Family-Owned Firms and Stock Returns: Evidence from the Chinese Stock Market

We examine stock return performance of Chinese family-firms vs. Nonfami-ly-firms, and the impact on family-firm returns of firm size, having a founder as CEO, and levels of family ownership. We model returns with the Capital Asset Pricing Model (CAPM), the Fama-French 3-factor model [1], and a new 5-factor model we modified from Miralles-Marcelo, Miralles-Quirós [2], which adds factors for debt and illiquidity. In previous studies, though the prevalence is for family-firm outperformance, the results have been mixed. Using CAPM or the Fama-French 3-factor model, we find that family-firms outperform nonfamily-firms, but with the 5-factor model that outperformance vanishes, and this is robust to controlling for the GFC. For the founder CEO, firm size and family ownership considerations, again previous research has been mixed. With CAPM and Fama-French 3-factor models, we find that founder CEOs outperform nonfounder, large firms outperform small firms, and higher family ownership outperforms lower ownership. However, with the 5-factor model, only size remains as having significant impact. This study helps to reconcile conflicting results in previous research.


Introduction
Since 1978, China's economy has undergone a huge transformation from being a planned economy to a market-oriented economy. According to the Chinese Family Business Survey and Analysis Report (2014), this market-oriented economy has encouraged development of a free market and the number of private firms increased dramatically between 2009 and 2015, from 6.64 million to There are two main motivations for why family firms might outperform non-family firms. First, according to behavioral theory regarding the ownership and management of family and non-family-owned firms, family-owned firms control higher cash flow rights than non-family-owned firms. This is driven by singular leadership and sustainable development [4]. In addition, family-owned firms are more likely to use interpersonal and value-based skills to make them more competitive and, when it comes to maximizing profits, they put the accumulation of wealth as a priority [5]. Therefore, it is significant for founders of family firms to make decisions which can make development sustainable from a long-term perspective [6]. Also, family-owned firms are concerned with their socio-emotional wealth, and family members prize their firm's reputation, its history, the continuation of its values and the perpetuation of the family dynasty [7]. As a result, family members may maximize their wealth by divesting external shareholders of their shares [8]. This can, however, lead to a lack of information transparency within a family business.
Secondly, based on agency theory, the separation of ownership and control in public firms can result in managers pursuing their own interests at the expense of shareholders' interests [9]. Since family shareholders' wealth is closely tied to their firms, they have more incentive to minimize agency problems by making optimal managerial decisions and bequeathing to the next generation [5].
There have been quite a few family vs. non-family studies done in various countries, both within and outside of China. In general, the results are quite mixed due to different sample countries, different time-periods, and different methodologies in particular return generation models.
It is the purpose of this paper to reconcile some of these results, with a common time frame that includes the GFC, incorporates control variables that have been used in other studies, and compare across three different methodologies.
To our knowledge, this is the first study to be similarly comprehensive. The main issues we address are: Family firms vs. Non-family performance, the impact of firm size, the impact of the CEO being the founder, and the impact of the level of family share ownership. Most of the studies find that family-run firms outperform non-family firms.
[2] [5] [8] [10] [11]. Lipiec [12] also finds that family firms outperform non-family, but only during crisis (he studies the GFC period); before and after the crisis, non-family firms outperform. A few studies find that non-family firms outperform family firms, but in response to specific events-Chang, Wu [13], Wong, Chang [14], and Sirmon, Arregle [15]-in response to innovations, venturing announcements, and threat of imitation, respectively. Fahlenbrach [16] and Adams, Almeida [17] find that founder CEO's outperform, while Villalonga and Amit [18] finds that family firms are better off only when the founder is the CEO or Chairman. Anderson and Reeb [5] and [19] find that family CEO's have positive influence, but Miralles-Marcelo, Miralles-Quirós [2] finds no effect from a family CEO.
As to size, the results are quite mixed. Isakov and Weisskopf [20] and Adams, Almeida [17] find that large family firms outperform small firms, but Anderson and Reeb [5] and Lipiec [12] find the opposite, with Miralles-Marcelo, Miralles-Quirós [2] finding that the family vs. non-family outperformance is particularly strong for small firms.
These and other studies have found a few other interesting effects with respect to social awareness and risk. Dyer and Whetten [21] find that family firms have higher Corporate Social Responsibility (CSR) than non-family firms, and Gómez-Mejía, Haynes [22] finds that family firms are more risk averse.
Within China, the studies are similarly mixed in results. Ding, Zhang [23] find that family-owned firms perform significantly better than state-owned enterprises. This outperformance was especially significant in agricultural and light industries, where family firms have had a presence longer than in other industries which have until recently been strictly state-owned enterprises.
Chinese family-owned firms may also be influenced by religious belief. Du [24] suggests that religious entrepreneurs in Chinese family-owned firms are more likely to participate in political affairs, which may potentially provide political advantages. In a similar vein, Yang and Jiang [25] shows that policy uncertainty has a significantly negative effect on a firm's performance, yet participation in political affairs tends to mitigate this impact. Further, Xu, Yuan [26] finds that founders of family-owned firms who have a solid government relationship are more likely to appoint a second-generation family member as chairman, CEO, or director, which, in turn, could reduce transfer costs, improving firm performance.
Cai, Luo [27] also finds that family firms outperform, but in particular that family CEOs have a positive impact. They suggest this is because family member CEOs can mitigate conflict between family owners as the interests between them are highly aligned, and this can enhance a firm's performance.
However, Luo and Liu [28], which also includes a GFC year in their data sample, finds that family-CEOs did not have a significant effect on abnormal returns. Agency costs have been addressed in some of the studies. Chang, Wu [13] contends that family-owned boards had a significantly negative relationship with a firm's abnormal returns, due to the hiring of unqualified employees simply because they are family members. Wong, Chang [14] shows that institutional ownership has a positive effect on family-firm performance, due to mitigating the negative effects of agency costs and nepotism in family-owned firms.
A study by Cao, Cumming [29] specifically examines the one-child policy, finding that it has negatively affected founders' expectations of going public, which, in turn, has reduced family firms' reinvestment rates. Moreover, it has also led to high agency costs due to the decreased probability of descendants in-

Definition of Family-Owned Firms
Anderson and Reeb [5] examines the level of equity ownership of the founding family and/or the presence of family members on the board of directors, but does not put a minimum limit on the threshold for family ownership. Villalonga and Amit [18] define a family-owned firm as being one where family members own at least 20% of shares. Anderson, Reeb [31] requires that a firm's founder or a member of the family should own at least 5% of the firm's equity, but does not require that the CEO is family. On the contrary, Fahlenbrach [16] defines a family-owned firm as one having a founder or co-founder as CEO.
Chang, Wu [13] concentrates on directed voting and cash-flow rights, requiring that for a firm to be defined as a family-owned, the family must have at least 20% of the firm ownership and at least half of the seats on the board. If a family-owned firm establishes a partnership with other family-owned firms, then the real controlling right belongs to the family with the largest amount of ownership or to their legal representatives. Barontini and Caprio [8] states that a firm can be defined as family-owned when at least 10% of ownership rights and more than 51% of direct voting rights are owned by family members.
In this study, we define a firm as family-owned whenever a founder and/or family members control at least 51% of voting rights and 10% of ownership rights.

Data
The data used in this research is from firms listed on the Chinese shanghai-A share market obtained from the CSMAR's database, with monthly data from December 2006 to November 2016. We require at least 24 months of returns, and that a firm is consistently either family or non-family during our sample period. Following Barontini and Caprio [8] and others, we exclude financial firms. Our final sample consists of 1041 publicly listed firms. Finally, to account for the impact of outliers, the sample has been winsorized, dropping the top and bottom 1%.

Research Hypotheses
As noted above, family-owned firms have higher ownership by family members and most have singular leadership [4]. Therefore, the owners of family firms are more likely to bring interpersonal and value-based skills to make their firms more competitive, which may, in turn, generate higher stock performance than non-family-owned firms. Therefore, the expectation is that there will be abnormal returns for family-owned firms. This leads to the below hypothesis: Hypothesis 1: Family-owned firms outperform non-family-owned firms.
The owners of family-owned firms, particularly founder-CEOs, treat their firms' wealth as a symbol of personal wealth [5]. As a result, they pay attention to decisions that affect the long-term, stable growth of the firm in the public market, which may improve the firm's ability to create potential investment opportunities. The expectation is that family firms with founder-CEOs will obtain abnormal returns. Thus: Hypothesis 2: Founder-CEOs of family-owned firms outperform non-founder-CEOs of family-owned firms.
Of the studies we have reviewed on the Chinese stock market, most find that larger-size has a positive effect on a firms' performance. The expectation is that larger family-owned firms have higher stock performance than smaller family-owned firms. Therefore, the third hypothesis is: Hypothesis 3: Bigger family-owned firms outperform smaller family-owned firms.
Family-owned firms show stronger performance when family owners have high levels of ownership. This is because family members in such firms have the incentive to make optimal development strategies for the long-term [27]. Therefore, the expectation is that family firms with higher levels of family ownership produce higher stock returns than family firms with low levels of family ownership. This leads to the fourth hypothesis: Hypothesis 4: Family-owned firms with high levels of family member ownership outperform family-owned firms with low levels of ownership.

Research Methodology
Many empirical studies of the U.S., European and other markets have used the Miralles-Marcelo, Miralles-Quirós [2] has noted the sensitivity of family firms to debt and illiquidity. As also noted by Amihud [35], illiquid stocks are more sensitive to market illiquidity, so we expect that family firms will likely be more sensitive to this. Miralles-Marcelo, Miralles-Quirós [2] also describe how debt may help explain returns, particularly for small firms, in light of the above mentioned agency issues (also described in Bhandari [36] and Fama and French [37]). Therefore, following Miralles-Marcelo, Miralles-Quirós [2], we add these two variables to the Fama-French model to create a new Five-Factor Model. The illiquidity metric is from Amihud [35], measured originally as absolute value of return/volume; we found this to be very small in our sample, so we used the natural log of this ratio. As with Miralles-Marcelo, Miralles-Quirós [2], the debt metric is the average of debt and market value for each stock.
To examine our hypotheses, we compare outperformance using three models: CAPM, the Fama-French 3-factor model, and our new 5-factor model. We use panel data methodology to allow investigation into individual securities, rather than potentially inducing bias with portfolios of firms stratified in some way 1 .
We employ this with random firm effects, as random effects models can include time invariant variables, and in our sample the family or nonfamily designation is time invariant; with fixed firm effects these variables are absorbed by the intercept.

Individual Impact of Family vs. Non-Family on Stock Returns
As noted earlier, we use three different models to examine the question of whether family firms outperform non-family-CAPM, Fama-French 3-Factor, New 5-Factor.
The Capital Asset Pricing Model (CAPM) , , , where: 1 To conserve space we do not report it here, but we also ran time series regressions using these models on portfolios, in a similar vein to 2. r it: R it -R ft is the excess stock return at time t, R m,t : R m,t -R ft is the excess market return at time t, SMB t is the difference between returns on small-and large-stocks with approximately the same weighted-average book-to-market equity at time t, HML t is the difference between returns on high-and low-BE/ME stock with approximately the same weighted-average size at time t [1], Debt it is the total debt/total market value at time t, Illiq it is the ratio of absolute stock return divided by its volume at time t [35],

The Effects of the Global Financial Crisis
For each of equations above, we add a dummy for GFC (=1 for years [2008][2009][2010] and an interaction term for family and GFC. The GFC dummy is a control; the interaction term, Fam*GFC, will reveal how family firms contribute during the GFC. Table 1 summarizes the sample structure of this research. It presents the number of family-owned and nonfamily-owned firms, the mean excess return for each and the difference in mean excess return for each year. On average, there are 998 firms per year, 23% of which are family-owned. Family-firms returns are significantly higher than nonfamily-firms in 4 of the 10 years. It is notable that three of these significant differences occurred during the GFC, consistent with the observation of Lipiec [12], that family firms outperform during a crisis.

Outperformance by Family-Owned Firms
The first test is to determine in broad terms if family firms outperform nonfamily firms. Table 3 reports results for the 3 models used, with Fam being the variable of interest. For the CAPM and Fama-French Three-factor models, the family firms outperform nonfamily firms by 0.5% and 0.6%, respectively, both significant at the 1% level. However, with the Five-Factor model including Debt and Illiquidity, both significant at 1% (negative and positive, respectively, as we would expect), the significance of Fam vanishes, suggesting that Illiquidity accounts for much of the outperformance.

Outperformance by Three Categories of Family-Owned Firms
For all categories, coefficients on the control variables are as expected, and consistent with previous research (e.g. Xie and Qu [30]). When a founder is the CEO, those firms outperform others under both the CAPM and 3-factor models, as with all family firms, but again not with the 5-factor model (see Table 4).
Comparing this with the family vs. nonfamily results of Table 3, this suggests that the Debt and Illiquidity impact on family firms dominates results, whether the CEO is the founder or not.
Large family-owned firms outperform small family firms with all 3 models (Table 5), significant at the 1% level. As before, with the 5-factor model, Debt is significantly negative and Illiquidity is significantly positive. Family ownership is   Notes: *, ** and *** denote significance at the 10%, 5% and 1% levels at a two-tail test, respectively. This table reports the panel data regression results on firm performance, using three models. In all cases, the dependent variable is firm return excess of the risk-free rate, BigFam is a dummy variable = 1 if the firm size is greater than the median of all family firms, Rm is market excess return, SMB and HML are the size and growth/value factors for the Fama-French 3-factor model, Debt is total debt divided by the total assets, and Illiquidity is the average return divided by total monthly volume (both at market level). There are 25,905 observations over the period 12/2006 to 11/2016. also positive and significant for CAPM and the 3-factor model, but not for the 5-factor model ( Table 6).
In terms of our hypotheses, the results are a bit mixed. For hypothesis 1, that family firms outperform nonfamily firms, for CAPM and the 3-factor model, we cannot reject. However, with the 5-factor model we find no significant evidence that family firms outperform. For hypothesis 2, that family firms with a founder CEO outperform other family firms, we cannot reject that family firms outperform under CAPM or the 3-factor model, but again with the 5-factor model we find no significant evidence that firms with founder CEOs outperform. For  Notes: *, ** and *** denote significance at the 10%, 5% and 1% levels at a two-tail test, respectively. This table reports the panel data regression results on firm performance, using three models. In all cases, the dependent variable is firm return excess of the risk-free rate, HiOwner is a dummy variable = 1 if the family share ownership is greater than 50%, Rm is market excess return, SMB and HML are the size and growth/value factors for the Fama-French 3-factor model, Debt is total debt divided by the total assets, and Illiquidity is the average return divided by total monthly volume (both at market level). There are

Impact of the Global Financial Crisis (GFC)
Lipiec [12] finds that family-owned firms in Warsaw perform significantly better  Tables 7-10).
In nearly all cases, for both the models and categories considered, we find results largely similar to those that do not consider the GFC. As before, the 5-factor model suggests that Debt and Illiquidity are important, but now after addressing the GFC, we reject hypotheses 1, 2 & 4 under the 5-factor model.
However for Size, we still find that large family firms outperform small family firms, even after addressing the GFC, with a coefficient of 2%, significant at the 1% level.

Conclusions
We examine monthly returns of family and nonfamily firms on the Shanghai    Notes: *, ** and *** denote significance at the 10%, 5% and 1% levels at a two-tail test, respectively. This table, similar to Table 6 but with GFC, reports the panel data regression results on firm performance, using three models. In all cases, the dependent variable is firm return excess of the risk-free rate, HiOwner is a dummy variable = 1 if the family share ownership is greater than 50%, Rm is market excess market excess return, SMB and HML are the size and growth/value factors for the Fama-French 3-factor model, Debt is total debt divided by the total assets, and Illiquidity is the average return divided by total monthly volume (both at market level), GFC is a dummy variable = 1 during years 2008-2010. There are 25,905 observations over the period 12/2006 to 11/2016. For the "within family firms" tests, larger firms outperform small firms with all three models. However, the CEO being a firm founder and family ownership level only have significant (and positive) impact under the CAPM and 3-factor models, not in the 5-factor model, again implying a strong impact from debt and illiquidity.
The GFC does not have much impact on our results regarding family and nonfamily firms. The GFC itself was of course significant, but at the margin, the impact of family during the GFC was not significant. For CAPM and the 3-factor Overall, our findings-particularly for the CAPM and Fama-French 3-factor models-corroborate the bulk of previous research, which shows that family firms outperform nonfamily firms, and that founder CEO's, large firms, and high ownership all play a part in outperformance. However, our addition of the 5-factor model helps to explain some of the contrary findings of other research, showing that size is the dominating factor in family firm outperformance.