different generational contexts, where family most commonly cited reason for investigating firms have adapted, evolved, and succeeded family business is the “observed dominance of across social, political, and economic changes these firms on the economic landscape of most into the modern era. The dynamism of family nations” (Sharma, 2004). For example, a 2006 businesses over time suggests when and where study on the S&P 500 found that announce- they have been relevant in the economic ment of an external CEO led to a subsequent history of a particular market. Family businesses share price increase, while the announcement were fundamental in the early development of another family manager led to a share price of close-knit, vertically integrated distribution decrease (Perez­-Gonzalez, 2006). In Thailand, channels that kept marketplaces such as Tsukiji widespread family involvement lowers firm fish market in Tokyo humming day-to-day, performance (Bertrand et al., 2008). In the dominated by families controlling a portion of case of family businesses where the founding the sourcing or distribution channels for their family is the largest shareholder, smaller catch (Bestor, 2004). Before World War II, shareholders could be subject to the implicit the big four zaibatsu, or family conglomerates, control that results from a dominant equity Sumitomo, Mitsui, Mitsubishi, and Yamada, owner. Investigating family businesses across controlled 76% of the total paid-in industrial different country contexts is significant because capital of Japan (Yamamura, 1967). Vertically family firms can be expected to operate integrated keiretsu, which replaced zaibatsu differently in unique economic contexts. As after they were disbanded in the post-­war Bertrand and Schoar (2006) discuss, observing period, capitalize on family relationships as how family firms behave in different markets sources of reciprocal information flow and

94 Family Business Performance in the Japanese Stock Markets Kerry Hammond long-term stability (Bestor, 2004). However, family system also supported the idea that the these structures have evolved over time, now entire nation was one family unified under the including both family and proxy-family structures emperor, and collective values and obligatory that achieve similar levels of collaboration, relationships formed the basis of relationship accountability, and vertical integration that building (Yoshino, 1968). the historical zaibatsu achieved. In Japan, the development of commerce The Japanese corporation is derived with legal protections began in the 1600s. By from the confluence of elements of western the 20th century, zaibatsu firms dominated the corporate governance practices and historical pre-war economy. The zaibatsu system was Japanese tradition. Bhappu (2000) argues the development of Japanese corporations that the Japanese family organization was during the feudal Tokugawa era into the fundamental to shaping the Japanese corporate Meiji era where the government decided to structures of apprentice and master, where sell enterprises to favored wealthy families, pseudo-­family relationships help establish providing these groups with a lasting, dynastic, expectations and behavioral norms. This is advantage (Yoshino, 1968). The firms were consistent with the argument of Colli (2003) characterized by close ties to the political that the legal and economic risk associated elite in the Japanese oligarchy and benefited with asymmetric information between business from preferential policies and privileged status. parties was part of the historical basis of Early beneficiaries like Mitsui and Mitsubishi family businesses in Japan. Before the Meiji were founded in the 17th century (Mitsui Restoration in 1868, a turbulent economic Ginko, 1926; Yamamura, 1967), and became environment combined with a legal system diversified conglomerates that benefited from unable to secure property owner shippositioned favorable economic policies throughout the the family organization as an ideal way to Meiji era until World War II (Bisson, 1954). safeguard business interests (Colli, 2003). The Indeed, Mitsui and Mitsubishi would become benefits of economic security that family units the top two zaibatsu until the war (Hirsch- conferred upon businesses in conjunction with meier and Yui, 1975). the Confucian-inspired bushido ethics code 1.3 Japanese Corporations in the of the Tokugawa era (the Tokugawa encom- Post-War Period passes the feudal period Pre-Meiji Restoration The zaibatsu were forcibly dissolved in 1868) emphasizing central teachings of under the Yasuda Plan after 1945 through loyalty and filial piety created an environ- elimination of zaibatsu­-controlled directors and ment conducive to the success of family firms managers in subsidiary firms, divestiture of (Yoshino, 1968). During the Meiji era, the security holdings, eliminating non­competitive

95 วารสารญี่ปุ่นศึกษา Japanese Studies Journal contracts, and concentrating monopoly­size We can observe their continued evolution of companies (Bisson, 1954). However, the family presence in the Tokyo Stock Exchange zaibatsu system was entrenched in the economic listings today, where only 2% of the Nikkei institutions of Japan, and despite the forced 225 members have founding family owners. dissolution of the zaibatsu after World War II, The legacy of the zaibatsu system Japanese corporations adopted new structures and native Japanese business culture is still that could act as proxies for the benefits relevant in today’s economy. Yoshikawa et al. previously conferred by the zaibatsu system (2007) find that there is no clear convergence through horizontally and vertically affiliated to or divergence from Anglo-American models firms called keiretsu (Yoshino, 1968). These of corporate governance. Instead, their findings firms can confer benefits to one another, such support that Japanese firms have selectively as risk diversification by acting as subcon- adopted features from western models and tractors to larger firms, specialization, and applied them to fit their context, and have contracts securing future business. been successful in doing so. Marshal (1967) The Doyukai was an organization explores the integration of “Anglo-American founded by 70 executives to create a unified capitalist creed” with “traditional Japanese managerial ideology in the post­-war period. By values,” citing businessmen who expressly the 1960s, these guidelines included a new reject profit for personal gain. Instead, the emphasis on independence and self-determina- post-war­ business class claimed that Japanese tion. This new emphasis on self-determination businessmen were motivated out of selfless- was contrary to the previous orientation of ness, patriotic devotion, and a willingness to business organizations. The Doyukai stressed sacrifice for the common good by contributing the primacy of functionalism and profession- to the industrialization of Japan. alism over the paternalism of the zaibatsu A key feature of the zaibatsu system system and affirmed the pursuit of profit; in corporate governance model was the separation contrast, Meiji-era businesses had deempha- of ownership and control, hiring professional sized profit as the central goal of business managers to run day-to-­day operations and (Yoshino, 1968). These pronouncements were subjecting family owners to the expertise of readily integrated into by Japanese corporate a board of directors. This resulting separation practice, adopted from the “managerial strand maintained personal loyalty to the family with of the American business ideology” (Yoshino, “clear-headed managers” (Hirschmeier and 1968). As a result, family firms became less Yui, 1975). This unique blend of capitalism prominent as influential forces within the into a pre­existing cultural context is indicative Japanese economy in the post­-war period. of the evidence found by Suehiro and

96 Family Business Performance in the Japanese Stock Markets Kerry Hammond

Wailerdsak (2004) that Japanese firms were According to Sheng (2009), the particularly successful at importing and Japanese role in the Asian Financial Crisis adapting Anglo-American corporate govern- has been understated, with the primary victims ance than their counterparts in Thailand, for of the Crisis being Thailand, South Korea, example. The development of a Japanese Indonesia, Malaysia, and greater China. Sheng model of corporate governance is visible in describes how the Japanese zero interest rate the continued prevalence of family firms in policy in place to combat the deflation of the the local stock market today. yen contributed to the trade conditions that led 1.4 Economic Globalization to currency bubbles in other Asian markets. In the post­-war period, family Receiving markets like Thailand, Indonesia, businesses in Japan saw their zaibatsu system and Malaysia ran trade deficits because of dissolved, bringing about a period of transition Japanese foreign direct investment in the for the formerly politically and economically region during the 1980s and 1990s until the dominant conglomerates. Instead of thoroughly Japanese yen began to depreciate against the ingrained vertical business organizations, dollar in 1995. Japanese banks pulled invest- horizontal webs were integral to success ment from other Asian markets, accelerating in the post-­war period. Keiretsu structures after the devaluation of the Thai baht. In part, developed in the post­-war period through the Asian Financial Crisis was directly related cross-shareholding and interlocking directo- to the withdrawal of funds by Japanese banks. rates as a way for firms to develop mutually Organizational evolution like the transition of beneficial patterns of allocation and return at zaibatsu into keiretsu complicates the legacy the expense of the competitive market (Lincoln of family firms in Japan as the corporate et al., 1996). Hideto and Haley (1983) environment has become increasingly integrated observe that keiretsu groupings joined by and sophisticated, with controls that improve anything from cross­-shareholding to personal stakeholder protection as well as institutions relationships at the executive level dominate like holding companies and trustee banks to Japanese trade, and that goods not produced or protect insider and external shareholders. The handled through keiretsu are nearly excluded Japanese role in the Asian Financial Crisis from the Japanese distribution market. Lincoln was likely complicated by keiretsu networks, et al. (1996) find that keiretsu system firms which can dictate the flow of capital into and exhibit a smoothing effect on their perform- out from Japanese banks that are affiliated. ance – independent firms may recover more 1.5 Hypothesis for the Case of Japan quickly from a downturn, but keiretsu system I hypothesized that family businesses’ firms appear to tax outperformers and earnings as measured by EBITDA (earnings guarantee the survival of troubled affiliates. 97 วารสารญี่ปุ่นศึกษา Japanese Studies Journal before interest, tax, depreciation, and create enduring relationships between producers amortization) will underperform non-family and distributors (Bestor, 2004), can actually businesses. Family structures today are less confer the structural benefits that used to be relevant to the performance of mature, public obtained through the proxy of family organiza- corporations where legal protections and tion in corporate management and ownership. succession can be secured by legislation and This hypothesis is contrary to the findings of professional managers. Whereas family ownership Anderson and Reeb (2003) on the S&P 500 and management used to be critical to main- and Allouche et al. (2008) and Saito (2008) taining control and securing company wealth on family firms in Japan. These authors found in the early stages of business growth and that certain types of family firms outperformed development, today these family structures are relative to non-family firms. not necessary. I hypothesize that family is no longer relevant to the relative performance of 2. Data and Methodology publicly traded Japanese firms because the The sample comprises earnings data protections and benefits once conferred by from the constituent members of the Japanese the family organization are now secured by Nikkei 225 index, reported from 2007 to 2013. legislation and financial regulation. The family Whether or not a company was a family business organization may be how businesses are was determined by examining current equity founded and grow, and this hypothesis does ownership data from Bloomberg and comparing not discount the importance family structures listed individuals with executive management, play in the early cultivation of a successful, executive board members, and founding family. diversified, multinational conglomerate. The This case-­by-case identification procedure was benefits of the complex keiretsu system can employed by prior research by Allouche et al. potentially penalize winners while rewarding (2008); Saito (2008); Suehiro and Wailerdsak losers, smoothing outcomes of family (2004) because of a lack of consensus for what businesses that are involved in the keiretsu defines a family businesses. After accounting system. for missingness, defined as where values for There is evidence to suggest that as variables in firm­-quarter observations were competitive pressures increase and markets not available, the final sample size was 3,489 become more developed and globalized, firm-­quarter observations from the Nikkei 225. alternative business organizational structures, In order to create quarterly lags up to 4 past such as a board of directors, an executive quarters (1 calendar year), the final samples management team (Suehiro and Wailerdsak, only included observations from 2008 to 2013. 2004), and patterns of vertical alignment that

98 Family Business Performance in the Japanese Stock Markets Kerry Hammond

2.1 Statement of the Research family managers. I focus on the former Question because of the relevance of equity shareholders 1. Do family-owned businesses out- to the market performance of the stock, where perform non-family-owned businesses using share price is my response variable of choice EBITDA as the measure of firm performance? for measuring firm performance. 2.2 Defining Family Businesses I utilize family equity share ownership No consensus definition of family as a proxy for control of the firm. Concen- business has been adopted for either private or trated ownership of outstanding equity can public firms (Sharma, 2004). Family influence allow family members to exert influence over encompasses a continuous spectrum of varying manager behavior by mitigating expropriation degrees of active and passive control, ranging (Demsetz and Lehn, 1985). The same influence from voting rights to the daily operations of a can be observed not only by family own- business. National laws and regulatory bodies ership, but by other large “blockholders” can affect what sort of family organizations (Anderson and Reeb, 2003) such as pension proliferate in a particular environment, such as funds or asset managers. Further, large family capital gains taxes or inheritance laws (Sharma, equity shareholders will retain significant 2004). Jaffe and Lane (2004) observe voting control and could continue to direct that many family firms never make it to the the strategic vision of the firm without active dynasty stage (post­-second­ generation). As family–member management in place. a result, I considered the Japanese market The converse situation is not true, because environments when determining how family managers will not be able to exert a comparable firms would be distinguished from non–family amount of influence over shareholders, who firms in this investigation, including the types could liquidate their positions and eliminate of businesses that are supported by each any financial risk they bear from the firm’s regulatory environment and the maturity of performance. Defining family businesses using the market. According to Bertrand and Schoar equity share ownership could result in a (2006), family firms are characterized by bias for outperformance of firms with high concentrated ownership, family control, and family equity share ownership. As Anderson often key management positions maintained by and Reeb (2003) discuss, a family might be the family. In this investigation, I define family more likely to retain shares if they expect businesses using equity share ownership by positive share performance because of future founding family members. Several other studies profitability or earnings growth. Conversely, a choose to define family businesses using either family might be more likely to sell or diminish equity ownership or the presence of key their equity ownership stake if they expect

99 วารสารญี่ปุ่นศึกษา Japanese Studies Journal future negative share price performance due as a sample and population of firms. Data to losses or negative earnings growth. were collected from Bloomberg, retrieved on In Japan, identifying family-owned 31 January 2014 Bloomberg L.P. (2014). All firms can be difficult due to the diversified statistics and results reported are from final positions of holding companies, trustee banks, data that eliminates firm-quarter observations asset managers, and investment banks across with missing values. Missingness (where the Nikkei 225 sample. Many Japanese firms observations are not available) is assumed to have a large main blockholder, like a holding occur at random. The sample of the Nikkei company or bank, as their largest shareholder. 225 members included 5,850 firm quarter For the case of Japan, I adopted a definition observations from 2007 to 2013, which was of family firms used by Saito (2008) to reduced to 4,346 observations following an identify family businesses where a founding analogous log-transformation to EBITDA. 6.2, family member is the largest shareholder. In Appendix: Justification for Excluding Negative addition, I identify firms with family business EBITDA, contains a summary of Welch’s legacies by encoding all Big Four Zaibatsu unpooled T-test results that demonstrates (Sumitomo, Mitsui, Mitsubishi, and Yasuda) the sample and family business groups are subsidiaries. Family businesses in Japan were not significantly different from one another encoded as a binary variable. Existing research with negative EBITDA values removed for (Saito, 2008; Suehiro, 2001; Suehiro and the purposes of this transformation. With the Wailerdsak, 2004) classify firms by type in exception of summary statistics, all data order to accommodate for performance analysis using multiple linear regression is differences stemming from the firm organization completed with the log-transformed EBITDA or succession pattern. I elect not to classify variable using non-­negative EBITDA family firms by type. In the data collection observations. process I attempt to control for various firm 2.3.1 Response Variables characteristics like firm size, firm age, and There are many different ways to firm valuation (for example, whether or not measure firm performance by assigning value the firm has traded at a historical premium). to or measuring the profitability of a company 2.3 Sample and Data using public, published financial data. I The data set utilizes financial data measure firm performance using accounting from the 225 member companies included in measures that are found on the company’s the Nikkei 225 index in order to represent the financial statements, earnings before interest, most influential and active public companies tax, depreciation, and amortization. When in the market by utilizing the benchmark index considering accounting performance reported

100 Family Business Performance in the Japanese Stock Markets Kerry Hammond on the financial statements, stock portfolio potential influence that institutional investors managers and analysts will be most concerned have on stock pricing. If an institutional investor, with a company’s ‘bottom-line’ earnings, such as a hedge fund manager, publicizes or after-tax earnings. Within an industry or his firm’s position on a particular oil and firm type segment, net income could be used gas company on Bloomberg, other investors instead to capture ‘bottom-line’ earnings; are more likely to emulate or consider the for many stock portfolio managers, they are position in their own portfolios. Further, most often making a buy or sell decision within institutional investors maintain analyst groups an industry or type of company, making that thoroughly research and seek out the this comparison appropriate. Because I am maximum return-on-investment positions for conducting an analysis across all industries their clients’ portfolios. However, institutional and firm types, I choose to examine ‘top-­line,’ ownership can also indicate a negative outlook or pre­tax earnings. The most common way on firm performance when stock sell-­offs of measuring this is using earnings before occur. Explanatory variables were included interest, tax, depreciation, and amortization in data collection with the goal of contribut- (EBITDA) because it allows for relatively ing to the explanation of firm performance simpler cross-sectional comparison since it in Japan. In reviewing existing research on eliminates the role of tax rates, financing, and family businesses, the Japanese stock markets, accounting decisions in earnings reporting. For and measuring firm performance and profit- the purposes of this investigation, EBITDA ability, I consider the following explanatory represents any reference to a firm’s profits. variables. I sought to capture equity share It is calculated by adding back interest, tax, ownership, industry, firm size, gearing (debt) depreciation, and amortization expenses to a ratios, and market valuation through these firm’s net income using the financial variables. All variables are continuous unless statements. For market performance, I collected otherwise indicated. share price data between January 3, 2007, and ● Family Business, a binary variable repre- September 27, 2013. I applied four quarters senting where a family member of the com- of lags on EBITDA as a response variable in pany’s founder is the largest shareholder order to control for serial correlation in the identified in the Bloomberg equity samples, resulting in the effective date range holders record. of the sample being reduced to 2008 to 2013. ● Blockholder Ownership, a binary variable 2.3.2 Explanatory Variables representing the a greater than 50% share Family equity share ownership is my of equity held by an institutional investor explanatory variable of interest because of the that can be identified via the Bloomberg

101 วารสารญี่ปุ่นศึกษา Japanese Studies Journal

equity holders record (e.g. hedge fund, ● Long-term Debt to Total Assets, a ratio pension fund, or holding company. The representing the company’s coverage of its 50% benchmark is adopted from the short­-term liabilities. This ratio could be methodology utilized by Saito (2008) in more normalized for one­-time expenses that identifying majority or controlling share- a company temporarily accumulates debt holders. for but subsequently pays off the following ● Trustee Bank, a binary variable representing quarter or year. It is computed by dividing where a trustee bank is the largest share- the long­-term debt by total assets. holder identified in the Bloomberg equity ● P/E Ratio, a ratio representing the company’s holders record. current valuation that can easily be compared ● Keiretsu, a binary variable representing to other companies on the market, allowing where keiretsu­-affiliated firms are the investors to evaluate whether a company largest shareholder identified in the is trading at a price more expensive (a Bloomberg equity holders record. premium) or cheaper (a discount) to peers ● Return­-on-­Assets (ROA), a performance based on profitability. It is computed by measure computed by dividing net income dividing share price by net income. for the quarter by total company assets. ● P/B Ratio, a ratio representing the company’s ROA allows shareholders to evaluate how current valuation that can easily be compared management is generating shareholder value to other companies on the market. It is per each unit of assets the company holds. computed by dividing share price by the ROA is consistent throughout the quarter. company’s book value. Instead of using I also utilize 3­-year and 5­-year average as net income, book value accounts for the a proxy for a lag on the ROA variable to firm’s asset value rather than profitability. control for serial correlation. 2.4 Statistical Methods ● Equity Shares Out, representing current 2.4.1 Modeling Techniques outstanding shares that can be traded freely Multiple linear regression is used in the stock market by foreign and domestic to explore the relationship between firm investors. accounting performance using EBITDA and ● Book Value per Share, representing how equity share ownership (R Core Team, 2012). much the company is worth according to Multiple regression is appropriate in this its assets and liabilities per share. analysis because there are sufficient robustness ● Total Debt to Total Assets, a ratio measures, including panel regression with representing the company’s coverage of its fixed effects and variable lags for serial liabilities. It is computed by dividing the correlation that can be employed to control total debt by total assets. 102 Family Business Performance in the Japanese Stock Markets Kerry Hammond for potential violations of the assumptions of Japan. Family firms on average earn more linear regression, including linearity, normality, among Nikkei 225 members in Japan. Overall, independence, and homoskedasticity (constant the Japanese sample contained 2.2% family variance) (Ramsey and Schafer, 1997). The firms among benchmark index members goal of this investigation is to draw associative, according to my definition of largest share- not predictive conclusions, regarding the holder a founding family member. relationship between family business and firm Family businesses comprise only 2.2% performance. My goal is to ensure that model of Nikkei 225 members. Japanese family firms assumptions are satisfied to ensure the validity were encoded as a binary variable because of of the least­ squares estimates by satisfying relatively more diversified equity ownership, linearity and serial correlation as best as greater presence of keiretsu or holding possible. Appendix: Regression Diagnostics companies as shareholders of corporate entities, illustrates the diagnostic plots for satisfaction and comparatively more opaque disclosure of of the assumption from the full and final ultimate beneficiaries of returns on shares held model specifications in the sample. by keiretsu or holding companies. The Nikkei 2.4.2 Final Model Selection 225 is annually re-balanced by the Nikkei The final model for interpretation group. Nikkei 225 members are selected for was selected using two primary criterion: sector balance from the First Section of the the Durbin-Watson test for serial correlation Tokyo Stock Exchange to act as a “barometer” (Durbin and Watson, 1950) and the Bayesian (Kurashina, 2003) for the Japanese economy. Information Criterion (Weakliem, 1999). The The family firms in the Nikkei 225 are in the former allows serial correlation in a least- auto parts and equipment, electronics, internet, ­squares regression model to be assessed, and retail, and telecommunications industries. the latter allows for model fitting with a criteria The summary statistics in the table below that accounts for the trade-­off between maxi- support including firm fixed effects to control mizing the likelihood function of the model for the large variance in EBITDAs exhibited while penalizing for the number of variables by Nikkei 225 members. In addition, binary included in the model. variables for other types of large blockholders 2.5 Summary Statistics were included and reveal that the largest Summary exploration of the aggregate shareholder of each Nikkei 225 member data reveals patterns that emerge among almost always can be categorized into a family family firms and non-family firms. In insider, trustee bank, holding company, or particular, the family firms exhibit different keiretsu. Only 60 firms out of 225, in fact, do industry affiliations and are different sizes in not have a largest shareholder of one of the

103 วารสารญี่ปุ่นศึกษา Japanese Studies Journal aforementioned classifications. Trustee banks with 105 firms out of 225 characterized by are the most common largest shareholder, this equity ownership pattern. Table 2.1 Japan Nikkei 225 Members Summary Statistics 2007-2013

3. Family Firm Performance Results (BIC) as the selection method. The BIC is In the case of Japan, I find that family approximated by BIC = −2 · ln(L)+ k·ln(n) businesses outperform non-family businesses. for large sample sizes of approximately n > 30 These results are in contradiction to my (Weakliem, 1999). Lower BIC values suggest hypothesis that family firms would underper- a better relative model fit when comparing form in Japan relative to non-family firms. models built from the same sample. In my This could be explained in part by existing analysis, I compare multiple linear regression research on family businesses in Thailand, models using BIC. where only certain types of family businesses In the univariate case, family businesses outperform non-family businesses (Suehiro earn more in Japan than non-family businesses, and Wailerdsak, 2004), and in Japan by the suggesting that there is a difference between long-term, strategic orientation of the family the firm types before controlling for other firms among Nikkei 225 members. In this background covariates or company-specific chapter, I examine these results in the effects. Using multiple linear regression, I univariate cases where I used Welch’s unpooled find that being a family owned business t-test for equal means, and then consider the is significant at the 0.01 level, indicating multiple linear regression case using a full that these firms perform differently than model with all variables retained and a final non-family businesses when controlling for model obtained using backward stepwise company-specific effects, blockholders, returns regression with Bayesian Information Criterion on assets for previous time periods, firm

104 Family Business Performance in the Japanese Stock Markets Kerry Hammond market size, book value, debt, and market significant predictors while retaining the valuation. In addition to the family business explanatory variables of interest–the family indicator being significant, using backward business indicator. When considering family stepwise regression and Bayesian Information explanatory variables in my final model, I Criterion for model selection, I find that the again observe the outperformance of family most significant predictor of log (EBITDA) businesses relative to non-family businesses in each case are a lag for EBITDA to control among Nikkei 225 members. for time dependence, return on assets, and 3.1 Japan book value of the company. I use these 3.1.1 Welch’s Unpooled T-test variables to build a final model of the most Results

Table 3.1 Japan Nikkei 225 Welch's Unpooled T-test Results

3.1.2 Welch’s Unpooled T-test 3.1.3 Multiple Linear Regression Interpretation Results In the case of Japan, family busi- Again, the model selection process nesses on average earn more than non-family was based on selecting a model that best businesses without controlling for background satisfied the assumptions of linear regression covariates like firm size, debt, and market while addressing serial correlation in the response valuation. In this case, the power of the statistical variable, EBITDA, as well as selecting the test could be improved with a more balanced most significant variables for a final model sample size. The Nikkei 225 sample only for interpretation using BIC. Among Nikkei contains 5 family businesses out of 225 total 225 members, it appears that there is greater firms. While this allows me to determine serial correlation as more of the EBITDA lags an exact mean for family business earnings are significant, suggesting that the current among Nikkei 225 members, the findings here earnings are dependent on past quarter reporting, may not be found to be statistically significant requiring both EBITDA lags and quarter fixed in a different sample of comparable Japanese effects to control for serial correlation and firms listed on the Tokyo Stock Market. minimize BIC.

105 วารสารญี่ปุ่นศึกษา Japanese Studies Journal

The table below contains the final in order to minimize BIC. These variables model selection using BIC and backward are left in the model in order to interpret the stepwise regression, maintaining binary indi- variable of interest, family business, in the cator variables FB and keiretsu in the model. context of keiretsu, which was significant at These variables would have otherwise been the 0.10 level. eliminated under backward stepwise regression

Table 3.2 Regressing Nikkei 225 EBITDAs using firm and time fixed effects with lagged EBITDA, backward stepwise

3.1.4 Multiple Linear Regression EBITDA. These results are noteworthy because Interpretation the family firms in the sample clearly outper- Due to the response variable EBITDA form non-family firms; these results would be is log-transformed, the significant family strengthened if they could be replicated if a business binary variable can be interpreted as new sample of firms from the Tokyo Stock having a multiplicative effect on the EBITDA Exchange or entire population of Japanese in the amount of eβ. Among Nikkei 225 firms listed on stock markets were included in members, the 5 family businesses outperform the investigation. Keiretsu, interestingly, is not non-family businesses at the 0.01 significance a significant explanatory variable in the final level when examining final model coefficients model selection. This is surprising given the in Table 19. Holding all else equal, a one-unit literature covering the extensive, far-reaching increase in the percentage of family ownership implications of horizontally inter-networked will result in an e2.0551-fold increase in and integrated firms across industries that dominate the economic landscape. 106 Family Business Performance in the Japanese Stock Markets Kerry Hammond

3.1.5 Multiple Linear Regression Selection Criteria Table 3.3 Nikkei 225 BIC Summary

Table 3.4 Nikkei 225 Durbin-Watson Summary

3.2 Conclusion: Results control could be less related to the bottom- The results of this investigation suggest line profitability of these companies. that family is positively associated with firm performance as measured by EBITDA in 4. Limitations and Discussion Japan. The results for the keiretsu indicator In this section, I examine the limitations variable are also noteworthy for having no of the study design and robustness of the significant relationship with explaining firm model results. Considering study design, I earnings, which appears contrary to extensive discuss the sample size as a limiting factor in literature on the importance of horizontally generalizing the results of the model and the networked firms. subjective nature of defining family business; These results are in contradiction to considering model robustnesss, I examine the my hypothesis that family firms would un- extent to which the assumptions of linear derperform in Japan. The findings imply that regression were satisfied and the potential the explanation for firm outperformance may role of missing data in the sample. Although continue to be relevant to firm profitability there are limitations on how generalizable in Japan. Family firms in the Nikkei 225 the results of this investigation are, I believe may have the long-term strategic guidance that my findings are defensible, rigorous, and necessary to succeed financially, while the corroborate the existing body of research on explanations for underperformance like nepotism listed family business performance in stock and seeking private benefits through family markets. 107 วารสารญี่ปุ่นศึกษา Japanese Studies Journal

4.1 Limitations Related to the sample size of the This investigation has four primary benchmark indices is the small proportion of areas of improvement that could strengthen the family businesses present in the Nikkei 225 results of future research, including improving at only 2% of the entire sample. With a total the sample size through a longitudinal study of sample size of 225 companies, the power of the entire population of Japanese firms listed statistical tests is compromised due to unbal- on the stock market, employing a consistent anced group sizes for family and non-family definition of family business in repeated studies, businesses. The family business coefficients ensuring model assumptions are satisfied, and from the Japan case multiple linear regression prudently addressing missing data. results are limited by the group size imba- 4.1.1 Limitations from the Sample lance. Statistical power is the probability of a Size test, like Welch’s unpooled t-test, correctly The use of the benchmark indices’ rejecting the null hypothesis of the test when members for Japan constitutes a subset of the null hypothesis is false (Cohen, 1992). the entire market deemed qualitatively by The null hypothesis of Welch’s unpooled t- investors to be an appropriate measure of test is that the difference in the two means overall financial market and macroeconomic is equal to zero, or that the means of the performance in the geography of interest two samples are the same, and compromised (Kurashina, 2003). Even though each stock power due to imbalanced group sizes could market creates benchmark indices to represent lead to a test that reports that the difference the macroeconomic climate, the criteria for a in means is significant when in the population public company to be included in a benchmark the difference is equal to zero. While in this vary from market to market. For example, investigation the sample (Nikkei 225) is also the Nikkei Corporation selects Nikkei 225 the full population of companies (Nikkei 225) members using a balance of industries and a of benchmark index members, relatively fewer size requirement. family businesses in the Japanese market in The definition of the sample using general could lead to incorrect conclusions the benchmark indices limits my findings to about family business performance if a sample being generalizable to the index members of of companies is taken for a statistical test. my sample. My results in Japanese Nikkei 225 4.1.2 Limitations from Defining samples could be strengthened by conducting Family Businesses an analogous analysis across the entire stock Research in family business is under market of Japan, allowing me to generalize constant development and improvement to the entire population of public companies. because of the iterative approach of researchers

108 Family Business Performance in the Japanese Stock Markets Kerry Hammond in study design and variable definition. was based on the question of whether family Research in different domiciles, geographies, ownership is a sufficient condition for control industries, and firm types requires an that could impact performance, regardless of appropriate definition of a family firm to be whether or not a family firm also has family constructed for the study design and question man- agers. For example, Saito (2008) of interest. This investigation was concerned found that firms family-owned and managed with the performance of family firms relative underperformed relative to non-family firms, to non-family firms, and as a result I drew while firms that were family-owned or managed inspiration from prior studies on firm per- by family members outperformed relative formance in financial markets (Suehiro, 2001; to non-family firms. This result motivated Suehiro and Wailerdsak, 2004; Anderson and my decision to examine only the ownership Reeb, 2003; Allouche et al., 2008; Saito, 2008) component of the family firm. The outper- in creating my definition for a family business. formance of family firms among the Nikkei I define family business in Japan as a firm 225 members demonstrates that it is possible where the largest shareholder is a founding that ownership is a sufficient condition to lead family member or descendant. A key limita- to this divergence in family firm performance. tion in all of the aforementioned studies is the Nonetheless, the lack of a consensus definition transparency, or sometimes lack thereof, of for listed family firms makes it difficult to published ownership data. Many listed firms in rigorously reproduce results and prove family Japan choose to utilize banks, keiretsu banks, firm outcomes as fact. Instead, each observa- or other financial institutions and holding tional study on family business performance companies as large equity shareholders, can contribute to the general rationale of the and it can obscure the researcher’s ability uniqueness of family business performance to determine who is receiving the capital relative to non-family businesses. gains from these outstanding shares. This 4.1.3 Robustness to Model As- will continue to be an area of improvement sumptions in family business research and will change The purpose of this investigation was as regulators alter their stances on transpa- to draw an association between explanatory rency of share ownership data. Prior research, variables and the response variable EBITDA, including Allouche et al. (2008), Anderson and to examine family business as explanatory and Reeb (2003), and Saito (2008), accounts variables of interest. The most critical violation for the generation of the shareholders as well of the assumptions of linear regression was as family member managers, and I chose not the serial correlation exhibited by the response to control for these factors. This decision variable EBITDA, a result of multi-year

109 วารสารญี่ปุ่นศึกษา Japanese Studies Journal firm-quarter observations for the same Linearity appears to be met in several companies. Using the Durbin-Watson test variables plotted against log (EBITDA), and for serial correlation in R (R Core Team, there is no indication of polynomial rela- 2012), I observed that there is significant tionships between log (EBITDA) and the serial correlation at the α = 0.05 level.6 This continuous explanatory variables. It appears was corroborated by autocorrelation function that homoskedasticity is violated in some of (ACF) plots generated in R, which indicated the response variables like return on assets that using a lagged response variable might or the valuation ratios, leading to incorrect be appropriate based on the periodicity of the standard error estimates. Nonetheless, because serial correlation. To control for this potential this investigation is concerned with associa- serial correlation, my final model featured tion, not prediction, I do not consider these additions of company fixed effects and fixed violations critical to the interpretation of the effect for time-quarter for Japanese firms. final model result and the explanatory variable Following the addition of these variables, of interest, family business. the Durbin-Watson test indicated that there 4.1.4 Limitations from Missingness was no further serial correlation. The Durbin- Any missing data was assumed to be Watson test for serial correlation is not at random. As a result, complete-case analysis without limitations, and a follow-up ACF (Little and Rubin, 1989) was pursued, plot indicates that certain past time-quarters discarding all cases where data was missing. are still significantly serially correlated at The creation of four quarters of historical the α = 0.05 level. The Durbin-Watson test EBITDA lags resulted in the removal of data can be misleading when incorporating more from the year 2007, making the effective than 5 regressors into the model, as is the sampling period from 2008 to 2013. Limitations case in this investigation (Savin and White, of complete-case analysis include the obvious 1977). Nonetheless, the Durbin-Watson test loss of information resulting from discarding is appropriate in cases where least-squares observations. For the purposes of this inves- estimates are sought, and while inferences tigation, missingness precluded the possibility may be compromised, the Durbin-Watson of including 2007 as a normalizing year in the test allows for the specification of a model context of firm performance before, during, for interpretation in this thesis. and after a financial crisis. Instead, the Besides the violation of independence remaining data captures the performance of of the errors, multiple linear regression is firms during and immediately after a financial robust to violations of normality except in crisis. Results from complete-case analysis can cases of extreme skewness or binary responses. be biased unless the missingness is perfectly

110 Family Business Performance in the Japanese Stock Markets Kerry Hammond at random (Little and Rubin, 1989). Exclud- sample. However, this inference cannot be ing the discard of year 2007 to create lags, generalized across the population of all listed possible violations of random missingness Japanese firms because the Nikkei 225 does include discarding negative EBITDA values. not represent a random sample of all listed These comprised less than 5% of the total firms (and cannot be balanced on background observations, but may not have occurred at covariates). While this study adds weight random. to the general validity of the argument that 4.2 Discussion: Implications for Japanese family firms outperform non-family Listed Family Businesses in Japan firms, it cannot rigorously prove or disprove In this thesis, I explored the relation- this claim (Smith, 1983). ship between family firms and performance. For the population of all listed Japanese The results of this investigation support the firms, I argue that my results support the continued relevance of family firms in the existing evidence that family firms are significant market and suggest that the family organiza- and outperform relative to non-family firms tion continues to play a significant role in and that they have external validity beyond the financial outcomes. This information can in- scope of this study based on the results from form investor decision-making when gauging Allouche et al. (2008) and Saito (2008), who how a potential investment might generate also found outperformance of family firms in returns. Family businesses generate returns the Nikkei 225. While there are limitations differently from non-family businesses, and to the power of the t-test and multiple linear this can be identified in part by utilizing regression because of the small number of publicly available equity share ownership data. family firms relative to non-family firms in Certain kinds of family businesses in the Nikkei 225, the results observed here are the Japanese stock markets have been shown compelling and corroborate prior findings of to outperform non- family businesses using other research (Allouche et al., 2008; Saito, different response variables by Allouche et al. 2008). Directions for future research may (2008) and Saito (2008), and these results include expanding the sample size to include have been further corroborated using EBITDA the entire Tokyo Stock Market and subsetting for performance and the Nikkei 225 as the the definition of family business by controlling firm sample. For the population of Nikkei 225 for different variations in ownership or members, being a family firmis a significant management structures (e.g. founding family indicator of performance, and these firms manager or descendant manager, descendant do outperform non-family businesses in the ownership).

111 วารสารญี่ปุ่นศึกษา Japanese Studies Journal

4.3 Conclusion of how family may become less relevant as The results of my investigation show the economic policy and financial institutions that family firms outperform non-family firms mature to provide less need for the structural among Nikkei 225 members. My results and control advantages of maintaining the highlight the unique performance outcomes of family unit. family businesses and demonstrate that family Family, in reputation, organization, businesses do indeed perform differently from and control, is a key component of creating and better than non-family businesses. The shareholder value, and should be considered results indicate that there is a relevant economic by investors and researchers when evaluating context for the success of firms based on company performance. As shown in my research, their family structure and are inconsistent family ownership can lead to firm outper- with my hypothesis that family firms would formance or underperformance, and family underperform in Japan. The greater profit- ownership should be investigated to determine ability of family firms in Japan’s Nikkei 225 its relevance in the context of interest. My could be explained by the strategic focus of results demonstrate that family ownership is firms where family members are the largest not necessarily better or worse for companies, shareholder and can control the firm’s business but has the potential to be beneficial in the decisions to optimize for long-term profit- right economic and organizational conditions. ability, for example. This research was inspired by the References influential, storied, and ongoing narrative of Allouche, J., Amann, B., Jaussaud, J., and family firms in the Japanese economy, and Kurashina, T. (2008). The impact of based on the observed difference between family control on the performance and listed family firms in the Japanese stock financial characteristics of family businesses market and their performance relative to in Japan: A matched-pair investigation. firms with large blockholders or other types Family Business Review, 21(4):315–329. of shareholders, it appears that family firms Anderson, R. C. and Reeb, D. M. (2003). could be at a beneficial stage in the cycle Founding-family ownership and firm of corporate business development and the performance: Evidencefrom the S&P 500. foundation of a successful corporation. The The Journal of Finance, 58:1301–1328. proportion of family firms in the Tokyo Bertrand, M., Johnson, S., Samphantharak, Stock Exchange’s benchmark index could be K., and Schoar, A. (2008). Mixing family suggestive of the prominence of family in with business: A study of Thai business the economy at large, and is also indicative groups and the families behind them. NBER Working Paper Series. 112 Family Business Performance in the Japanese Stock Markets Kerry Hammond

Bertrand, M. and Schoar, A. (2006). The role Faccio, M. and Lang, L. H. (2002). The of family in family firms. Journal of ultimate ownership of western European cor- Economic Perspectives, 20:73–96. porations. Journal of Financial Economics, Bestor, T. C. (2004). Tsukiji: The fish 65(3):365–395. market at the center of the world. Gregory, A. (1982). The Japanese economy and Berkeley, California: University of California business system: Patterns and influences Press. upon growth. Japanese-American Business Bhappu, A. D. (2000). The Japanese family: Administration and Cultural Studies An institutional logic for Japanese corpo- Program, Graduate School of Business rate networks and Japanese management. Administration, New York University. The Academy of Management Review, Hideto, I. and Haley, J. O. (1983). Anti- 25:109–115. competitive practices in the distribution Bisson, T. (1954). Zaibatsu dissolution in of goods and services in Japan: The Japan. Berkeley, California: University problem of distribution keiretsu. Journal of California Press. of Japanese Studies, 9:319–334. Bloomberg L.P. (2014). SET 50 and Nikkei Hirschmeier, J. and Yui, T. (1975). The 225 data from 01/03/07 to 06/01/13. development of Japanese business. London: Retrieved December 11, 2013 and January George Allen &Unwin. 31, 2014, from Bloomberg database. Jaffe, D. T. and Lane, S. H. (2004). Sustaining Cohen, J. (1992). A power primer. Psycho- a family dynasty: Key issues facing mul- logical Bulletin, 112(1):155. tigenerational business and investment- Colli, A. (2003). The history of family owning families. Family Business Review, business, 1850-2000. New York: Cambridge 17:81–98. University Press. Kurashina, T. (2003). Family kigyono keieigaku Demsetz, H. and Lehn, K. (1985). The (Management studies on family business). structure of corporate ownership: Causes Tokyo: Keizai Shimbunsha (in Japanese). and consequences. The Journal of Political Lincoln, J. R., Gerlach, M. L., and Ahmad- Economy, 1155–1177. jian, C. L. (1996). Keiretsu networks and Durbin, J. and Watson, G. S. (1950). Test- corporate performancein Japan. American ing for serial correlation in least squares Sociological Review, 61:67–88. regression. i. Biometrika, 37(3-4):409– Little, R. J. and Rubin, D. B. (1989).The 428. analysis of social science data with missing values. Sociological Methods & Research, 18(2-3):292–326.

113 วารสารญี่ปุ่นศึกษา Japanese Studies Journal

Marshal, B. K. (1967). Capitalism and 2000s. New York: University of Cam- nationalism in prewar Japan: The ideology bridge Press. of the business elite, 1868-1941. Stanford, Shleifer, A. and Vishny, R. W. (1997). A California: Stanford University Press. survey of corporate governance. The Mitsui Ginko (1926). The Mitsui Bank, a brief Journal of Finance, 52:737–783. history: Jubilee commemoration. Tokyo: Smith, T. M. F. (1983). On the validity of Mitsui Bank. inferences from non-random sample. Perez-Gonzalez, F. (2006). Inherited control Journal of the Royal Statistical Society. and firm performance. The American Series A (General), 394–403. Economic Review, 96:1559–1588. Suehiro, A. (2001). Family business gone R Core Team (2012). R: A language and wrong? Ownership patterns and corporate environment for statistical computing. R performance in Thailand. Tokyo: Asian Foundation for Statistical Computing, Development Bank Institute. Vienna, Austria. ISBN 3-900051-07-0. Suehiro, A. and Wailerdsak, N. (2004). Family Ramsey, F. L. and Schafer, D. W. (1997). business in Thailand: Its management, The statistical sleuth: A course in methods governance, and future challenges. ASEAN of data analysis. Belmont, California: Economic Bulletin, 21:81–93. Duxbury Press. Weakliem, D. L. (1999). A critique of the Saito, T. (2008). Family firms and firm Bayesian information criterion for model performance: Evidence from Japan. The selection. Sociological Methods & Research, Journal of the Japanese and International 27(3):359–397. Economies, 22:620–646. Yamamura, K. (1967). The founding of Mit- Savin, N. E. and White, K. J. (1977). The subishi: A case study in Japanese business Durbin-Watson test for serial correla- history. The Business History Review, tion with extreme sample sizes or many 41:141–160. regressors. Econometrica: Journal of the Yoshikawa, T., Tsui-Auch, L. S., and Econometric Society, 1989–1996. McGuire, J. (2007). Corporate governance Sharma, P. (2004). An overview of the field of reform as institutional innovation: The family business studies: Current status and case of Japan. Organization Science, directions forthe future. Family Business 18:973–988. Review, 17:1–36. Yoshino, M. (1968). Japan’s managerial Sheng, A. (2009). From Asian to global system: Tradition and innovation. Cambridge, financial crisis: An Asian regulator’s view Massachusetts: The MIT Press. of unfettered finance in the 1990s and

114