Printer Friendly

The global financial crisis, family control, and dividend policy.

Using newly collected data on the ultimate ownership structure ofpublicly traded firms in nine East Asian economies, we find that family control is negatively related to the dividend payout ratio. Family firms are less (more) likely to increase (omit) dividends than non-family firms. These negative associations between family firms and dividend policy are more pronounced during the recent global financial crisis, suggesting that controlling families have incentives to expropriate more firm resources during crises than in normal times.

**********

Theory suggests that family owners, as large shareholders, may act as monitors and thus reduce agency and information asymmetry problems stemming from the atomistic structure of corporate shareholdings (Shleifer and Vishny, 1986). However, this conclusion may not hold outside the United States, where institutions provide minority shareholders less protection and hence expropriation is more likely to take place among family firms (Almeida and Wolfenzon, 2006).

The literature on the link between dividends and family control generally focuses on a single country with mixed results. For instance, prior work shows that family firms tend to pay lower dividends than non-family firms in Austria (Gugler, 2003), Germany (Gugler and Yurtoglu, 2003), Hong Kong (Chen et al., 2005), and the United States (Hu, Wang, and Zhang, 2008). In contrast, Setia-Atmaja, Tanewski, and Skully (2009) and Isakov and Weisskopf (2015) find that family firms exhibit higher dividend payouts in Australia and Switzerland, respectively. Pindado and de la Torre (2008) find that family ownership is irrelevant to dividends in Spain. Overall, the literature has little to say about cross-country variation in the association between dividends and family ownership. (1) Given the prevalence of family firms around the globe (Anderson, Duru, and Reeb, 2009, 2012; Carney and Child, 2013), examining the dividend policy of family firms in a cross-country setting can improve our understanding of the value relevance of family control, potentially adding insight into the mixed evidence at the single country level.

Accordingly, we investigate the differences in dividend policy between family and non-family firms in nine East Asian countries around the 2008-2009 global financial crisis. To conduct our analysis, we use Carney and Child's (2013) newly collected data on the ultimate ownership structure of publicly traded firms in Asia. Family firms provide an ideal testing ground because they are particularly vulnerable to unanticipated liquidity shocks (Almeida et al., 2012). Owners of family firms are subject to significant idiosyncratic risk (Anderson and Reeb, 2003; Faccio, Marchica, and Mura, 2011), mostly because family owners tend to retain control over management and hold highly undiversified investments in the firm. (2) Furthermore, incentives to expropriate rents and extract private benefits of control are exacerbated during a crisis (Johnson et al., 2000).

Using 4,285 firm-year observations representing 923 unique firms from nine countries (Indonesia, Japan, Hong Kong, Malaysia, Philippines, Singapore, Thailand, South Korea, and Taiwan) over 2006-2010, we first investigate the extent to which dividend payouts of family firms differ from those of non-family firms. We find that across East Asian economies, family firms are associated with lower dividend payouts than non-family firms. This finding is robust to sensitivity tests. In particular, our results remain unchanged when we employ propensity score matching (PSM), use different samples, and consider alternative proxies for dividend payout policy. We also find that the negative association between family control and dividend payout holds only for firms with more pronounced agency problems (proxied by free cash flows).

Next, we examine the impact of family control on changes in dividend policy and find that family firms are less likely to increase dividends and more likely to decrease and omit dividends, supporting the view that family firms are associated with more pronounced agency problems.

When we focus on the crisis period (2008-2009), we find that the negative association between family-controlled firms and dividend payout ratios is more pronounced, complementing the evidence in Lins et al. (2013) that controlling families tend to cut productive investment or divert resources to a greater extent during financial crises than during normal times. This result lends support to Masulis, Pham, and Zein (2011), who show that family firms use their resources more efficiently under normal business conditions.

We complete our analysis by examining how family firms employ the resources that are not distributed to shareholders. The literature suggests that the negative association between family firms and dividend payouts is due to precautionary motives aimed at preserving corporate resources. Here firms build their cash reserves to alleviate financing frictions (Keynes, 1936) and buffer investments from temporary financing shocks (e.g., Opler et al., 1999; Harford, Mansi, and Maxwell, 2008; Brown and Petersen, 2011). Under this hypothesis, family firms are expected to have higher cash reserves and investment. Our results show instead that even when profitability increases, family firms actually reduce cash holdings and cut investment expenditures, suggesting that the dividend policy of family firms cannot be explained by precautionary motives, but rather is a manifestation of agency problems and extraction of private benefits.

The findings of this study imply that the corporate governance changes that have taken place in East Asia since the 1997 financial crisis (Carney and Child, 2013) do not appear to have constrained the corporate behavior of family firms during the most recent crisis. This conclusion is consistent with the Faccio, Lang, and Young (2001) observation that "crony capitalism" in East Asian economies is politically, rather than institutionally, determined. More broadly, our evidence contributes to the literature on the relation between family control and dividend policy in a cross-country sample, thereby adding to the debate on the extent to which family control serves the family's interests to the detriment of outside shareholders, as well as to the growing line of research that examines the corporate effects of the 2008-2009 financial crisis (e.g., Campello, Graham, and Harvey, 2010; Duchin, Ozbas, and Sensoy, 2010; Lins et al., 2013).

The rest of this study is organized as follows. In Section I, we develop our hypotheses. In Section II, we describe our research design and sample construction. In Section III, we report the results, and in Section IV we conclude.

I. Hypotheses

Our hypotheses are rooted in work that suggests dividends bind managers to a long-term commitment to disgorge future free cash flows and expose managers to increased monitoring by the external market, reducing the risk of insider expropriation (Rozeff, 1982; Easterbrook, 1984; Jensen, 1986; Faccio, Lang, and Young, 2001). Accordingly, dividend-paying firms are expected to have fewer agency problems. (3)

Based on this line of reasoning, we distinguish two views on the relation between family control and dividends. On the one hand, the positive view of family firms (Anderson and Reeb, 2003) suggests that family firms have fewer agency problems than non-family firms. Thus, this positive view implies that family firms pay higher dividends than other firms. On the other hand, prior empirical evidence (e.g., Claessens et al., 2002; Fan and Wong, 2002; Barth, Gulbrandsen, and Schone, 2005; Attig et al., 2006; Boubakri, Guedhami, and Mishra, 2010) shows that family firms tend to be associated with more agency problems than non-family firms, due to greater use of control-enhancing mechanisms and excessive board representation (e.g., La Porta et al., 1998; Faccio, Lang, and Young, 2001; Faccio and Lang, 2002). Thus, this negative view of family firms implies that family-controlled firms pay lower dividends than other firms.

Because our sample firms are located in countries with weak protection of minority shareholder interests (Claessens, Djankov, and Lang, 2000; Carney and Child, 2013), our first hypothesis is based on the negative view of family firms:

H1: Family firms pay lower dividends than non-family firms.

However, extant theory also suggests that firms with more pronounced agency and information problems pay higher dividends to signal private information about the firm's future prospects (Bhattacharya, 1979; Miller and Rock, 1985) or to bond themselves to act in the interests of shareholders (Rozeff, 1982). (4) Under this alternative explanation, the positive view of family firms suggests that family firms do not need the bonding or signaling benefits of an aggressive dividend policy and hence pay lower dividends than non-family firms, whereas the negative view of family firms suggests that family firms pay higher dividends than non-family firms to signal or bond themselves more aggressively. (5)

To distinguish the agency problem explanation from the alternative (i.e., signaling or bonding) explanation, we first examine the dividend policy of family firms according to the extent of their agency problems as proxied by their free cash flows. If the dividend policy of family firms is driven by agency problems, we would expect family firms with more pronounced agency problems to pay lower dividends than other family firms. This leads to our second hypothesis:

H2: Family firms with more pronounced agency problems (i.e., more free cash flow) pay lower dividends than other family firms.

To further distinguish the explanation driving the dividend policy of family firms, we examine the effect of family control on changes in dividends. Under the agency problem explanation, an increase in dividends binds managers to pay out higher dividends in future periods (e.g., Jensen, 1986), whereas a decrease in dividends increases the free cash flow available to managers to divert to their private benefit. The negative view of family firms implies in turn that their managers will be less inclined to increase dividends. This leads to our third hypothesis:

H3: Family firms are negatively (positively) associated with increases (decreases) in dividends.

Building on this test, we examine the sensitivity of dividend changes in family firms to changes in their profitability. The negative view under the agency problem explanation predicts that family firms with increased profitability are inclined to decrease dividends to increase managerial discretion with respect to the use of corporate resources. In contrast, the negative view under the signaling/bonding explanation predicts that family firms with increased profitability increase dividends to signal their favorable prospects. In terms of the agency explanation, our fourth hypothesis is thus:

H4: Family firms with increased profitability are negatively (positively) associated with dividend increases (decreases).

In an additional test to distinguish the agency and signaling/bonding explanations, we examine the extent to which an unexpected liquidity shock--the 2008-2009 financial crisis--affects the dividend policy of family firms. Lins et al. (2013) argue that a liquidity shock can lead family-controlled firms, compared to other firms, to engage in actions that aim to preserve the family's benefits of control at the expense of outside shareholders. The recent financial crisis can thus be used as a natural experiment that helps alleviate endogeneity concerns. Boubakri et al. (2010) find a positive association between family control and firms' equity financing costs following the 1997 Asian crisis, lending support to the argument that agency costs are more pronounced during times of crisis. Consistent with the expropriation hypothesis, Bae et al. (2012) find that poorly governed Asian firms experience a large decrease in value during the 1997 Asian crisis but perform better during the recovery period. Johnson et al. (2000) further suggest that rent expropriation can be exacerbated during a crisis, and Mitton (2002) posits that corporate governance may be especially critical during periods of economic distress. Building on these insights, the negative view of family firms under the agency problems explanation predicts that family firms pay lower dividends during the recent financial crisis. This leads to our fifth hypothesis:

H5: During the 2008-2009 financial crisis, family firms pay lower dividends than non-family firms.

II. Data and Research Design

A. Sample Selection and Ownership Data

To investigate the impact of family firms on dividend payout policy, we first compile ownership data for a sample of firms from nine East Asian economies (Hong Kong, Indonesia, Japan, Korea, Malaysia, Philippines, Singapore, Taiwan, and Thailand) used by Carney and Child (2013). The data set identifies the ultimate controlling shareholders as well as their ultimate cash flow (ownership) and voting (control) rights as of 2008. We hand-match the ownership data set with financial data from the Compustat Global database for 2006-2010. We drop firms with insufficient financial data to construct the regression variables and eliminate financial firms (Standard Industrial Classification [SIC] codes between 6000 and 6999) from our analysis. This procedure leaves us with 4,285 observations representing 923 unique firms over 2006-2010.

B. Variables

1. Dividend Variable

Following prior research (e.g., Chay and Suh, 2009), we use the dividend payout ratio, DIV/E, as our primary measure of dividends. DIV/E is computed as the ratio of cash dividends to net income before extraordinary items. We winsorize DIV/E at 0 and 1, and set DIV/E equal to 1 for cash dividends that are larger than net income, and 0 for negative net income firms that do not pay cash dividends. (6) We consider alternative dividend payout measures in robustness tests.

2. Family Control

Following Boubakri et al. (2010) and Lins et al. (2013), we proxy for family control using FAMILY, a dummy variable equal to 1 if the ultimate owner is a family, and 0 otherwise. Carney and Child (2013) identify ultimate owners at the 10% and 20% thresholds. We consider ultimate owners at the 10% threshold in our main analysis. (7)

3. Control Variables

Our selection of firm-level control variables closely follows Brockman and Unlu (2009). Specifically, we control for RE/TE, the ratio of retained earnings (RE) to total assets (AT); TE/TA, the ratio of shareholders' equity (CEQ) to total assets; ROA, return on assets, computed as the ratio of net income before extraordinary items (IB) to total assets; SGR, logarithmic sales growth, computed as log([SALE.sub.t]/[SALE.sub.t-1]); SIZE, the natural logarithm of total assets in millions of $US; and CH/TA, the ratio of cash and short-term investments (CHE) to total assets. (8) To mitigate the effect of outliers, we winsorize all ratios at the 1% and 99% levels. In addition to these firm-level variables, we control for country and industry-year effects, to account for country- and industry-specific trends, in all our regressions. (9) We do not include firm fixed effects because the cross-sectional variation in FAMILY should capture the heterogeneity in dividend payout policy across firms, and including firm fixed effects would remove all cross-sectional variation from the data (Zhou, 2001; Lemmon, Roberts, and Zender, 2008). Further details on variable construction are provided in the Appendix.

C. Descriptive Statistics

Table I summarizes our sample distribution by year, country, and industry (using Campbell's, 1996, industry classification). Among East Asian countries, Taiwanese firms have the largest representation in our sample (15.52%) followed by Malaysian firms (13.16%). In contrast, only 6.09% and 9.40% of sample firms are in the Philippines and Indonesia, respectively. Our sample also shows diversification across industries, with 21.05% in consumer durables, 13.16% in basic industries, 11.11% in food and tobacco, and 10.18% in utilities. In sum, Table 1 suggests that the distribution of our sample has sufficient variation across years, countries, and industries to identify a link between family ownership and dividend payout policy in an international context.

Table II presents descriptive statistics by country. We note that the highest average dividend payout ratio is in Thailand (52.2%), followed by Japan (46.9%), Taiwan (45.9%), and Malaysia (44.8%), while the Philippines has the lowest average dividend payout ratio (30.8%). Interestingly, however, the Philippines has the highest proportion of firms with a family as the ultimate owner (78.5% of sample firms at the country level), followed by Korea (63.5%). Similarly, although Japan displays the lowest proportion of family firms (11.2%), it has the second highest dividend-payout ratio. These descriptive figures provide preliminary evidence of a negative link between family control and dividend payouts.

A casual examination of the distribution of other firm characteristics shows that Japanese and Malaysian firms have the highest retained earnings (RE/TE of 0.640 and 0.426, respectively), while firms in Korea and Hong Kong have the lowest (RE/TE of 0.089 and 0.115, respectively). On average, firms have a TE/TA higher than 50% in most countries, with the exception of firms in Indonesia, Japan, and Korea. The fastest growing firms are in Indonesia (sales growth rate of 15%) followed by firms in Singapore (13.8%). Firms in Japan exhibit the lowest sales growth (1.9%) and the lowest profitability (ROA of 3.4%) compared to firms in other countries in the region. In sum, the figures reported in Table II again suggest that our sample contains sufficient heterogeneity in dividend policy and firm ownership across East Asian countries to examine our research question.

Table III presents descriptive statistics at the firm level and correlations among our key regression variables. In Panel A, we observe that our sample firms have an average dividend payout ratio of 41.4%. Furthermore, for 43.3% of our sample firms the ultimate shareholder is a family. In Panel B we note that, in line with HI, family control (FAMILY) is negatively correlated with DIV/E, suggesting that family firms pay lower dividends. The pairwise correlation coefficients between the key regression variables are relatively low, indicating that multicollinearity is not likely to affect our multivariate regression results.

III. Empirical Results

A. Univariate Analysis

To further explore differences between family and non-family firms, in Table IV we examine differences in means (Panel A) and medians (Panel B). In Panel A, we observe that the dividend payout ratio is significantly higher for non-family firms (45.0%) than for family firms (36.7%). This result is in line with our preliminary evidence in Table II and Table III, Panel B, indicating that family firms tend to pay lower dividends than their non-family counterparts. This finding does not appear to be explained by firm profitability or growth, as the profitability (ROA) and sales growth (SGR) of family firms are indistinguishable from those of non-family firms, leaving open the possibility that rent expropriation may explain dividend payout policy in family firms. The findings based on differences in medians reported in Panel B are generally consistent with the differences-in-means results in Panel A. One noticeable exception is that the median sales growth of family firms is higher than that of non-family firms.

Although the results of Table IV are interesting in their own right, they do not control for other variables that could affect firms' dividend policy. We conduct multivariate analysis next.

B. Multivariate Analysis

1. Family Firms and Dividend Policy: Main Evidence

To examine the impact of family control on firms' dividend payout ratio (DIV/E), we estimate the following Tobit regression model:

DIV/E = [[alpha].sub.0] + [[alpha].sub.1] FAMILY + [[alpha].sub.2]RE/TE + [[alpha].sub.3]TE/TA + [[alpha].sub.4]ROA + [[alpha].sub.5]SGR + [[alpha].sub.6]SIZE + [[alpha].sub.7]CH/TA + Fixed Effects + [epsilon], (1)

where inferences are based on standard errors clustered by firm. Table V reports the results.

We begin in Column (1) of Table V by estimating the above specification with country and industry-year effects. The estimated coefficient on FAMILY is negative and significant at the 1% level, suggesting that in line with H1, family firms pay lower dividends than non-family firms. When we repeat this analysis using industry and country-year effects in Column (2), we find that the effect of FAMILY on DIV/E remains negative and significant at the 1% level. Thus, for brevity we focus on country and industry-year effects in subsequent analysis. The coefficients on the control variables exhibit the expected signs. For instance, retained earnings and firm size load significantly and positively whereas sales growth loads significantly and negatively on dividend payouts.

Because family firms may be fundamentally different from non-family firms (e.g., Table IV shows that family firms are significantly smaller and have higher equity-to-asset ratios than non-family firms), one may ask whether the negative relation between family control and dividends is driven by family control or other firm characteristics. To address this question, we employ PSM to control for differences in characteristics between family and non-family firms and estimate family firm treatment effects. We start by running a probit model of FAMILY on firm characteristics as well as fixed effects (country and industry-year). We then match each family firm to the non-family firm with the closest score (with replacement). (10) The advantage of using a matched sample is that family and non-family firms are similar with respect to firm characteristics included in the probit model, which allows us to isolate the effect of family firm affiliation on dividend payouts. The results, reported in Column (3) of Table V, are in line with our main results. Specifically, FAMILY remains negatively related to the propensity to pay dividends, and significant at the 5% level.

2. Family Firms and Dividend Policy: Competing Explanations

Although our results highlight a negative relation between family control and dividend payouts (H1), they are not necessarily consistent with an agency explanation, in line with the negative view of family firms. In particular, the negative link between family firms and dividends could reflect less need to pay dividends due to lower signaling/bonding motives, in line with the positive view of family firms. To distinguish which of these explanations drives our results on the relation between family control and dividends, we begin by examining the extent to which family firms with more pronounced agency problems--that is, high free cash flows and few investment opportunities (e.g., Lang, Stulz, and Walkling, 1991; Chen, Chen, and Wei, 2011)--pay lower dividends than other firms (H2).

Following Lins et al. (2013), we measure a firm's cash flow as the difference between operating income before depreciation and capital expenditures deflated by total assets. Note that if family firms suffer from agency problems, their observed free cash flow might be endogenous. To circumvent this endogeneity issue, we adopt the methodology of Chen et al. (2011) and employ a firm's residual cash flow, calculated as the residual from a regression of cash flow on FAMILY, as this measure is orthogonal to family control. To capture a firm's investment opportunities we use its industry median Tobin's Q, where Tobin's Q is equal to total assets minus shareholders' equity plus market capitalization, all deflated by total assets, as the industry median value should be largely unaffected by the firm's ownership structure.

To empirically test H2, we follow the standard practice of estimating regressions on subsamples sorted by agency costs (e.g., Chen et al., 2011; Lins et al., 2013). More specifically, each year we assign firms to one of four subsamples according to whether their residual cash flow is lower (higher) than the median and whether their median industry Q is lower (higher) than the median; firms with above-median residual cash flows and below-median industry Q are identified as having pronounced agency problems. We then estimate our main regression separately for the four subsamples. The results are reported in Table VI. Interestingly, we find that FAMILY loads significantly and negatively only for the subsample of firms suffering from pronounced agency problems. Providing some support for H2, this finding based on split-sample analysis is consistent with the agency explanation for the negative relation between family control and dividends, suggesting a negative view of family firms. (11)

In Table VII, we further examine the link between family control and changes in dividends, and the sensitivity of dividend changes in family firms to changes in firm profitability. H3 predicts that family firms are negatively (positively) associated with increases (decreases) in dividends, and H4 predicts that family firms with increased profitability are negatively (positively) associated with dividend increases (decreases). To test these predictions, we create DIV_JNCR (DIV_DECR), which equals 1 if cash dividends in year t are higher (lower) than cash dividends in year t - 1, and 0 otherwise, and DIV_OMIT, which equals 1 if cash dividends in year t are 0 and cash dividends in year t - 1 are positive, and 0 otherwise. Columns (1) and (2) report results for increases in dividends, and Columns (3) and (4) ((5) and (6)) correspond to dividend decreases (omissions).

In Column (1), we find that family firms are negatively associated with increases in dividends. In particular, we document a negative and significant (at the 1% level) effect of FAMILY on the likelihood of a dividend increase {DIV_INCR), suggesting that family firms are less likely to increase dividend payouts. In Column (2), we further find that family firms with increased profitability are associated with a lower likelihood of a dividend increase than other firms. On the decrease side, we find that family firms are positively associated with decreases in dividends, with FAMILY loading positively and significantly (at the 1% level) on the likelihood of a dividend decrease (DIV_DECR). We do not find a significant effect of the interaction between family firms and changes in profitability on the likelihood of decreasing dividends. Turning to dividend omissions, we find that FAMILY loads positively and significantly on DIV_OMIT, suggesting that family firms tend to not only reduce dividends (as reported in Columns (3) and (4)), but to omit them as well. Equally important, we document a positive and significant coefficient on the interaction between family control and changes in firm profitability (FAMILY x [DELTA]ROA) on DIV_OMIT. Overall, these results are in line with H3 and H4, and hence suggest that dividend policy in family firms is more likely to be driven by agency problems than the alternative signaling/bonding explanation.

Although the evidence we document above on the link between family control and dividend payout policy (in terms of both levels and changes) is on the whole persuasive, in the following section, we further improve our understanding of this relation by examining the extent to which the recent global financial crisis affects this relation.

3. Family Firms and Dividend Policy: The Effect of the Global Financial Crisis

In Table VIII, we investigate the effect of the recent global financial crisis on family firms' dividend payout policy. H5 predicts that family firms pay lower dividends during the financial crisis. To test this prediction, we run our analysis introducing a dummy variable (D(2008-2009)) for the crisis years (2008 and 2009) and the interaction between this variable and FAMILY (FAMILY x D(2008-2009)). Although FAMILY loses its statistical significance, FAMILY x D (2008-2009) bears significantly and negatively on the level of dividends paid out of earnings.

The novel finding of Table VIII, which is consistent with H5, suggests that family firms in East Asia decreased their dividend payouts during the recent financial crisis. This result supports the argument that resource preservation and rent expropriation in family firms are more pronounced during financial shocks than during normal times. This evidence also suggests that the governance reforms that took place in East Asia following the 1997 financial crisis have not significantly altered the corporate behavior of family firms.

C. Additional Analyses and Robustness Checks

In this subsection, we subject our main findings to sensitivity tests to assess the robustness of our conclusions.

1. Effect of Family Control on Cash and Investment

Our results so far indicate that family firms pay lower dividends than non-family firms. This raises the question of how families employ the resources they do not distribute to shareholders. As we discuss in the Introduction, firms with increased profitability may not distribute dividends in order to accumulate cash and thereby buffer investments from transitory financing shocks (e.g., Opler et al., 1999; Harford et al., 2008; Brown and Petersen, 2011). To the extent this is the case, family firms with increased profitability are expected to be associated with more cash reserves and more investment. To test this conjecture, we examine the impact of family firms on cash holdings and capital expenditures. We report the results in Table IX. We find that, on average, family firms do not have a statistically significant effect on either cash holdings or capital expenditures. However, when we interact FAMILY with the change in return on assets (AROA), the coefficients on the interaction terms enter with a negative and statistically significant sign. This indicates that when profitability increases, family firms tend to reduce cash holdings and cut investment expenditures. Together with our main finding that family firms pay fewer dividends, the results of Table IX indicate that family firms are associated with more appropriation of corporate resources than non-family firms.

2. Alternative Payout Variables

We assess whether our results continue to hold if we use alternative payout variables. First, we consider dividends deflated by assets (DIV/A). Second, we follow Brockman and Unlu (2009) and use DIV/S on the grounds that sales are less prone to opportunistic manipulation than earnings. Third, we employ total payout, that is, cash dividends and stock repurchases, deflated by earnings (TOTP/E) as alternative dependent variable. The results reported in Table X show that the estimated coefficients on FAMILY remain largely unchanged, confirming our findings above based on DIV/E.

3. Sample Composition

Recall from Table I that some countries account for more than 15% of our sample firms (e.g., Taiwan), while others (e.g., the Philippines) account for only 6.41%. In Panel A of Table XI, we explore whether the exclusion of any given country alters our core findings. To do so, we rerun our main regression model by sequentially excluding each of our sample countries. Our main finding of a negative association between family firms and dividend payouts remains unchanged.

To further assess whether sample composition is driving our main evidence, we reestimate our regression model using weighted regressions, where countries with a large number of observations are given less weight. The results, reported in Panel B of Table XI, confirm our main evidence that family ownership is significantly and negatively related to firms' dividend policy (at the 1% level). Our control variables retain their sign and significance, suggesting that our evidence is not driven by sample composition.

IV. Conclusion

Notwithstanding prior research on the role that ultimate corporate ownership structure plays in corporate outcomes, few studies examine the dividend payout policy of family firms in an international context. Our work fills this void by documenting the importance of family ownership in shaping firms' dividends policy in East Asian firms, with a particular focus on the crisis period. By examining the impact of this unexpected liquidity shock on family firms' dividend payout policy, we also add to the nascent literature on the effects of the financial crisis.

In this article, we use newly collected data on the ultimate ownership structure of publicly traded firms in East Asia. We compare the dividend payout ratios of family and non-family firms from 2006 to 2010 and find strong evidence that family firms are negatively associated with the amount of dividends paid out of earnings. We also show that family firms are more likely to decrease and omit dividends. The negative effect of family firms on dividend payouts is more pronounced during financial crises, lending support to the view that rent expropriation and private benefits of control in family firms are exacerbated during financial shocks. In a follow-up analysis, we observe that the negative association between family control and dividend payout holds only for firms with more pronounced agency problems.

Finally, we examine how family firms employ the resources they do not distribute to shareholders. If family firms retain earnings for precautionary purposes aimed at preserving corporate resources, we expect them to have higher cash reserves and investment. Inconsistent with this precautionary motive explanation, our findings suggest that even when profitability increases, family firms actually reduce cash holdings and cut investment expenditures, suggesting that the dividend policy of family firms is a manifestation of agency problems and extraction of private benefits.

Overall, our results indicate that family firms in East Asia are plagued by agency problems that the crisis only helped to exacerbate. It is interesting that the previous Asian crisis of 1997, largely blamed on corporate governance failures, does not seem to have induced improvements in the governance of East Asian firms this time around.

Appendix
Table A1. Variable Definitions and Data Sources

This table provides variable definitions and data sources.
Compustat data items are reported in parentheses.

Variable        Definition                        Source

DIV/E           Dividend payout defined as        Authors' calculations
                  the ratio of cash dividends     based on Compustat
                  (DV) to net income before       data
                  extraordinary items (IB).
                  DIV/E equals 1 for cash
                  dividends that are larger
                  than net income, and 0 for
                  negative net income firms
                  that do not pay cash
                  dividends.
DIV_INCR        Dummy variable equals 1 if        As above
                  cash dividends in year t are
                  higher than cash dividends
                  in year t - 1, and 0
                  otherwise.
DIV_DECR        Dummy variable equals 1 if        As above
                  cash dividends in year t are
                  lower than cash dividends in
                  year t - 1, and 0 otherwise.
DIV_OMIT        Dummy variable equals 1 if a      As above
                  dividend-paying firm in year
                  t - 1 does not pay dividends
                  in year 1, and 0 otherwise.
DIV/A           Ratio of cash dividends (DV)      As above
                  to total assets (AT).
DIV/S           Ratio of cash dividends (DV)      As above
                  to sales (SALE).
TOTP/E          Total payout defined as the       As above
                  ratio of the sum of cash
                  dividends (DV) and stock
                  repurchases (PRSTKC) to net
                  income before extraordinary
                  items (IB).
FAMILY          Dummy variable equals 1 if        Authors' calculation
                  the largest ultimate owner      based on Carney and
                  at the 10% threshold is a       Child's (2013) data
                  family, and 0 otherwise.
RE/TE           Ratio of retained earnings        Authors' calculations
                  (RE) to common stockholders'    based on Compustat
                  equity (CEQ).                   data
TE/TA           Ratio of common                   As above
                  stockholders' equity (CEQ)
                  to total assets (AT).
ROA             Return on assets computed as      As above
                  ratio of net income before
                  extraordinary items (IB) to
                  total assets (AT).
SGR             Logarithmic sales growth          As above
                  computed as
                  log([SALE.sub.t]/
                  [SALE.sub.t-1]).

SIZE            Natural logarithm of total        As above
CH/TA             assets in millions of $US.      As above
                  Ratio of cash and short-term
                  investments (CHE) to total
                  assets (AT).
Residual cash   Residual from a regression        As above
  flow            of cash flow on FAMILY. Cash
                  flow is defined as the ratio
                  of the difference between
                  operating income before
                  depreciation (OIBDP) and
                  capital expenditures (CAPX)
                  to total assets (AT).
Industry Q      The median industry Q in          As above
                  year t. Q is the ratio of
                  total assets (AT) minus
                  stockholder's equity (CEQ)
                  plus market capitalization
                  (CSHOC x PRCCD), all
                  deflated by total assets
                  (AT).
[DELTA]ROA      The change in ROA from year       As above
D (2008-2009)     r - 1 to year t. Dummy          As above
                  variable equals 1 for 2008
                  and 2009, and 0 otherwise.
CAPX/TA         Ratio of capital                  As above
                  expenditures (CAPX) to total
                  assets (AT).


References

Almeida, H., M. Campello, B. Laranjeira, and S. Weisbenner, 2012, "Corporate Debt Maturity and the Real Effects of the 2007 Credit Crisis," Critical Finance Review 1, 3-58.

Almeida, H.V. and D. Wolfenzon, 2006, "A Theory of Pyramidal Ownership and Family Business Groups," Journal of Finance 61, 2637-2680.

Anderson, R.C., A. Duru, and D.M. Reeb, 2009, "Founders, Heirs, and Corporate Opacity," Journal of Financial Economics 92, 205-222.

Anderson, R.C., A. Duru, and D.M. Reeb, 2012, "Investment Policy in Family Controlled Firms f Journal of Banking and Finance 36, 1744-1758.

Anderson, R.C. and D.M. Reeb, 2003, "Founding-Family Ownership and Firm Performance: Evidence from the S&P 500," Journal of Finance 58, 1301-1328.

Attig, N., W-M. Fong, L. Lang, and Y. Gadhoum, 2006, "Effects of Large Shareholding on Information Asymmetry and Stock Liquidity," Journal of Banking and Finance 30, 2875-2892.

Bae, K.H., J.S. Baek, J.K. Kang, and W.L. Liu, 2012, "Do Controlling Shareholders' Expropriation Incentives Imply a Link between Corporate Governance and Firm Value? Theory and Evidence," Journal of Financial Economics 105, 412-435.

Barth, E., T. Gulbrandsen, and P. Schone, 2005, "Family Ownership and Productivity: The Role of Owner-Management," Journal of Corporate Finance 11, 107-127.

Bhattacharya, S., 1979, "Imperfect Information, Dividend Policy, and 'The Bird in the Hand' Fallacy," Bell Journal of Economics 10, 259-270.

Boubakri, N., O. Guedhami, and D. Mishra, 2010, "Family Control and the Implied Cost of Equity: Evidence before and after the Asian Financial Crisis," Journal of International Business Studies 41, 451-474.

Brockman, P. and E. Unlu, 2009, "Dividend Policy, Creditor Rights, and the Agency Costs of Debt," Journal of Financial Economics 92, 276-299.

Brown, J.R. and B.C. Petersen, 2011, "Cash Holdings and R&D Smoothing," Journal of Corporate Finance 17, 694-709.

Campbell, J.Y., 1996, "Understanding Risk and Return," Journal of Political Economy 104, 298-345.

Campello, M., J. Graham, and C. Harvey, 2010, "The Real Effects of Financial Constraints: Evidence from a Financial Crisis," Journal of Financial Economics 97, 470-497.

Carney, R.W. and T.B. Child, 2013, "Changes to the Ownership and Control of East Asian Corporations between 1996 and 2008: The Primacy of Politics," Journal of Financial Economics 107, 494-513.

Chay, J.B. and J. Suh, 2009, "Payout Policy and Cash-Flow Uncertainty," Journal of Financial Economics 93, 88-107.

Chen, K.C.W., Z. Chen, and K.C.J. Wei, 2011, "Agency Costs of Free Cash Flow and the Effect of Shareholder Rights on the Implied Cost of Equity Capital," Journal of Financial and Quantitative Analysis 46, 171-207.

Chen, Z., Y. Cheung, A. Stouraitis, and A.W.S. Wong, 2005, "Ownership Concentration, Firm Performance and Dividend Policy in Hong Kong," Pacific-Basin Finance Journal 13, 431-449.

Claessens, S., S. Djankov, J. Fan, and L. Lang, 2002, "Disentangling the Incentive and Entrenchment Effects of Large Shareholding," Journal of Finance 57, 2741-2771.

Claessens, S., S. Djankov, and L. Lang, 2000, "The Separation of Ownership and Control in East Asian Corporations," Journal of Financial Economics 52, 81-112.

Duchin, R., O. Ozbas, and B. Sensoy, 2010, "Costly External Finance, Corporate Investment, and the Subprime Mortgage Credit Crisis," Journal of Financial Economics 97, 418-435.

Easterbrook, F., 1984, "Two Agency-Cost Explanations of Dividends," American Economic Review 14, 650-659.

Faccio, M. and L. Lang, 2002, "The Ultimate Ownership of Western European Corporations," Journal of Financial Economics 65, 365-395.

Faccio, M., L. Lang, and L. Young, 2001, "Dividends and Expropriation," American Economic Review 91, 54-78.

Faccio, M., M.-T. Marchica, and R. Mura, 2011, "Large Shareholder Diversification and Corporate Risk-Taking," Review of Financial Studies 24, 3601-3641.

Fan, J. and T.J. Wong, 2002, "Corporate Ownership Structure and the Informativeness of Accounting Earnings in East Asia," Journal of Accounting and Economics 33, 401-425.

Guedhami, O., J. Pittman, and W. Saffar, 2009, "Auditor Choice in Privatized Firms: Empirical Evidence on the Role of State and Foreign Owners," Journal of Accounting and Economics 48, 151-171.

Gugler, K., 2003, "Corporate Governance, Dividend Payout Policy, and the Interrelation between Dividends, R&D, and Capital Investment," Journal of Banking and Finance 27, 1297-1321.

Gugler, K. and B.B. Yurtoglu, 2003, "Corporate Governance and Dividend Payout Policy in Germany," European Economic Review 47, 731-758.

Harford, J., S. Mansi, and W. Maxwell, 2008, "Corporate Governance and a Firm's Cash Holdings," Journal of Financial Economics 87, 535-555.

Hu, Y., D.D. Wang, and S. Zhang, 2008, "Founding Family Ownership, Management and Payout Policy," Annual Meeting of American Accounting Association, Anaheim, CA.

Isakov, D. and J-P. Weisskopf, 2015, "Pay-Out Policies in Founding Family Firms," Journal of Corporate Finance 33, 330-344.

Jensen, M.C., 1986, "Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers," American Economic Review 76, 323-329.

Johnson, S., P.D. Boone, A. Breach, and E. Friedman, 2000, "Corporate Governance in the Asian Financial Crisis," Journal of Financial Economics 58, 141-186.

Keynes, J.M., 1936, The General Theory of Employment, Interest, and Money, London, Macmillan.

Lang, L.H.P., R.M. Stulz, and R.A. Walkling, 1991, "A Test of the Free Cash Flow Hypothesis: The Case of Bidder Returns," Journal of Financial Economics 29, 315-335.

Lang, ML, K. Lins, and D. Miller, 2004, "Concentrated Control, Analyst Following, and Valuation: Do Analysts Matter Most When Investors Are Protected Least?" Journal of Accounting Research 42, 589-623.

La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. Vishny, 1998, "Law and Finance," Journal of Political Economy 106, 1113-1155.

La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. Vishny, 2000, "Agency Problems and Dividend Policies around the World," Journal of Finance 55, 1-33.

Lemmon, M., M. Roberts, and J. Zender, 2008, "Back to the Beginning: Persistence and the Cross-Sectional Distribution of Capital Structure," Journal of Finance 63, 1-37.

Lins, K.V, P. Volpin, and H.F. Wagner, 2013, "Does Family Control Matter? International Evidence from the 2008-2009 Financial Crisis," Review of Financial Studies 26, 2538-2619.

Masulis, R.W., P.K. Pham, and J. Zein, 2011, "Family Business Groups around the World: Financing Advantages, Control Motivations, and Organizational Choices," Review of Financial Studies 24, 3556-3600.

Miller, M. and K. Rock, 1985, "Dividend Policy under Asymmetric Information," Journal of Finance 40, 1031-1051.

Mitton, T., 2002, "A Cross-Firm Analysis of the Impact of Corporate Governance on the East Asian Financial Crisis," Journal of Financial Economics 64, 215-241.

Opler, T., L. Pinkowitz, R. Stulz, and R. Williamson, 1999, "The Determinants and Implications of Corporate Cash Holdings," Journal of Financial Economics 52, 3-46.

Pindado, J., I. Requejo, and C. de la Torre, 2012, "Do Family Firms Use Dividend Policy as a Governance Mechanism? Evidence from the Euro Zone," Corporate Governance: An International Review 20, 413-431.

Pindado, J. and C. de la Torre, 2008, "Financial Decisions as Determinants of Ownership Structure: Evidence from Spanish Family Controlled Firms," Managerial Finance 34, 868-885.

Rozeff, M.S., 1982, "Growth, Beta and Agency Costs as Determinants of Dividend Payout Ratios," Journal of Financial Research 5, 249-259.

Setia-Atmaja, L., G.A. Tanewski, and M. Skully, 2009, "The Role of Dividends, Debt, and Board Structure in the Governance of Family Controlled Firms," Journal of Business Finance and Accounting 36, 863-898.

Shleifer, A. and R. Vishny, 1986, "Large Shareholders and Corporate Control," Journal of Political Economy 94, 461-488.

Zhou, X., 2001, "Understanding the Determinants of Managerial Ownership and the Link between Ownership and Performance: Comment," Journal of Financial Economics 62, 559-572.

Najah Attig, Narjess Boubakri, Sadok El Ghoul, and Omrane Guedhami *

We thank an anonymous reviewer, Ruiyuan Chen, Figen Gunes Dogan (MFA discussant), Dusan Isakov, Jacob Kleinow (SWFA discussant), Mark Mietzner (EFA discussant), Raghavendra Rau (Editor), and participants at the 2015 Eastern Finance Association meetings (New Orleans), 2015 Southwestern Finance Association conference (Houston), and 2015 Midwest Finance Association Meeting (Chicago) for constructive comments and suggestions. We are grateful to Richard Carney for providing the ultimate ownership data. We appreciate generous financial support from Canada's Social Sciences and Humanities Research Council and excellent research assistance from He Wang.

* Najah Attig is an Associate Professor at Saint Mary's University in Halifax, Canada. Narjess Boubakri is a Professor at the American University of Sharjah in Sharjah, United Arab Emirates. Sadok El Ghoul is an Associate Professor at the University of Alberta in Edmonton, Canada. Omrane Guedhami is an Associate Professor at the University of South Carolina in Columbia, SC.

(1) Using data from Faccio, Lang, and Young (2001), Pindado, Requejo, and de la Torre (2012) find that in eurozone countries, family firms distribute higher and more stable dividends than non-family firms.

(2) Lins, Volpin, and Wagner (2013) argue that the survival of a family empire can be threatened by unexpected liquidity shocks. They state that an exogenous financial shock "moves firms out of equilibrium in a way that magnifies both the benefits and costs of family control" (p. 2584).

(3) For instance, La Porta et al. (2000) find that firms in countries with strong investor protection pay higher dividends than firms in countries with weak investor protection.

(4) We thank an anonymous reviewer for suggesting this alternative and the empirical strategy to test it.

(5) Dividends as a signal of private information should be more relevant for family firms than non-family firms, as the former are likely to have more pronounced information asymmetry problems (e.g., Attig et al., 2006; Anderson et al., 2009).

(6) We thank the reviewer for suggesting this procedure, which minimizes sample attrition due to negative net income.

(7) Our results are robust to using the 20% threshold (untabulated).

(8) Compustat data items are reported in parentheses.

(9) In Table V, we control for industry and country-year effects. We find similar results, and thus for brevity we report results using only country and industry-year effects in the subsequent analysis.

(10) Results with no replacement are similar (untabulated).

(11) Alternatively, one may employ an approach similar to a "difference-in-difference" procedure, that is: 1) create a dummy (AGENCY) for firms with above-median residual cash flows and below-median industry Tobin's Q, 2) augment the regression with AGENCY and its interaction with FAMILY, and 3) estimate the regression on the full sample. Our split- sample design, which closely follows Chen et al. (2011) and Lins et al. (2013), helps avoid multicollinearity stemming from high correlations between the test variables (here, FAMILY and AGENCY) and their interaction term, especially when the interaction comprises a dummy variable (e.g., Lang, Lins, and Miller, 2004; Guedhami, Pittman, and Saffar, 2009). In our case, the correlations between the interaction term and FAMILY and AGENCY are 0.42 and 0.61, respectively.
Table I. Sample Breakdown by Year, Country, and Industry

This table presents the sample distribution by year, country, and
industry (according to Campbell's, 1996, classification). The sample
comprises 4,285 observations representing 923 unique firms from 2006
to 2010.

Year                   N       %

2006                   867    20.23
2007                   878    20.49
2008                   873    20.37
2009                   850    19.84
2010                   817    19.07
Total                4,285      100

Country                N       %

Hong Kong              438    10.22
Indonesia              403     9.40
Japan                  534    12.46
Korea                  458    10.69
Malaysia               564    13.16
Philippines            261     6.09
Singapore              439    10.25
Taiwan                 665    15.52
Thailand               523    12.21
Total                4,285      100

Industry               N       %

Petroleum             152    3.55
Consumer durables     902    21.05
Basic industry        564    13.16
Food and tobacco      476    11.11
Construction          261    6.09
Capital goods         334    7.79
Transportation        306    7.14
Utilities             436    10.18
Textiles and trade    248    5.79
Services              189    4.41
Leisure               205    4.78
Other industries      212    4.95
Total                4,285    100

Table II. Descriptive Statistics Country

This table presents country means of the regression variables.
DIV/E is dividend payout defined as the ratio of cash dividends
(DV) to net income before extraordinary items (IB) (Compustat data
items are reported in parentheses). DIV/E equals 1 for cash
dividends that are larger than net income, and 0 for negative net
income firms that do not pay cash dividends. FAMILY is a dummy
variable equal to 1 if the largest ultimate owner at the 10%
threshold is a family, and 0 otherwise. RE/TE is the ratio of
retained earnings (RE) to common stockholders' equity (CEQ). TE/TA
is the ratio of common stockholders' equity (CEQ) to total assets
(AT). ROA is return on assets computed as ratio of net income
before extraordinary items (IB) to total assets. SGR is logarithmic
sales growth computed as log([SALE.sub.t]/[SALE.sub.t-1]). SIZE is
the natural logarithm of total assets in millions of $US. CH/TA is
the ratio of cash and short-term investments (CHE) to total assets
(AT). The sample comprises 4,285 observations representing 923
unique firms from 2006 to 2010.

Country         DIV/E   FAMILY   RE/TE   TE/TA

Hong Kong       0.329   0.532    0.115   0.586
Indonesia       0.312   0.541    0.264   0.472
Japan           0.469   0.112    0.640   0.449
Korea           0.359   0.635    0.089   0.397
Malaysia        0.448   0.557    0.426   0.547
Philippines     0.308   0.785    0.208   0.513
Singapore       0.407   0.560    0.391   0.541
Taiwan          0.459   0.140    0.160   0.523
Thailand        0.522   0.375    0.353   0.540
All countries   0.414   0.433    0.303   0.508

Country          ROA     SGR    SIZE    CH/TA

Hong Kong       0.040   0.099   6.629   0.198
Indonesia       0.068   0.150   5.969   0.141
Japan           0.034   0.019   9.592   0.141
Korea           0.046   0.122   8.281   0.129
Malaysia        0.075   0.108   6.405   0.149
Philippines     0.056   0.109   5.326   0.141
Singapore       0.065   0.138   6.210   0.182
Taiwan          0.063   0.090   7.228   0.198
Thailand        0.069   0.061   5.739   0.116
All countries   0.058   0.096   6.945   0.157

Table III. Descriptive Statistics and Correlation Matrix

This table presents descriptive statistics (Panel A) and a
correlation matrix (Panel B) for the regression variables. DIV/E is
dividend payout, defined as the ratio of cash dividends (DV) to net
income before extraordinary items (IB) (Compustat data items are
reported in parentheses). DIV/E equals 1 for cash dividends that
are larger than net income, and 0 for negative net income firms
that do not pay cash dividends. FAMILY is a dummy variable equal to
1 if the largest ultimate owner at the 10% threshold is a family,
and 0 otherwise. RE/TE is the ratio of retained earnings (RE) to
common stockholders' equity (CEQ). TE/TA is the ratio of common
stockholders' equity (CEQ) to total assets (AT). ROA is return on
assets, computed as ratio of net income before extraordinary items
(IB) to total assets. SGR is logarithmic sales growth, computed as
log([SALE.sub.t]/[SALE.sub.t-1]). SIZE is the natural logarithm of
total assets in millions of $US. CH/TA is the ratio of cash and
short-term investments (CHE) to total assets (AT). The p-value is
reported in parentheses below each correlation coefficient. The
sample comprises 4,285 observations representing 923 unique firms
from 2006 to 2010.

Panel A. Descriptive Statistics

            Mean          Q1        Median

DIV/E      0.414        0.112        0.339
FAMILY     0.433        0.000        0.000
RE/TE      0.303        0.129        0.400
TE/TA      0.508        0.356        0.496
ROA        0.058        0.020        0.053
SGR        0.096       -0.024        0.088
SIZE       6.945        5.588        6.718
CH/TA      0.157        0.064        0.121

             Q3           SD

DIV/E      0.682        0.349
FAMILY     1.000        0.496
RE/TE      0.648        0.640
TE/TA      0.665        0.208
ROA        0.096        0.081
SGR        0.210        0.396
SIZE       8.302        1.899
CH/TA      0.212        0.130

Panel B. Correlation Matrix

           DIV/E        FAMILY       RE/TE       TE/TA

DIV/E        1
FAMILY    -0.118 ***       1
          (0.000)
RE/TE    0.172 ***    -0.049 ***       1
          (0.000)      (0.001)
TE/TA      0.005      0.055 ***    0.141 ***       1
          (0.755)      (0.000)      (0.000)
ROA        -0.011       -0.002     0.354 ***   0.356 ***
          (0.458)      (0.911)      (0.000)     (0.000)
SGR      -0.137 ***     0.011       -0.004     -0.065 ***
          (0.000)      (0.493)      (0.775)     (0.000)
SIZE     0.094 ***    -0.213 ***   0.185 ***   -0.372 ***
          (0.000)      (0.000)      (0.000)     (0.000)
CH/TA     0.026 *       -0.021       0.016     0.391 ***
          (0.095)      (0.173)      (0.299)     (0.000)

            ROA          SGR         SIZE        CH/TA

DIV/E
FAMILY

RE/TE

TE/TA

ROA          1

SGR      0.174 ***        1
          (0.000)
SIZE     -0.125 ***     -0.023         1
          (0.000)      (0.139)
CH/TA    0.250 ***      -0.015     -0.154 ***      1
          (0.000)      (0.314)      (0.000)

*** Significant at the 0.01 level.

* Significant at the 0.10 level.

Table IV. Univariate Tests

This table presents differences in means (Panel A) and differences
in medians (Panel B) for the regression variables across non-family
and family firms. E is dividend payout defined as the ratio of cash
dividends (DV) to net income before extraordinary items (IB)
(Compustat data items are reported in parentheses). DIV/E equals 1
for cash dividends that are larger than net income, and 0 for
negative net income firms that do not pay cash dividends. RE/TE is
the ratio of retained earnings (RE) to common stockholders' equity
(CEQ). TE/TA is the ratio of common stockholders' equity (CEQ) to
total assets (AT). ROA is return on assets computed as ratio of net
income before extraordinary items (IB) to total assets. SGR is
logarithmic sales growth computed as
log([SALE.sub.t]/[SALE.sub.t-1]). SIZE is the natural logarithm of
total assets in millions of $US. CH/TA is the ratio of cash and
short-term investments (CHE) to total assets. The sample comprises
4,285 observations representing 923 unique firms from 2006 to 2010.

        FAMILY = 0   FAMILY = 1     Difference
           (1)          (2)       (3) = (2) - (1)     t-test

Panel A. Differences in Means

DIV/E     0.450        0.367          -0.083         -7.807 ***
RE/TE     0.330        0.267          -0.063         -3.217 ***
TE/TA     0.498        0.521           0.023          3.621 ***
ROA       0.058        0.058           0.000          0.112
SGR       0.093        0.101           0.008          0.685
SIZE      7.299        6.483          -0.816        -14.256 ***
CH/TA     0.159        0.154          -0.005         -1.364

Panel B. Differences in Medians

DIV/E     0.383        0.273          -0.110         -8.258 ***
RE/TE     0.401        0.400          -0.001         -2.710 ***
TE/TA     0.485        0.512           0.027          3.772 ***
ROA       0.052        0.054           0.002          0.119
SGR       0.080        0.098           0.018          2.974 ***
SIZE      7.225        6.295          -0.930        -14.037 ***
CH/TA     0.124        0.115          -0.009         -2.271 **

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

Table V. Family Firms and Dividend Payout

This table presents fixed-effects Tobit regression results. The
dependent variable is DIV/E, dividend payout defined as the ratio
of cash dividends (DV) to net income before extraordinary items
(IB) (Compustat data items are reported in parentheses). DIV/E
equals 1 for cash dividends that are larger than net income, and 0
for negative net income firms that do not pay cash dividends. The
independent variables are as follows. FAMILY is a dummy variable
equal to 1 if the largest ultimate owner at the 10% threshold is a
family, and 0 otherwise. RE/TE is the ratio of retained earnings
(RE) to common stockholders' equity (CEQ). TE/TA is the ratio of
common stockholders' equity (CEQ) to total assets (AT). ROA is
return on assets, computed as ratio of net income before
extraordinary items (IB) to total assets. SGR is logarithmic sales
growth, computed as log([SALE.sub.t]/[SALE.sub.t-1]). SIZE is the
natural logarithm of total assets in millions of SUS. CH/TA is the
ratio of cash and short-term investments (CHE) to total assets
(AT). In Column (1), the regression controls for country effects
and industry-year effects. In Column (2), the regression controls
for industry effects and country-year effects. In Column (3), the
regression is performed on a propensity-score-matched sample. We
run a probit model of FAMILY on firm characteristics as well as
country effects and industry-year effects. We then match each
family firm to the non-family firm with the closest score (with
replacement). The p-values based on standard errors clustered by
firm are reported in parentheses.

                        Industry-Year   Country-Year   PSM Sample
                           Effects        Effects
                             (1)            (2)           (3)

FAMILY                   -0.069 ***      -0.069 ***    -0.067 **
                           (0.006)        (0.006)       (0.029)
RE/TE                     0.204 ***      0.203 ***     0.201 ***
                           (0.000)        (0.000)       (0.000)
TE/TA                      -0.075          -0.067        -0.102
                           (0.274)        (0.322)       (0.264)
ROA                        -0.290          -0.321        -0.308
                           (0.153)        (0.113)       (0.239)
SGR                      -0.161 ***      -0.153 ***    -0.188 ***
                           (0.000)        (0.000)       (0.000)
SIZE                      0.027 ***      0.028 ***     0.035 ***
                           (0.001)        (0.001)       (0.004)
CH/TA                     0.231 **        0.237 **      0.274 **
                           (0.016)        (0.014)       (0.038)
Intercept                  -0.088          0.020         -0.243
                           (0.476)        (0.859)       (0.116)
Country effects              Yes             No           Yes
Industry-year effects        Yes             No           Yes
Industry effects             No             Yes            No
Country-year effects         No             Yes            No
N                           4,285          4,285         2,321
Pseudo-[R.sup.2]            0.115          0.116         0.118

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

Table VI. Family Firms and Dividend Payout:
The Role of the Agency Costs of Free Cash Flow

This table presents fixed effects Tobit regression results. The
dependent variable is DIV-E, dividend payout defined as the ratio of
cash dividends (DV) to net income before extraordinary items (IB)
(Compustat data items are reported in parentheses). DIV-E equals 1
for cash dividends that are larger than net income, and 0 for
negative net income firms that do not pay cash dividends. The
independent variables are as follows. FAMILY is a dummy variable
equal to 1 if the largest ultimate owner at the 10% threshold is a
family, and 0 otherwise. RE-TE is the ratio of retained earnings
(RE) to common stockholders' equity (CEQ). TE-TA is the ratio of
common stockholders' equity (CEQ) to total assets (AT). ROA is
return on assets computed as ratio of net income before
extraordinary items (IB) to total assets. SGR is logarithmic sales
growth, computed as log([SALE.sub.t]-[SALE.sub.t-1]). SIZE is the
natural logarithm of total assets in millions of $US. CH-TA is the
ratio of cash and short-term investments (CHE) to total assets (AT).
Each year, firms are assigned to one of four subsamples according to
whether their residual cash flow is lower (higher) than the median
and whether their industry Q is lower (higher) than the median. Cash
flow is defined as the ratio of the difference between operating
income before depreciation and capital expenditures to total assets.
Residual cash flow is the residual from a regression of cash flow on
FAMILY. Q is the ratio of total assets minus stockholder's equity
plus market capitalization, all deflated by total assets. The p-
values based on standard errors clustered by firm are reported in
parentheses.

                              Low Residual Cash Flow

                         Low Industry Q   High Industry Q
                              (1)               (2)

FAMILY                       -0.090           -0.092
                            (0.136)           (0.105)
RE/TE                      0.230 ***         0.368 ***
                            (0.001)           (0.000)
TE/TA                        -0.091            0.001
                            (0.526)           (0.995)
ROA                        -1.043 **          -1.276 **
                            (0.026)           (0.011)
SGR                        -0.176 ***          0.021
                            (0.000)           (0.722)
SIZE                       0.049 ***         0.066 ***
                            (0.008)           (0.001)
CH/TA                        0.377            0.445 *
                            (0.165)           (0.063)
Intercept                   -0.429 *          -0.391
                            (0.098)           (0.106)
Country effects               Yes               Yes
Industry--year effects        Yes               Yes
N                            1,283              810
Pseudo-[R.sup.2]             0.125             0.175

                              High Residual Cash Flow

                         Low Industry Q   High Industry Q
                              (3)               (4)

FAMILY                     -0.089 ***         -0.052
                            (0.009)           (0.106)
RE/TE                      0.153 ***           0.079
                            (0.001)           (0.133)
TE/TA                        -0.004           -0.112
                            (0.969)           (0.241)
ROA                         -0.606 *         0.583 ***
                            (0.056)           (0.046)
SGR                        -0.252 ***       -0.526 ***
                            (0.000)           (0.000)
SIZE                         -0.014           -0.007
                            (0.281)           (0.553)
CH/TA                       0.346 **          -0.033
                            (0.010)           (0.799)
Intercept                    0.274           0.473 ***
                            (0.104)           (0.006)
Country effects               Yes               Yes
Industry--year effects        Yes               Yes
N                            1,120              975
Pseudo-[R.sup.2]             0.185             0.255

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.

Table VII. Family Firms and Dividend Changes

This table presents fixed effects logit regression results. In
Columns (1) and (2), the dependent variable is DIV_INCR, a dummy
variable equal to 1 if cash dividends in year t are higher than
cash dividends in year t - 1, and 0 otherwise. In Columns (3) and
(4), the dependent variable is DIV_DECR, a dummy variable equal to
1 if cash dividends in year t are lower than cash dividends in year
t - 1, and 0 otherwise. In Columns (5) and (6), the dependent
variable is DIV_OMIT, a dummy variable equal to 1 if a
dividend-paying firm in year t - 1 does not pay dividends in year
t, and 0 otherwise. The independent variables are as follows.
FAMILY is a dummy variable set to 1 if the largest ultimate owner
at the 10% threshold is a family, and 0 otherwise. RE/TE is the
ratio of retained earnings (RE) to common stockholders' equity
(CEQ) (Compustat data items are reported in parentheses). TE/TA is
the ratio of common stockholders' equity (CEQ) to total assets
(AT). ROA is return on assets, computed as ratio of net income
before extraordinary items (IB) to total assets. SGR is logarithmic
sales growth, computed as log([SALE.sub.t]/[SALE.sub.t-1]). SIZE is
the natural logarithm of total assets in millions of $US. CH/TA is
the ratio of cash and short-term investments (CHE) to total assets
(AT). [DELTA]ROA is the change in ROA from year t - 1 to year t.
The p-values based on standard errors clustered by firm are
reported in parentheses.

                              DIV_NCR                   DIV_DECR

                        (1)          (2)          (3)          (4)

FAMILY               -0.186 **    -0.199 **    0.232 ***    0.236 ***
                      (0.029)      (0.020)      (0.006)      (0.005)
FAMILYx[DELTA]ROA                 -2.035 **                   -0.595
                                   (0.043)                   (0.526)
RE/TE                0.469 ***    0 437 ***    0.576 ***    0.574 ***
                      (0.001)      (0.001)      (0.000)      (0.000)
TE/TA                  0.119        0.119      -0.585 **    -0.606 **
                      (0.602)      (0.606)      (0.013)      (0.010)
ROA                  7.027 ***    7.456 ***    -2.962 ***   -2.846 ***
                      (0.000)      (0.000)      (0.000)      (0.000)
SGR                    0.157        0.155      -0.362 ***   -0.358 ***
                      (0.161)      (0.166)      (0.001)      (0.001)
SIZE                 0.153 ***    0.158 ***     0.052 *       0.050
                      (0.000)      (0.000)      (0.096)      (0.109)
CH/TA                  0.181        0.122        -0.029       -0.031
                      (0.587)      (0.715)      (0.932)      (0.926)
Intercept            -2.151 ***   -2.175 ***   -1.667 ***   -1.652 ***
                      (0.000)      (0.000)      (0.005)      (0.006)
Country effects         Yes          Yes          Yes          Yes
Industry-year
  effects               Yes          Yes          Yes          Yes
N                      4,217        4,205        4,217        4,205
Pseudo-[R.sup.2]       0.124        0.126        0.0873       0.0874

                              DIV_OMIT

                         (5)           (6)

FAMILY                 0.431 *      0.456 **
                       (0.051)       (0.040)
FAMILYx[DELTA]ROA                   3.885' *
                                     (0.025)
RE/TE                   0.079         0.151
                       (0.564)       (0.313)
TE/TA                  -0.144        -0.096
                       (0.810)       (0.876)
ROA                  -5.489 ***    -6.344 ***
                       (0.000)       (0.000)
SGR                    -0.122        -0.104
                       (0.635)       (0.692)
SIZE                   -0.057        -0.061
                       (0.426)       (0.393)
CH/TA                   0.027        -0.023
                       (0.975)       (0.979)
Intercept            -17.315 ***   -17.554 ***
                       (0.000)       (0.000)
Country effects          Yes           Yes
Industry-year
  effects                Yes           Yes
N                       2,844         2,839
Pseudo-[R.sup.2]        0.124         0.129

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.

Table VIII. Family Firms and Dividend Payout during
the 2008-2009 Global Financial Crisis

This table presents fixed effects Tobit regression results. The
dependent variable is DIV/E, dividend payout defined as the ratio
of cash dividends (DV) to net income before extraordinary items
(IB) (Compustat data items are reported in parentheses). DIV/E
equals 1 for cash dividends that are larger than net income, and 0
for negative net income firms that do not pay cash dividends. The
independent variables are as follows. FAMILY is a dummy variable
equal to 1 if the largest ultimate owner at the 10% threshold is a
family, and 0 otherwise. D(2008-2009) is a dummy variable equal to
1 for 2008 and 2009, and 0 otherwise. RE/TE is the ratio of
retained earnings (RE) to common stockholders' equity (CEQ). TE/TA
is the ratio of common stockholders' equity (CEQ) to total assets
(AT). ROA is return on assets, computed as ratio of net income
before extraordinary items (IB) to total assets. SGR is logarithmic
sales growth, computed as log([SALE.sub.t]/[SALE.sub.t-1]). SIZE is
the natural logarithm of total assets in millions of $US. CH/TA is
the ratio of cash and short-term investments (CHE) to total assets
(AT). The p-values based on standard errors clustered by firm are
reported in parentheses.

                            (1)

FAMILY                    -0.032
                          (0.206)
FAMILY x D/2008-2009)     -0.092 ***
                          (0.000)
RE/TE                      0.204 ***
                          (0.000)
TE/TA                     -0.075
                          (0.269)
ROA                       -0.288
                          (0.154)
SGR                       -0.160 ***
                          (0.000)
SIZE                       0.027 ***
                          (0.001)
CH/TA                      0.228 **
                          (0.018)
Intercept                 -0.092
                          (0.459)
Country effects             Yes
Industry-year effects       Yes
N                          4,285
Pseudo-[R.sup.2]           0.116

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

Table IX. Family Firms, Cash Holdings, and Investment

This table presents fixed-effects regression results. In Columns
(1) and (2), the dependent variable is CH/TA, the ratio of cash and
short-term investments (CHE) to total assets (AT) (Compustat data
items are reported in parentheses). In Columns (3) and (4), the
dependent variable is CAPX/TA, the ratio of capital expenditures
(CAPX) to total assets (AT). The independent variables are as
follows. FAMILY is a dummy variable equal to 1 if the largest
ultimate owner at the 10% threshold is a family, and 0 otherwise.
RE/TE is the ratio of retained earnings (RE) to common
stockholders' equity (CEQ). TE/TA is the ratio of common
stockholders' equity (CEQ) to total assets (AT). ROA is return on
assets, computed as ratio of net income before extraordinary items
(IB) to total assets. SGR is logarithmic sales growth, computed as
log([SALE.sub.t]/[SALE.sub.t-1]). SIZE is the natural logarithm of
total assets in millions of $US. The p-values based on standard
errors clustered by firm are reported in parentheses.

                               CH/TA                  CAPX/TA

                          (1)         (2)         (3)         (4)

FAMILY                   -0.006      -0.006      0.002       0.002
                        (0.410)     (0.405)     (0.534)     (0.583)
FAMILY x [DELTA]ROA                -0.176 ***              -0.084 ***
                                    (0.000)                 (0.000)
RE/TE                  -0.014 ***  -0.016 ***    0.000       -0.000
                        (0.004)     (0.001)     (0.844)     (0.888)
TE/TA                  0.192 ***   0.188 ***     -0.007      -0.008
                        (0.000)     (0.000)     (0.368)     (0.307)
ROA                    0.307 ***   0.346 ***   0.062 ***   0.077 ***
                        (0.000)     (0.000)     (0.002)     (0.000)
SGR                      -0.008      -0.008     0.007 **    0.007 **
                        (0.204)     (0.193)     (0.014)     (0.013)
SIZE                   -0.006 **   -0.006 **     0.001       0.001
                        (0.021)     (0.021)     (0.487)     (0.453)
CH/TA                                          -0.068 ***  -0.070 ***
                                                (0.000)     (0.000)
Intercept              0.091 ***   0.092 ***   0.065 ***   0.065 ***
                        (0.006)     (0.006)     (0.000)     (0.000)
Country effects           Yes         Yes         Yes         Yes
Industry-year effects     Yes         Yes         Yes         Yes
N                        4,285       4,272       4,279       4,266
Adj. [R.sup.2]           0.240       0.244       0.112       0.116

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

Table X. Robustness to Alternative Payout Variables

This table presents fixed-effects Tobit regression results. In
Column (1), the dependent variable is DIV/A, the ratio of cash
dividends (DV) to total assets (AT) (Compustat data items are
reported in parentheses). In Column (2), the dependent variable is
DIV/S, the ratio of cash dividends (DV) to sales (SALE). In Column
(3), the dependent variable is TOTP/E, total payout defined as the
ratio of the sum of cash dividends (DV) and stock repurchases
(PRSTKC) to net income before extraordinary items (IB). TOTP/E is
equal to 1 if total payout is larger than net income. TOTP/E is
equal to 0 for negative net income firms that do not pay cash
dividends nor repurchase shares. The independent variables are as
follows. FAMILY is a dummy variable equal to 1 if the largest
ultimate owner at the 10% threshold is a family, and 0 otherwise.
RE-TE is the ratio of retained earnings (RE) to common stockholders'
equity (CEQ). TE/TA is the ratio of common stockholders' equity
(CEQ) to total assets (AT). ROA is return on assets computed as
ratio of net income before extraordinary items (IB) to total assets.
SGR is logarithmic sales growth computed as log([SALE.sub.t]/
[SALE.sub.t-1]). SIZE is the natural logarithm of total assets in
millions of $US. CH/TA is the ratio of cash and short-term
investments (CHE) to total assets (AT). The p-values based on
standard errors clustered by firm are reported in parentheses.

                         DIV/A        DIV/S       TOTP/E
                           (1)         (2)         (3)

FAMILY                  -0.006 ***  -0.007 *    -0.039 **
                         (0.003)     (0.098)     (0.045)
RE/TE                   0.010 ***   0.021 ***    0.050 **
                         (0.000)     (0.000)     (0.012)
TE/TA                    -0.002     0.075 ***     0.060
                         (0.789)     (0.000)     (0.245)
ROA                     0.358 ***   0.364 ***   -1.390 ***
                         (0.000)     (0.000)     (0.000)
SGR                     -0.019 ***  -0.034 ***  -0.122 ***
                         (0.000)     (0.000)     (0.000)
SIZE                      0.000     0.008 ***     0.009
                         (0.625)     (0.000)     (0.191)
CH/TA                   0.034 ***     0.027     0.253 ***
                         (0.001)     (0.204)     (0.000)
Intercept               -0.025 **   -0.115 ***  0.291 ***
                         (0.025)     (0.000)     (0.007)
Country effects            Yes         Yes         Yes
Industry-year effects      Yes         Yes         Yes
N                         4,285       4,285       4,285
Pseudo-[R.sup.2]         -0.248      -0.265       0.276

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.

Table XI. Robustness to Sample Composition

This table presents fixed-effects Tobit regression results. The
dependent variable is DIV/E, dividend payout defined as the ratio
of cash dividends (DV) to net income before extraordinary items
(IB) (Compustat data items are reported in parentheses). DIV/E is
equal to 1 if cash dividends are larger than net income. DIV/E is
equal to 0 for negative net income firms that do not pay cash
dividends. The independent variables are as follows. FAMILY is a
dummy variable set to 1 if the largest ultimate owner at the 10%
threshold is a family, and 0 otherwise. RE/TE is the ratio of
retained earnings (RE) to common stockholders' equity (CEQ). TE/TA
is the ratio of common stockholders' equity (CEQ) to total assets
(AT). ROA is return on assets, computed as ratio of net income
before extraordinary items (IB) to total assets. SGR is logarithmic
sales growth, computed as log([SALE.sub.t]/[SALE.sub.t-1]).
SIZE is the natural logarithm of total assets in millions of $US.
CH/TA is the ratio of
cash and short-term investments (CHE) to total assets (AT). Panel A
presents regressions that sequentially exclude each country from
the sample. Panel B shows weighted regressions where the weights
equal the inverse of the number of observations in each country.
The p-values based on standard errors clustered by firm are
reported in parentheses.

                          Panel A. Results after Sequentially
                                  Excluding Each Country

                        Hong Kong (1)   Indonesia (2)   Japan (3)

FAMILY                   -0.089 ***      -0.111 ***     -0.104 ***
                           (0.000)         (0.000)       (0.000)
RE/TE                     0.201 ***       0.195 ***     0.217 ***
                           (0.000)         (0.000)       (0.000)
TE/TA                      -0.050        -0.134 * *       -0.014
                           (0.487)         (0.066)       (0.860)
ROA                        -0.187          -0.172         -0.085
                           (0.384)         (0.424)       (0.684)
SGR                      -0.186 ***      -0.171 ***     -0.167 ***
                           (0.000)         (0.000)       (0.000)
SIZE                      0.017 **         0.013 *      0.033 ***
                           (0.018)         (0.084)       (0.000)
CH/TA                       0.120           0.156         0.170
                           (0.220)         (0.114)       (0.101)
Intercept                   0.149          0.203 *        -0.019
                           (0.175)         (0.064)       (0.881)
Country effects              Yes             Yes           Yes
Industry-year effects        Yes             Yes           Yes
N                           3,847           3,882         3,751
Pseudo-[R.sup.2]           0.0938          0.0952         0.0863

                        Korea (4)    Malaysia (5)   Philippines (6)

FAMILY                  -0.100 ***    -0.092 ***      -0.085 ***
                         (0.000)       (0.000)          (0.000)
RE/TE                   0.190 ***     0.192 ***        0.183 ***
                         (0.000)       (0.000)          (0.000)
TE/TA                     -0.019        -0.067         -0.123 *
                         (0.803)       (0.388)          (0.076)
ROA                       0.069         -0.068          -0.173
                         (0.738)       (0.744)          (0.388)
SGR                     -0.172 ***    -0.154 ***      -0.175 ***
                         (0.000)       (0.000)          (0.000)
SIZE                    0.021 ***     0.025 ***          0.011
                         (0.009)       (0.001)          (0.119)
CH/TA                     0.118         0.067            0.138
                         (0.234)       (0.500)          (0.141)
Intercept                 0.064         0.035          0.232 **
                         (0.580)       (0.745)          (0.034)
Country effects            Yes           Yes              Yes
Industry-year effects      Yes           Yes              Yes
N                         3,827         3,721            4,024
Pseudo-[R.sup.2]          0.0933        0.0945          0.0888

                        Singapore (7)   Taiwan (8)   Thailand (9)

FAMILY                   -0.099 ***     -0.072 ***    -0.082 ***
                           (0.000)       (0.005)       (0.001)
RE/TE                     0.178 ***     0.187 ***     0.178 ***
                           (0.000)       (0.000)       (0.000)
TE/TA                      -0.080         -0.122        -0.079
                           (0.292)       (0.102)       (0.299)
ROA                        -0.115         -0.171        -0.227
                           (0.593)       (0.423)       (0.299)
SGR                      -0.155 ***     -0.183 ***    -0.169 ***
                           (0.000)       (0.000)       (0.000)
SIZE                      0.019 ***     0.021 ***     0.035 ***
                           (0.009)       (0.004)       (0.000)
CH/TA                     0.200 **        0.132       0.268 ***
                           (0.049)       (0.216)       (0.008)
Intercept                   0.151         0.138         -0.005
                           (0.188)       (0.200)       (0.965)
Country effects              Yes           Yes           Yes
Industry-year effects        Yes           Yes           Yes
N                           3,846         3,620         3,762
Pseudo-[R.sup.2]           0.0884         0.0994        0.103

                            Panel B.
                            Weighted
                         Regression (10)

FAMILY                      -0.068 **
                             (0.013)
RE/TE                       0.207 ***
                             (0.000)
TE/TA                         -0.041
                             (0.588)
ROA                           -0.263
                             (0.211)
SGR                         -0.156 ***
                             (0.000)
SIZE                        0.037 ***
                             (0.000)
CH/TA                        0.236 **
                             (0.027)
Intercept                     -0.190
                             (0.131)
Country effects                Yes
Industry-year effects          Yes
N                             4,285
Pseudo-[R.sup.2]              0.120

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.
COPYRIGHT 2016 Financial Management Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2016 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Attig, Najah; Boubakri, Narjess; Ghoul, Sadok El; Guedhami, Omrane
Publication:Financial Management
Article Type:Report
Geographic Code:90ASI
Date:Jun 22, 2016
Words:12289
Previous Article:Product market competition and financial decisions during a financial crisis.
Next Article:News-specific price discovery in credit default swap markets.
Topics:

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |