MASTER THESIS FINANCE · SEPTEMBER 2010
CORPORATE CASH RATIOS IN FAMILY AND
NON-FAMILY FIRMS
EVIDENCE FROM THE EUROPEAN UNION
A.B. LIER1
Abstract
This study examines whether target cash ratios and the speed with which cash ratios are adjusted to their targets, differ between family and non-family firms. I perform a random effects panel analysis on a balanced sample of 1178 publicly listed family and non-family firms in the European Union, as per January 1st 1995, over 2003 – 2008. In contrast to the assumption of ownership stability frequently encountered in the empirical cash literature, I check for ownership consistency in multiple years. This may explain why, in contrast to other studies, I find that family firms maintain similar cash ratios compared to non-family firms. Additionally, I find that family firms adjust their cash ratios more slowly to their target ratios than their non-family counterparts. This latter finding is in accordance with family firms’ increased monitoring abilities and incentives to limit managerial discretion. In line with previous literature, I find that younger and smaller firms and those with less non-cash working capital have higher cash ratios. Additionally, I find that more leverage and higher and less volatile cash flows result in lower cash ratios. These results are fairly consistent for both family and non-family firms.
JEL code: G32, G34, G35
Keywords: Cash ratio; target; adjustment speed, family firm; non-family firm
1. Introduction
The topic of corporate cash holdings has for a long time been neglected in the empirical literature due to the theorems of Modigliani and Miller (1958), amongst others. In recent years, however, this topic has gained more attention (Opler, Pinkowitz, Stulz and Williamson, 1999; Dittmar, Mahrt-Smith and Servaes, 2003; Ozkan and Ozkan, 2004), possibly due to significant and rapidly growing corporate cash ratios and acknowledged limitations of market perfection. Bates, Kahle and Stulz (2009) show that the average cash ratio has increased in the United States from 10.5% in 1980 to 23.2% in 2006. This is in line with the observation made by Von Eije (2010), who has shown an increase in the mean cash ratio from 11% to 18% in the European Union (EU) over the 1989 - 2005 period.
1 University of Groningen, Faculty of Economics & Business, MSc Business Administration, S1355619
An important strand of cash-literature focuses on the determinants of target cash ratios.2 For example, Kim, Mauer and Sherman (1988) analyze the determinants of cash ratios for a sample of companies located in the United States (US). They find that these firms maintain higher cash ratios when facing higher costs of external capital, higher variability of future cash flows and a lower return on physical assets. Opler et al. (1999) find higher cash ratios in US firms with strong growth opportunities and riskier cash flows, whereas large firms and those with high credit ratings tend to hold lower cash ratios. International studies show that the countries’ legal structure also impacts cash ratios. Dittmar, Mahrt-Smith and Servaes (2003) examine firms from 45 countries and find that corporations in countries where shareholder rights are not well protected hold up to twice the cash ratio compared to corporations in countries with good shareholder protection.
When examining cash ratios, most studies indirectly assume that companies are homogenous entities with regards to their ownership structure. Yet some studies include a measure of ownership concentration in their analysis. Guney, Ozkan and Ozkan (2003) and Ferreira and Vilela (2004) observe that the presence of large shareholders can curb managerial discretion which leads to lower cash ratios. Zhang (2005) on the other hand finds that increased ownership concentration is positively correlated with cash ratios, supporting the view that there are private benefits of control not necessarily shared by minority shareholders.
It is my thesis that it is the identity of large shareholders along with their priorities and preferences, not ownership concentration per se, that determines cash ratios. In this respect, I focus on the difference between family and non-family firms, as many authors claim that family firms differ substantially from non-family firms (James, 1999; Claessens, Djankov and Lang, 2000; Morresi, 2010). Moreover family firms are one of the major forms of ownership structure around the world (La Porta, Lopez-De Silanes, Shleifer and Vishny, 2000).
In this study, I examine a balanced panel of 1178 publicly listed family and non-family firms in the European Union, as per January 1st 1995, over 2003 – 2008. I consider firms in which family shareholders hold a minimum of 20% of the total voting rights in addition to being the largest shareholder as family firms. All other firms, in which family shareholders hold no more than 10% of the total voting rights, are defined as non-family firms. In contrast to the assumption of ownership stability frequently encountered in the empirical literature, all firms are checked for ownership consistency in multiple years, 2005 and 2008. Using random effects panel regressions on the static trade-off model, I find in contrast to the observations of Ozkan and Ozkan (2004) and Ginglinger and Saddour (2007), whom both assume ownership stability, that family and non-family firms maintain similar cash ratios. Subsequently, I estimate a partial
2 The other strand of literature focuses on the relation between cash ratios and firm performance (Harford, 1999; Mikkelson and
adjustment trade-off model, where I find that both firm types adjust their cash ratio to approximately 48% of their three-year-average historical cash ratio. When using a better specified target, I observe that cash ratios of family firms adjust more slowly to the specified target ratio compared to non-family firms, with 21% versus 31% respectively. This latter finding is in accordance with family firms’ increased monitoring abilities and incentives to limit managerial discretion. Furthermore, I find that younger and smaller firms and those with less non-cash working capital have higher cash ratios. Additionally, I find that more leverage and higher and less volatile cash flows result in lower cash ratios. These results are fairly consistent for both family and non-family firms.
My findings contribute to the literature on cash holding decisions of firms in several ways. First, to my knowledge no other study has previously specifically addressed the impact of firm ownership types, especially family firms, on cash ratios. In other finance literature, concerning dividends payment for instance, a distinction has since long been made between family and non-family firms (Faccio, Lang and Young, 2001; Setia-Atmaja, Tanewski and Skully, 2009). There is reason to believe this difference does also impact cash ratios. Second, I show that the standard assumption of ownership stability, frequently encountered in the empirical literature (Opler et al., 1999, Kalcheva and Lins, 2007), is not appropriate. In this study I disregard nearly 20% of the final sample due to changes in ownership type over a four year period. Third, in contrast to most other studies examining cash ratios, I consider the endogeneity problem in the empirical analysis of cash ratios. It is highly likely that cash holdings of firms can also affect some of the firm-specific characteristics such as leverage and market-to-book ratios. Therefore I use one-year lagged variables to correct for possible endogeneity.
The remainder of this study is structured as follows: in section 2, I discuss the theoretical framework on which the hypotheses, developed at the end of that section, will be based. Section 3 addresses the data, followed by a description of the methodology in section 4. In section 5 I present an overview of the results and conclude with a discussion of the obtained results and their position in the existing literature in section 6.
2. Theoretical foundations and hypotheses
invest in liquid assets would leave shareholder wealth unchanged (Opler et al., 1999). Yet, the capital structure irrelevance principle of Modigliani and Miller (1958) only holds under stringent assumptions3 and the existence of market imperfections implies that there is, in fact, a target cash ratio that balances costs and benefits and ultimately maximizes firm value.
The empirical literature emphasizes two models on target liquidity ratios: the static trade-off model and the financing hierarchy model (Opler et al., 1999; Lins, Servaes and Tufano, 2008; Bates, Kahle and Stulz, 2009). The static trade-off model is more frequently supported in describing target cash ratios (Kim, Mauer and Sherman, 1998; Opler et al., 1999, Bruinshoofd and Kool, 2004). However, many proxies used to test the predictions of the static trade-off model are identical to the financing hierarchy model. This makes it difficult to ultimately discriminate between both models (Jani, Hoesli and Bender, 2004; Lins, Servaes and Tufano, 2008). Bruinshoofd and Kool (2004; 2009) and Jani, Hoesli and Bender (2004) argue that these two models in practice complement each other. They both explain the process that determines target corporate cash ratios, but they do so over different time horizons. The static trade off model defines the long run desired ratio while the financing hierarchy model is the driving force of the short run dynamics of cash ratios.
The static trade-off model states that in order to maximize shareholder wealth, firm management should set a firm’s cash ratio at such a level that its marginal benefit to the shareholders equals its marginal cost (Almeida, Campello and Weisbach, 2004). Keynes (1936) divides the benefits of cash in the transaction motive and the precautionary motive. The transaction motive implies that converting a non-cash financial asset into non-cash is accompanied by transaction costs. In order to finance day-to-day operations, firms prefer to retain some liquidity in order to avoid these costs (Bates, Kahle and Stulz, 2009; Ramirez and Tadesse, 2009). The precautionary motive suggests that firms retain cash within the firm in order to avoid raising external funds, when this is expensive or unavailable (Kim, Mauer and Sherman, 1998). The general costs to holding cash lies in the lower rate of return due to existence of a liquidity premium and tax disadvantages (Opler et al. 1999; Lee and Powel, 2010). Equating the marginal benefits and costs will theoretically lead to a target cash ratio. However, the target cash ratio may be affected by the fact that various stakeholders of a firm frequently view the costs and benefits of cash differently. Jensen and Meckling (1976) argue that management has a greater preference for cash, because it reduces firm risk and increases managerial discretion. Also, conflicts of interest may exist between majority and minority shareholders.
In this study it is my thesis that the identity of the controlling shareholder is a major factor influencing cash ratios. I specifically concentrate on cash ratios in family and non-family firms and
3 The capital structure irrelevance theory assumes no taxes or transaction costs exist and that individuals and corporations borrow
therefore first examine the definitions of a family firm frequently encountered in the empirical literature. Subsequently, I examine the agency costs theories and how these impact target cash ratios and the adjustment speed with which cash ratios are adjusted to their target in family and non-family firms. Finally, I discuss the variables which may influence the target cash ratios and the adjustment speed.
Defining a family firm
The empirical literature does not utilize a consistent description of what constitutes a family firm. Most studies state that when a firm has a shareholder holding more than a certain minimum of voting or cash flow rights, the firm has a controlling shareholder. If the largest controlling shareholder is a family, this firm is characterized as a family firm (La Porta, Lopez-de-Silanes, Shleifer and Vishny, 1999; Faccio and Lang, 2002; Claessens, Djankov and Lang, 2002; Ginglinger and L’Her, 2006). However, the minimal percentage of voting or cash flow rights which must be held in order to classify as a controlling shareholder differs greatly. Peng and Jiang (2010) use a 5% threshold, whereas Croci, Doukas and Gonenc (2010) use a 10% cut-off and Ginglinger and Saddour (2007) use a 33% minimum. These three studies all contain a voting right minimum in order for the largest shareholder to be deemed to have effective control over the firm. The frequent emphasis on voting rights follows from the fact that shareholders are able to obtain control over a firm in excess of their cash flow rights via pyramid structures, cross-shareholdings and via the acquisition of shares with superior voting rights. The indirect voting rights obtained through pyramid structures and cross-shareholdings are equally important as direct voting rights according to La Porta et al. (1999), Faccio and Lang (2002) and Ozkan and Ozkan (2004). These authors also argue that if the directly controlling shareholder is controlled by a controlling shareholder, the latter should be classified as the firm’s ultimate owner. If this is a family, the firm is then classified as family owned. Overall, the empirical literature most commonly uses a minimum of 20% control to identify controlling shareholders (Claessens, Djankov and Lang, 2002; Faccio and Lang, 2002). The wide diversity of family firm definitions encountered in empirical literature limits the comparison of results found in different studies. Nevertheless, Croci, Doukas and Gonenc (2010) find consistent results when using either a 10% or 20% control threshold.
Agency costs and corporate cash holdings
Shareholders versus management
Berle and Means (1932) first pointed to the conflicts of interest that exist between shareholders and managers. Because managers maintain insulated positions within a firm, information asymmetry may exist between them and the firm’s shareholders. The degree of information asymmetry depends on the presence of a large shareholder, for whom it is cost-efficient to monitor management. The absence of a large shareholder and the ensuing free-rider dilemma4 will cause a firm’s management to become entrenched. One of the potential conflicts between management and shareholders regards the amount of cash to be retained within the firm. Jensen (1986) argues that management prefers to increase the amount of funds under their control, allowing them to obtain perquisites and make inefficient investment decisions. In line with this statement, Shleifer and Vishny (1997) and Guney, Ozkan and Ozkan (2003) find that management, if given the opportunity, may act selfishly by maximizing their own wealth, power and prestige at the expense of shareholders. Moreover, Tong (2006) finds that more risk-averse managers prefer to retain higher cash ratios. Overall, Kalcheva and Lins (2007) find that cash ratios are higher in firms with higher managerial entrenchment. In contrast to the overwhelming international evidence, Harford, Mansi and Maxwell (2008) provide evidence for the US-market suggesting that entrenched managers are more likely to build excess cash balances, but subsequently spend excess cash quickly. When the above is the case and underinvestment occurs (Amihud and Lev, 1981), holding cash becomes more costly. Dittmar and Mahrt-Smith (2007) show that shareholders view the value of an additional dollar of cash held by a poorly governed firm lies between $0.42 and $0.88, whereas the value of cash lies well above one dollar when appropriately managed.
Family firms are, however, less likely to be confronted with agency costs of this nature. First, family firms frequently have a member of their family in the top-management or board of directors of the firm resulting in less information asymmetry between them (Burkart, Panunzi, Shleifer, 2003). Second, even if there is no active participation by the family in the management of the firm, the amount of information opacity is supposedly still reduced, because, a family’s wealth is closely linked to the continued welfare and performance of the firm. Anderson and Reeb (2003a) and Holmen and Nivrozhkin (2004) document that families in family firms, in respectively the US and Sweden, have more than 50% of their total wealth invested in the firms they control. It is likely that these families require being adequately informed. Finally, Casson (1999) shows that families want to preserve their company for the next generation. For all these reasons, Anderson, Mansi and Reeb (2003) and Ben-Amar and Andre (2006) argue that families, compared to other types of shareholders, are better monitors of managers.
4 The free-rider dilemma states that when there are multiple small shareholders, a combined incentive to monitor management will
In conclusion, interest alignment between family shareholders and management, reduced information opacity as well as increased monitoring incentives cause family firms to be subject to fewer agency costs than non-family firms. Therefore cash ratios in family firms are expected to be lower than in non-family firms.
Majority versus minority shareholders
Although controlling shareholders have an incentive to monitor management and curb their discretion, they also have private benefits of control not necessarily shared by the minority shareholders (Faccio, Lang and Young, 2001). According to DeAngelo and DeAngelo (2000) private benefits are extracted through special dividends, extreme compensation schemes and business dealings with co-controlled companies (tunneling). This is easier when higher cash ratios are maintained and consequently, controlling shareholders may prefer their firms to retain more cash. In line with these statements, Dittmar, Mahrt-Smith and Servaes (2003) find evidence suggesting that firms have higher cash ratios in countries with more agency problems. Moreover, Pinkowitz, Stulz and Williamson (2006; 2007) find that cash is worth less to minority shareholders of firms in countries with low shareholder protection.
Conflicts with minority shareholders are further exacerbated when the controlling shareholder has voting rights in excess of cash flow rights (Zhang, 2005), e.g. as a result of dual class shares, pyramid structures and/or cross-holdings. Especially, as the costs of expropriation, translated in a lower market value of the firm, are mostly borne by those expropriated (Attig, Fischer and Gadhoum, 2004). In line with this statement, Barclay and Holderness (1989), Johnson, La Porta, Lopez-de-Silanes and Shleifer (2000) and Claessens, Djankov and Lang (2002) show that voting shares trade at a substantial premium over non-voting shares. Potential shareholders are therefore less inclined to invest in these firms, hence making external finance more difficult to obtain.
firms appear to reduce agency costs associated with minority shareholder expropriation as these firms are more valuable than non-family firms.
Overall, it cannot be conclusively determined whether or not family firms expropriate minority shareholders more than non-family firms. However, the findings in the empirical literature tend to agree that they do. If this is indeed the case, it is to be expected that family firms hold higher cash ratios than their non-family counterparts.
Leverage
The previous subsections discussed the financing relationship between a firm and its shareholders. However, firms receive financing from debt holders as well. In contrast to non-family firms, family firms have a preference for debt financing as this provides family shareholders with several advantages. First, family shareholders have a desire to retain a control over their firm which would otherwise be diluted when accessing the equity capital market (Burkhart, Panunzi and Shleifer, 2003). Second, due to more information opacity within the firm, family firms face a higher cost of equity (Brav, 2009). Third, because banks have a comparative advantage over public debt in reducing information asymmetry and agency costs, families, by issuing bank debt, can reduce their cost of external financing (Diamond, 1984). Fourth and finally, because families wish to someday transfer the family firm to the next generation, they have a preference for low-risk financing strategies (Croci, Doukas and Gonenc, 2010).
The preference for debt financing in family firms, matches the preference of debt holders to invest in family firms (Ferreira and Vilela, 2004). This is a consequence of the low-risk investment strategy of family firms which is preferred over the risk-seeking investment strategy of non-family firms. As a result, Croci, Doukas and Gonenc (2010) show that credit markets provide family firms with more long-term debt and less short-term debt than non-family firms.
Growth opportunities
Various empirical studies have shown that the existence of growth opportunities positively affects cash ratios (Opler et al., 1999; Ferreira and Vilela, 2004; Ozkan and Ozkan, 2004). The reason for this positive correlation is twofold. First, Myers and Majluf (1984) point out that firms whose value is largely determined by growth options experience greater information asymmetry. Second, growth firms experience higher agency costs of debt. These costs arise because risk-averse debt holders are averse to their funds being used to finance high-risk growth investments. Myers (1977) thus argues that growth firms dependent on debt may pass up valuable investment opportunities. Consequently, growth firms prefer to hold larger cash ratios in order to avoid the increased costs of external finance and to be able to continue investing in valuable investment opportunities.
Family firms in the UK exhibit significantly higher growth opportunities than non-family firms (Alexandrou and Sudarsanam, 2001). This finding has been confirmed for France by Ginglinger and Saddour (2007) and for the whole of Europe by Croci, Doukas and Gonenc (2010), by examining respectively Tobins-Q and yearly sales growth. Therefore it is to be expected that family firms have higher cash ratios compared to non-family firms in order to finance their growth.
Company size
Cash flow and cash flow variability
Most studies examining the impact of family control on company performance find that they consistently outperform non-family firms (Anderson and Reeb, 2003b; Jirapon and DaDalt, 2009). It would then seem reasonable to assume that family firms generate relatively higher cash flows. Opler et al. (1999), Ozkan and Ozkan (2004) and Ferreira and Vilela (2004) all find that higher cash flows lead to higher cash ratios. Additionally, Almeida, Campello and Weisbach (2004) show that financially constrained firms have a positive cash flow sensitivity of cash, while this is not the case for financially unconstrained firms. According to these studies, this is the result of firms preferring internally generated funds in order to avoid having to access the expensive capital markets. However, according to Kim, Mauer and Sherman (1988) and Garcia-Teruel and Martinez-Solano (2008) cash flows serve as a substitute for cash holdings as they provide a direct source of liquidity. In their view, cash ratios within firms are expected to be lower as fewer external funds have to be raised and the risk of having to pass up valuable investment opportunities is reduced. Therefore, the final effect on cash ratios of the higher cash flows commonly observed in family firms remains unclear.
An important factor that has to be considered when discussing cash flows is its variability. It can be expected that firms with highly volatile cash flows will experience cash shortfalls in more states of nature. Minton and Schrand (1999) find that firms with higher cash flow volatility permanently forgo investment opportunities when these cannot be directly financed. Opler et al. (1999) show that firms tend to hold higher cash ratios if their industry average cash flow volatility is higher. Yet, family firms maintain superior relationships with their customers, vendors and investors compared to non-family firms (Petersen and Rajan, 1994) which may result in less variable cash flows. However, Ginglinger and Saddour (2007) show that family firms in France experience more cash flow volatility. The latter study is, to my knowledge, the only observation in the empirical literature regarding the cash flow volatility of family firms. In as far as this observation is representative for the European Union as a whole, I expect that family firms maintain higher cash ratios in order to avoid financial distress.
Hypotheses
firms. Additionally, the smaller size of family firms along with their significant growth opportunities leads them to be more information-opaque than their non-family counterparts. Moreover, the preference of family firms to maintain control over their company by using debt instead of equity to finance their investments leads them to have higher default risks. All these factors negatively influence the accessibility to external finance and increase the demand for internally-held cash. However, the presumably higher cash flows and easier access to debt financing in family firms could to some extent mitigate these higher costs of external finance. What the combined effect of these determinants is on corporate cash ratios remains unclear. The first hypothesis of this study is as follows:
H0. Family firms maintain similar cash ratios compared to non-family firms. H1. Family firms maintain different cash ratios compared to non-family firms.
In case the null hypothesis is rejected and family firms are found to maintain higher cash ratios, I assume this to be a consequence of their increased tendency to expropriate minority shareholder wealth. However, when cash ratios are found to be lower in family firms I assume this to be a consequence of either reduced managerial discretion or an increased tendency to protect minority shareholders compared to non-family firms. It is impossible to discriminate between the latter two effects, because no appropriate proxy can be obtained to measure each effect independently.
According to Bruinshoofd and Kool (2002; 2009), the static trade-off model can be employed to determine the long-term desired target cash ratio. At the same time, this model suggests that firms instantaneously readjust their annual cash ratios to their target cash ratios. However, it is highly likely that firms are affected by transaction and other adjustment costs, which cause a delay in the adjustment from the current cash ratio to the desired ratio. Indeed, Bruinshoofd and Kool (2002) in the Dutch market find adjustment speeds varying between 21% and 68%, depending on the target used, whereas Ozkan and Ozkan (2004) show that firms in the UK adjust to 60% of their target cash ratio yearly. Besides the other determinants of cash ratios previously described, the degree of agency costs is also likely to influence adjustment costs. When agency costs are high, the costs of external finance increase which are likely to make firms wary not to stray too far from their target cash ratio. Because family and non-family firms are expected to be influenced by a different degree of agency costs, the second hypothesis of this study is:
Again, it cannot be beforehand predicted if adjustment speeds differ between both firm types, and if they do, which firm type will adjust quicker than the other. It is my assumption, that if adjustment speeds in family firms are significantly higher, this to be a consequence of their increased tendency to expropriate minority shareholder wealth. If adjustment speeds are however lower in family firms, I assume this to be a consequence of either reduced managerial discretion or an increased tendency to protect minority shareholders compared to non-family firms. Again, it is impossible to discriminate between the latter two effects, because no appropriate proxy can be obtained to measure each effect independently.
3. Data
Sample selection
In order to test the proposed empirical analysis of corporate cash ratios, this study examines a panel of 7068 firm-year observations representing 1178 firms publicly listed in the fifteen European countries that made up the European Union as per January 1st 1995 (EU-15) over the 2003 – 2008 period.5 The EU is chosen due to the presence of relatively many and large family shareholders as shown by La Porta et al. (1999) and Faccio and Lang (2002), enhancing the testability of the hypotheses postulated in this study. The focus on these EU-15 countries and the exclusion of later EU-member states follows from the fact that the EU expanded on May 1st 2004, which lies within the data-obtainment years. Data are collected from 2001 onward, as this study employs one-year lagged variables to correct for possible endogeneity and because variables are included which are measured as the change over time. Although it would be preferable to include the most recent data, the final year of examination, 2008, is chosen because of the large amount of missing data for the year 2009.
All firms of the initial sample are obtained from the Amadeus Database of Bureau Van Dijk, which focuses specifically on European firms and contains extensive financial data and the most current ownership information. Previous versions of Amadeus provide ownership information for earlier years which enables me to test the standard assumption of ownership stability proposed by Opler et al. (1999). Regrettably, the versions differ in the scope of firms examined. Additionally, the search engine of Amadeus changed in 2003, providing consistent information acquisition only from this point onward. Given these limitations, I select all companies listed on the two largest available database-years6, 2005 and 2008. Subsequently, all financial firms, measured by two-digit NACE Rev. 1.1 codes (65 through 67) are excluded as their core business involves marketable securities. Finally, all companies with missing
5 The members of the EU-15 are Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, The
Netherlands, Portugal, Spain, Sweden and the United Kingdom (UK).
6 The following versions of Amadeus are used: May 2005: Release 135.0, version 36.0, 8.017.125 companies, May 2008: Release
shareholder information are dropped after which the remaining firms are then classified as either family or non-family firms.
In line with most empirical literature, this study defines a family firm as a firm in which an individual or family holds at least 20% of the total voting rights and is also the largest shareholder.7 Amadeus divides shareholder ownership into direct and total voting rights, where the latter is the summation of direct and indirect voting rights. Frequently only the direct voting rights are known. However, the literature on tunneling in pyramid and cross-holding structures demonstrates the importance of indirect voting rights. Additionally, because individuals or families often invest in their company through holding companies8, I select, where known, total voting rights instead of direct voting rights. Amadeus furthermore does not define an individual or a family separately and frequently divides families into their corresponding members. I therefore extract from Amadeus all firms in which one individual or family obeys the family firm definition and a second list of firms in which all named individuals and/or families together follow the family firm definition. The first list is regarded as the base sample, to which firms listed on the latter list are added, after they are manually checked to see if individuals and/or families have the same last name. Based on the last name, voting rights are summed and it is again verified if the firm adheres to the definition of a family firm employed in this study.9 Furthermore, there are firms for which Amadeus contains information that a family or individual holds over 25% of the total voting rights and is also the largest shareholder, but of which the actual percentage is unknown.10 I include these firms into the family firm sample. If a firm does not meet the definition of a family firm I classify it as a non-family firm. I only include in the examination of corporate cash holdings those non-family firms in which one individual or family holds less than 10% of the total voting rights. When a firm cannot be classified, it is discarded from the sample. Finally, firms which are not listed in Amadeus during either 2005 or 2008 are removed, because it cannot be determined if they consistently belong to either group.
7 Besides numerical values, Amadeus sporadically defines shareholders as Majority Owners (>50%), Wholly Owned (>98%) or
Negligible (<0,01%). I replace these words by the minimum voting right percentage indicated between brackets and when voting rights are not available (n.a.) they are considered to be zero.
8 In 2005, Peugeot was majority owned by the family Peugeot by 43,86% of the total voting rights. This firm would not have been
classified as a family firm, when having focused merely on direct voting rights.
9 Because family members do not necessarily have to have the same last name, it is possible that the calculated voting rights are
negatively biased. Still, I feel this is the best way to create an honest family firm sample, considering that Amadeus is not always very meticulous in classifying shareholders. Frequently shareholders with the same name are mentioned twice with a similar voting right percentage. This would create false positive-family firms. Example: year 2005: Birse Group PLC mentions Mr. P.M. Birse and P.M. Birse as shareholders both with 12.41% of the voting rights.
10 These companies are found through searching for individual or family ultimately owned companies in Amadeus and selecting
In contrast to La Porta et al. (1999), Faccio and Lang (2002) and Kalcheva and Lins (2007)11 I find that ownership concentration is susceptible to change as 565 companies, 18% of the remaining sample, are not consistently defined in 2005 and 2008 (see Appendix A for a detailed breakdown). I only retain those firms which consistently belong to either firm type in order to obtain a pure family and non-family sample. Still, I do assume that once firms are consistently controlled over 2005 through 2008, they are consistently controlled during the entire sample-period.
For the remaining firms, company level financial and accounting information is retrieved for all variables from the most current Amadeus-database. Companies with one or more missing observations are removed from the sample to create a balanced dataset. After applying these stringent criteria, companies from Austria, Denmark, Ireland and Luxembourg disappear from the final sample. Table 1 shows a numerical representation of the process described above. Of the remaining 1178 firms, 432 are family firms whereas 746 are non-family firms.
(Please insert Table 1 here)
Dependent variable
In this study I use the cash ratio as the dependent variable. Similarly to Ozkan and Ozkan (2004) the variable CASH is calculated by scaling the sum of cash and cash equivalents by total assets.12 To remove outliers, CASH is winsorized at the 1st and 99th percentile. Table 2 gives an overview of the cash ratio per country over 2003 – 2008. It should be noted that there is a large spread regarding the number of observations included per country. This is a direct effect of the number of publicly listed firms in a country in addition to the stringent inclusion criteria of this study. I observe a mean cash ratio of 11% in the EU, which is slightly lower compared to Ferreira and Vilela (2004) who find a relatively stable mean cash ratio of 15% in the European Monetary Union over 1987 through 2000. The higher mean value in their analysis is most probably due to normalizing cash and cash equivalents by total assets minus cash and cash equivalents rather than total assets. Furthermore, Table 2 shows considerable differences between cash ratios in the different countries. While France has the highest mean cash ratio of 0.15, Portugal has the lowest mean cash ratio of 0.03. These differences can be caused by varying accounting standards, development of capital markets or corporate governance patterns (Ferreira and Vilela, 2004).
11 These authors all assume that ownership concentration can be assumed to be stable over time. Thomsen and Pedersen (2000)
have tested this simplifying assumption for a subsample of 329 of the 446 largest European companies. They argue that ownership structure is indeed stable over the 1990-1995 period with a year by year correlation coefficient of 0.95 – 0.96. Yet, this high year by year correlation could lead to great changes in ownership over a longer period.
12 Opler, Pinkowitz, Stulz and Williamson (1999) use a slightly different definition, by adjusting the denominator by subtracting
Still, my findings are in line with previous literature (Guney, Ozkan and Ozkan, 2003; Ozkan and Ozkan, 2004; Ginglinger and Saddour, 2007).
(Please insert Table 2 here)
Independent variables
I introduce multiple independent variables into the regressions that can affect cash ratios. Most independent variables have been discussed previously in the literature review. For others, the empirical literature does not propose a specific relationship with either family or non-family firms, though they are important determinants of cash ratios. For a detailed overview of the variables and the calculations, see Appendix B.
The firm’s debt ratio, leverage (LEV), is obtained by scaling total debt by total assets. Following Ozkan and Ozkan (2004), I proxy growth opportunities (MTB) by the market-to-book ratio. A company’s
cash flow (CFLOW) is calculated by dividing pre-tax profits plus depreciation by total assets. In line with
Guney, Ozkan and Ozkan (2003) the cash flow variability (VAR) is calculated as the standard deviation of pre-tax profits plus depreciation over the current and two previous years divided by average total assets over this period. By taking a three year moving average instead of a constant average of the entire sample-period13 it is possible to better capture the variability of the cash flows over time. Company size (SIZE) is measured by the natural logarithm of total assets. I introduce the variable liquidity (LIQ) calculated as the ratio of working capital, less cash and cash equivalents, to total assets into the regressions. Dittmar, Mahrt-Smith and Servaes (2003) consider the presence of liquid assets other than cash and cash equivalents as a substitute for the latter. The firm’s age (AGE), defined as the natural logarithm of the difference between the year of analysis and the year of incorporation, has been shown to impact cash ratios. Croci, Doukas and Gonenc (2010) show that internal financing is more important for young firms as they are generally considered to be high growth firms with uncertain future cash flows and a non-established credit reputation. It would follow that younger firms generally maintain higher cash ratios than their older counterparts. Guney, Ozkan and Ozkan (2003) stress the importance of controlling for the degree of capital expenditures as a firm’s cash holding policy could simply be a function of its capital expenditures. I measure the variable investment (CAPEX) by scaling capital expenditures by total assets. Furthermore, I use the degree of capital market development (MCAP) as a control variable. A measure for capital market development is obtained by scaling the market capitalization of a given stock market by the country’s total GDP. This data is obtained from the World Bank. Following Ferreira and Vilela (2004) and
13 Kim, Mauer and Sherman (1998), Minton and Schrand (1999) and Ozkan and Ozkan (2004) calculate cash flow variability as
Chang and Noorbakhsh (2006) it is expected that firms in countries which are more financially developed have easier access to external financing. These firms will consequently be able to hold lower cash ratios. Additionally, I correct for the country’s legal system. The empirical literature has shown a strong relationship between the degree of shareholder protection and cash ratios. Besides company governance, a country’s legal system has shown to be an important proxy of shareholder protection. In Europe, two legal systems can generally be defined: common law and civil law. La Porta, Lopez-de-Silanes, Shleifer and Vishny (1997) show that common law countries generally offer shareholders more protection than civil law countries. When sufficiently protected, shareholders have more incentives to invest and consequently this lowers the costs of external financing and expands the capital market. Dittmar, Mahrt-Smith and Servaes (2003) show, in research spanning more than 11.000 firms in 45 countries, that firms in common law countries (here only firms from the UK) hold up to 35% less cash than civil law countries (all other countries). To control for the degree of shareholder protection, I include a dummy variable (COMMON) for the country’s legal system. This dummy variable takes the variable of 1 in common law countries and 0 in civil law countries. Finally, I add country and industry dummies in all regressions in order to control for country and industry differences. The extent of country differences regarding cash ratios has already been determined in Table 2.
In order to characterize the firms in the final sample, Table 3 reports the descriptive statistics of the independent variables mentioned above. To remove outliers, all variables are winsorized at their 1st and 99th percentile. It can be seen that the sample is made up of a relatively fair mix of young and old firms which have a considerable size. The average size of a company, measured by total assets, amounts to €230 million of which more than 50% is financed through debt. The average market-to-book ratio indicates that the sample firms have considerable growth opportunities with a market value on average twice their book value. Cash flows over the 2003 – 2008 period are on average approximately 6% of total assets with a year-to-year average variation of 5%. Compared to, for instance, the United States with a market capitalization of 1.34 times its total GDP, the capital market ratio of Western-Europe is only 0.86. Overall, the obtained sample has similar characteristics compared to other studies describing cash ratios in European countries (Ozkan and Ozkan, 2004; Ferreira and Vilela, 2004; Gingliner and Saddour, 2007).
(Please insert Table 3 here)
relationship with CASH, in contradiction to my assumption. The highest correlation of 0.41 exists between CFLOW and INV.
In Table 4, I present the descriptive statistics of all variables according to firm type. When examining this table it immediately becomes clear that family firms do indeed differ dramatically from non-family firms as all variables, except for mean capital expenditures and cash ratios, are significantly different at a 1% level. In accordance with Croci, Doukas and Gonenc (2010), family firms are significantly younger than non-family firms. Family firms are also considerably smaller. In line with Ozkan and Ozkan (2004) higher liquidity levels are observed in family firms. In conflict with Croci, Doukas and Gonenc (2010) but in conformity with Gingliner and Saddour (2007) I further find that family firms are less leveraged than non-family firms and that family firms have on average lower cash flows than their non-family counterparts. The latter finding may be intertwined with the lower growth opportunities and less risky cash flows present in family firms in this sample. The lower variability of these cash flows can possibly be explained by the importance of consumer and vendor relationships in family firms which could result in more steady cash flows (Ali, Chen and Radhakrishnan, 2007).
(Please insert Table 4 here)
4. Methodology
In this study I analyze whether family and non-family firms can be considered homogeneous entities when describing their individual cash ratios, as is currently the case in the empirical literature. Additionally, I examine whether the adjustment speeds to target cash ratios differ between family and non-family firms. These hypotheses are tested by estimating the models developed here using random effects panel regressions on the balanced panel dataset described in the previous section.
Having specified the estimation method, I present here the two versions of the static trade-off model used in this study. I first perform a regression on all independent variables in order to determine what the effect of these variables is on cash ratios. This general version of the static trade-off model also allows the obtainment of the average cash ratios of both family and non-family firms. Following Opler et al. (1999), the basic model is specified as follows:
(1) Yit = α0 + ∑βnXit + ηi + υit
given by ηi and υit, where ηi captures firm heterogeneity and υi represents white noise. I estimate this equation for the entire sample and for each firm type independently. By including a dummy variable resembling family firms (FAMILY) in the regression considering the entire sample, I can determine whether cash ratios between family and non-family firms differ significantly. It should be noted that the proposed relationship between cash ratios and the independent variables can, in theory, go both ways. I therefore correct for endogeneity by including only one-year lagged variables of the independent variables in the regression, except for AGE.
Having used the previous specification to obtain average cash ratios for both types of firms, I subsequently focus on the second hypothesis regarding the adjustment speed with which firms adjust their cash ratios to a certain target. The normal static trade-off model assumes that firms can instantaneously adjust their cash ratios back towards their target ratio. However, Bruinshoofd and Kool (2004), amongst others, have shown that firms adjust to approximately 21% - 68% of their target ratio in a given year, depending on the target ratio used. I employ the equation first developed by Lintner (1956), who used it to explain the speed with which firms adjust their dividend payout to their target payout ratio, to obtain the target-adjustment speed of both family and non-family firms. The partial-target adjustment model is specified as follows:
(2) ∆Yit = α0 + λ(Yit* - Yit-1) + ηi + υit
Where ∆Yit represents the change in CASH over one time period, α0 is a constant and λ indicates
the speed with which cash ratios in the previous period (Yit-1) adjust to their long run target (Yit*). The error terms are given by ηi and υit, where ηi captures firm heterogeneity and υi represents white noise. In order for cash ratios of firms to regress towards their target, λ has to be greater than 0, but lower than 1, implying positive adjustment costs. Unfortunately, the target cash ratio (Y*) is unobservable. Multiple possible targets have been introduced over the past decades. In accordance with Shyam-Sunder and Myers (1998), Opler et al. (1999) naturally consider the average historical cash ratio as the target ratio. Bruinshoofd and Kool (2004) following Opler et al. (1999) use three modified outcomes of equation (1) as the target ratio. Because of these possible different approaches, I will use two different estimates of target cash ratios. First, I model target cash ratios in a certain year as the average of the cash ratio’s in the three previous years. This model is specified as follows:
(2.1) ∆Yit = α0 + λ(Yit* - Yit-1) + ηi + υit
Secondly, I will employ equation (2) where the target cash ratio, from now on denoted the specific target ratio, is a function of the independent variables used in equation (1). This equation is as follows:
(2.2) ∆Yit = α0 + λ(Y* - Yit-1) + ηi + υit
Yit* = α1 + ∑βnXit
Equation 2.2 can be rewritten as:
(2.3) ∆Yit = α0+ λα1 + λβnXit - λYit-1 + ηi + υit
Where ∆Yit represents the change in CASH over one time period, α0 and α1 are constants and λ indicates the speed with which cash ratios in the previous period (Yit-1) adjust to their long run target ratios. λβn denotes the vector of the corresponding parameter estimates of each of the included independent variables (Xit). βn/λ in turn is a measure for the long-term effect of the variable on cash ratios. The error terms are given by ηi and υit, where ηi captures firm heterogeneity and υi represents white noise. Equations (2.1) and (2.3) are estimated for the entire sample and for both firm types independently. By including a dummy variable resembling family firms (FAMILY) in the regression considering the entire sample, I can determine whether long-run cash ratios between family and non-family firms differ significantly, when adjusting for their possible different adjustment speeds. Because the dependent variable ∆Yit contains Yit-1, a variable also encountered on the right-hand side of the equation, autocorrelation between with the error terms can occur. 14 To determine whether this autocorrelation will lead to biased standard errors, I employ the Durbin-Watson (DW) to test the null hypothesis of no autocorrelation.15 Furthermore, I also employ one-year lagged variables for the remaining independent variables in the regression, except for AGE, in order to correct for possible endogeneity.
5. Results
Univariate analysis
I first conduct a univariate analysis in order to determine if there are significant differences in the independent variables studied between firms in relation to their cash ratios. Table 5 displays the means of all variables according to the quartiles of CASH and firm type. I perform an F-test of the means of each
14 I prefer to correct for autocorrelation by using the one-year lag of Y
it-1, being Yit-2, on the right hand side of the equation.
However, the use of this lagged variable leads to non-interpretable results. Therefore this preferred solution is not chosen,
15 The 1% confidence interval for which the null hypothesis of no autocorrelation will not be rejected lies between the Durbin
variable to determine whether these means differ from each other over all quartiles and provide the p-value of this test.
In general, it can be observed that the characteristics of firms which hold more cash (Q4) are significantly different from those with less cash (Q1). In line with my expectations I find that firms which are older, larger, have less growth opportunities and have more liquidity retain lower cash ratios. Less leveraged firms together with firms experiencing higher cash flows and more variability of these flows are found to have higher cash ratios. In contrast to my expectations I find that firms in common law countries have higher cash ratios and that the degree of development of the capital market is also positively related to cash ratios. It should be noted that the effects are not monotonic for all variables and consistent in both firm types. For example, size and leverage are non-monotonic in family and non- family firms, whereas cash flow and cash flow variability show a non-linear pattern only in non-family firms. This suggests that merely comparing the highest and lowest quintiles of cash ratios cannot present a clear picture of the relation between cash ratios and firm characteristics or the difference between firm types. I therefore focus on the multivariate analysis to provide a clearer picture.
(Please insert Table 5 here)
Multivariate analysis – Cash ratios
Table 6 shows the results of equation (1) using either country dummies or the legal-dummy (COMMON). This regression of the static-trade-off model, under the assumption of instantly mean reverting cash ratios, is used to test the null hypothesis that cash ratios do not differ significantly between family and non-family firms. Additionally, this model is able to show what determinants influence cash ratios in both firm types. Because I am interested in the difference between family and non-family firms, I estimate equation (1) for the entire sample (ALL) and subsequently for family firms (FAM) and non-family firms (NON-FAM) separately. In all models time and industry dummies are included, except for one in order to avoid multicollinearity. For the sake of brevity, the coefficients on these dummy variables are not shown. I take the year 2004 as the base year, the transport sector as the reference industry and Sweden as the country of reference. Furthermore, all independent variables are lagged by one year in order to correct for possible endogeneity problems. I show the p-values of the coefficients in brackets and use asterisks to indicate whether coefficients differ between both firm types at a specified significance level.
incorporating the legal-dummy, family firms appear to have significantly lower cash ratios at a 10% significance level compared to non-family firms. This would imply that family firms, overall, face lower agency costs than their non-family counterparts. However, when replacing the legal-dummy by country-dummies, the significant difference in cash ratios between both firm types disappears. The inclusion of country-dummies provides a more detailed analysis of cash ratios than merely including the legal-dummy, as seen by the higher adjusted R2 of these former models. Therefore, I interpret the different outcomes as indicating that no difference between cash ratios and subsequently overall agency costs exist between both firm types. Either way, these findings contradict the results of Ozkan and Ozkan (2004) and Ginglinger and Saddour (2007), who find a positive effect of family firms on cash ratios. Because the dummy COMMON only includes the UK, the inconsistent findings for the FAMILY-dummy could hint towards the existence of a different attitude towards maintained cash ratios between countries. Consequently, examining the EU as a whole might not give representative answers for countries independently. I will examine this issue in the robustness checks performed later on.
When switching attention to the control variables, it can be clearly seen that besides the degree of cash flows and capital market development, there are no significant differences between the independent variables and how they influence cash ratios in family and non-family firms. This is a remarkable finding as the empirical literature would lead us to believe that there are ample differences between both firm types which could impact cash ratios. However, the lack of differences between both firm types support the finding that cash ratios do not differ significantly between family and non-family firms.
With regards to the effect of leverage (LEV), there is strong support for the negative relation between leverage and cash ratios (Kim, Mauer and Sherman, 1998; Opler et al., 1999; Ferreira and Vilela, 2004; Ozkan and Ozkan, 2004). I find at a 1% significance level, that more leveraged family and non-family firms hold lower cash ratios. As far as leverage is a proxy for the ease with which firms can access the debt-market, this negative relation is justifiable. Also, the negative coefficient of leverage may indicate that the costs of holding cash increase with the amount of debt financing.
It can further be observed that firm size (SIZE) negatively influences cash ratios. Again this effect is equally significant for both types of firms. According to the established literature on cash ratios, larger firms should theoretically, from a transaction motive perspective, be able to hold lower cash ratios due to economies of scale. Yet, Kim, Mauer and Sherman (1998), Guney, Ozkan and Ozkan (2003), Ozkan and Ozkan (2004) and Garcia-Teruel and Martinez-Solano (2007) have not been able to confirm this suspicion empirically. Size is also frequently used as a proxy for agency costs. Garcia-Teruel and Martinez-Solano (2007) argue that small firms are more likely to be influenced by asymmetric information also implying a negative relationship with cash ratios. Without being able to differentiate between either effect I confirm both.
The empirical literature is at odds when discussing the impact of cash flow (CFLOW) on cash ratios. According to Opler et al. (1999), Ozkan and Ozkan (2004) and Ferreira and Vilela (2004) higher cash flows should have a positive impact on cash ratios. In contrast Kim, Mauer and Sherman (1988) claim a substitution effect between cash and cash flows. In line with these latter authors I find that higher cash flows lead to lower cash ratios when examining the overall sample. This effect is significant at a 1% level. However, when differentiating between family and non-family firms, it can be seen that cash ratios of family firms are not influenced by their cash flows, whereas non-family firms hold significantly lower cash ratios when their cash flows increase.
Besides the flow of cash, its variability (VAR) is also crucial. I find this effect to be significant at a 1% level and the variable to have the largest coefficient of around 0.155. This would indicate that firms with highly volatile cash flows want to protect themselves from becoming financially constrained or having to give up valuable investment opportunities. These results hold when examining family and non-family firms separately.
The proposed substitution effect of liquid assets (LIQ) other than cash and cash equivalents with cash is also confirmed. This finding is in line with most other financial literature (Opler et al., 1999; Dittmar, Mahrt-Smith and Servaes, 2003; Ozkan and Ozkan, 2004; Garcia-Teruel and Martinez-Solano, 2008).
as five out of six regressions yield insignificant coefficients for AGE, the impact of a firms age on CASH is likely to be minor.
An interesting observation is that firms in common law countries have significantly higher cash ratios compared to civil law countries. It seems that this effect is limited to non-family firms at a 10% significance level. Overall, the empirical literature assumes that common law countries have better shareholder protection than their civil law counterparts. Translated to the arena of cash, this would normally imply that firms in common law countries are forced to retain lower cash ratios in order to not use these funds to obtain private benefits. Moreover, firms in common law countries generally have more easily access to the capital market. Yet, a possible explanation for the observed positive correlation resides in the fact that minority shareholders in common law countries allow firms to increase their cash ratios in order to fund attractive investment opportunities as they are less likely to be expropriated (Harford, Mansi and Maxwell, 2008).
(Please insert Table 6 here)
Multivariate analysis − Adjustment speed
Table 7 shows the results of regression equation (2.1) used to obtain the adjustment speed of family and non-family firms, where the preceding three year CASH-average is used as the current year target cash ratio. Table 8 shows the results of regression (2.3) in which I use equation (1) to specify the target cash ratio. I estimate equation (2.1) and (2.3) again on three samples: the entire sample and on both firm types independently. In all models time, industry and country or legal dummies are included, except one in order to avoid multicollinearity. For the sake of brevity the coefficients on these dummies, except COMMON, are not shown. I take the year 2004 as the base year, the transport sector as the industry reference and Sweden as the country of reference. I show the p-values of the coefficients in brackets and use asterisks to indicate whether the coefficients between both firm types differ at a specified significance level. Before describing both tables, it should be denoted that all regressions have a DW-statistic which does not allow me to reject the null hypothesis of no autocorrelation. Therefore I assume that all models contain unbiased standard errors.
difference in adjustment speeds between family and non-family firms to their historical target cash ratios. On the basis of Table 7 the null hypothesis that family and non-family firms do not have significantly different adjustment speeds cannot be rejected.
Table 8 shows the results of equation (2.3), where the specific target cash ratio is modeled. Considerably lower than the average adjustment speed of 48% observed when using an historical target, I find in this table that the average adjustment speed declines to 27%. Yet, this observation is in line with Bruinshoofd and Kool (2004) who use a similar definition of the target cash ratio and find an adjustment speed of approximately 23-26% in Dutch listed firms. Furthermore, in contrast to the results of Table 7, I find that family firms on average have lower adjustment speeds of 21% versus 31% compared to non-family firms. Considering the agency theory framework, this implies that non-family firms are subject to fewer agency costs and consequently are able to more easily obtain external financing, decreasing the necessity to quickly adjust to their target cash ratio compared to non-family firms. Overall, these findings reject the null hypothesis of similar adjustment speeds between firm types and the acceptance of the alternative in which family firms adjust their cash ratio more slowly towards their target ratio.
The lower adjustment speeds observed in family firms seem in agreement with the finding that family firms possibly hold lower cash ratios, previously observed in Table 6. However, this result cannot be confirmed in Table 8 as the average target cash ratios are not significantly different from zero in both firm types, as indicated by the dummy FAMILY. The latter result is further emphasized by the absence of differences in coefficients impacting target cash ratios in family and non-family firms. Although the impact of the independent coefficients on target cash ratios is initially obscured through the multiplication by the adjustment speed of the corresponding firm type, no difference between variables is found after correcting for adjustment speed.16 This finding is again in accordance with Table 6, where also no significant differences in independent variables between both firm types can be observed.
For the most part, the independent variables have the same sign as previously discussed with regards to Table 6. However, there are some differences when comparing Tables 6 and 8. In Table 8, the impact of cash flows on CASH is now significant at 10% for both firm types, whereas size is no longer of influence on target cash ratios in family firms. Additionally, the coefficient for LIQ has changed from a negative to a positive sign for both firm types.
(Please insert Table 7 here) (Please insert Table 8 here)
16 By performing a Wald-test on the coefficients scaled by the appropriate adjustment speed, the impact of only the variable itself
Robustness check
The previous tables have hinted that family firms maintain similar or even lower cash ratios than their non-family counterparts and that they adjust to their target cash ratios slower. These results are obtained when analyzing the European Union as a whole. Only firms in the UK, being a common law country, seem to hold higher cash ratios. In this section I again analyze the average cash ratios and adjustment speeds of family and non-family firms, but now per country to see whether the previous results hold. I only include France, Germany, Greece and the UK in this analysis as there are more than 800 available observations for each of these countries.
Table 9 shows the determinants of cash ratios per country in addition to their average cash ratios. In line with my previous findings, I find that family firms in France, Greece and the UK hold similar cash ratios compared to their non-family counterparts. In Germany, however, family firms have higher cash ratios at a 10% significance level, implying an increased tendency to expropriate minority investors compared to non-family firms. Additionally, it can be observed that different variables have a significantly different impact on cash ratios in family than in non-family firms in the four countries. In France, family firms which are older, less leveraged and maintain less liquidity have lower cash ratios than non-family firms. In Germany, only leverage and company size are more negatively correlated with cash ratios in family firms, whereas in Greece firm age, capital expenditures and size impact cash ratios differently compared to non-family firms. Finally, in the UK only leverage seems to impact cash ratios more than in non-family firms. Additionally, this is the only country in which family firms have a tendency to have higher cash ratios, instead of lower cash ratios, when their leverage increases.
the associated firm type adjustment speed, the Wald-test shows that only difference exists. In France family firms appear to hold higher cash ratios when they are confronted with similar growth opportunities compared to non-family firms.
(Please insert Table 9 here) (Please insert Table 10 here)
6. Discussion
The aim of this study is to examine whether target cash ratios and the speed with which cash ratios are adjusted to the firm’s target ratios differ between family and non-family firms. To this end, I meticulously construct a panel dataset of 1078 firms, consisting of 432 family firms and 746 non-family firms located in the European Union over 2003 – 2008. In contrast to most empirical cash literature (Opler et al., 1999; Ozkan and Ozkan, 2004; Ginglinger and Saddour, 2007), all included firms are checked for ownership consistency in more than one year.
First, I employ the basic static trade-off model and find that family firms do not maintain significantly different cash ratios compared to non-family firms. One specification of the model even suggests that family firms hold lower cash ratios. These results are in conflict with the findings of Ozkan and Ozkan (2004) and Ginglinger and Saddour (2007) whom both find that family firms maintain higher cash ratios in respectively the UK and France. Proof for the insignificant difference between cash ratios of both firm types resides in the fact that family and non-family firms are similarly affected by the multiple independent variables. Translated to the agency costs framework, the similarity in average cash ratios implies that family and non-family firms overall have similar agency costs. Whether agency costs affect both firm types equally or whether agency costs of managerial discretion are lower in family firms but offset by the increased agency costs of minority shareholder expropriation compared to non-family firms is not possible to determine. Furthermore, in line with the empirical literature, I find that younger and smaller firms and those with less non-cash working capital have higher cash ratios. Additionally, I find that more leverage and higher and less volatile cash flows result in lower cash ratios. In contrast to Ferreira and Vilela (2004) the degree of capital market development has no effect on cash ratios. These before mentioned effects are similar for family and non-family firms. Moreover, I find that when examining the sample as a whole, firms have significantly higher cash ratios in common law countries. This effect seems to be limited to non-family firms.
is in line with Ozkan and Ozkan (2004) and Bruinshoofd and Kool (2004) who find that there are significant adjustment en transaction costs involved with adjusting cash ratios. I find that family firms adjust their cash ratios more slowly towards their target ratio compared to non-family firms. Because no single determinant influences one firm type more or less than the other, the effect is assumed to be caused by different degrees of agency costs present in both firms. This implies that family firms are subject to fewer agency costs and consequently are able to more easily obtain external financing, decreasing the necessity to quickly adjust to their target cash ratio compared to non-family firms.
Finally, I test whether the results obtained when examining the EU-15 are still valid when analyzing countries separately. Considering the basic static trade-off model, I find that only family firms in Germany have a significantly higher cash ratio compared to non-family firms. This effect seems to disappear when the partial adjustment model is examined. Also, the slower adjustment speed observed in family firms when examining the entire European Union cannot be replicated on individual country level.
Limitations
During the process of this study, I encountered several limitations which could ultimately lead to biased results. First, the selection of family and non-family firms is based on ownership percentages obtained from the Amadeus database. Because family shareholders are frequently split up into their corresponding members in this database, shareholders which do not have the same last name, but belong to the same family, are disregarded. Furthermore, Amadeus does not always list all shareholders or has the most recent ownership data available. Taken together, these factors could all lead to a misclassification of firm types and ultimately bias the obtained results.
Second, in this study only firms listed on a stock exchange are included. As these firms all have to obey to stringent rules and regulations, the pre-listing differences between family and non-family firms are likely to decrease. Furthermore, when family firms become listed, they frequently attract a more professional management team, adding to their resemblance of a non-family firm. Ultimately, these two factors could explain the lack of differences observed between both firm types.
