Performance of Swedish listed family-
firms
Paper within Finance
Author: Andreas Rasku
Tutor: Agostino Manduchi
Master thesis in International financial analysis (30 hp)
Title: Performance of Swedish listed family firmsAuthor: Andreas Rasku
Tutor: Agostino Manduchi
Date: 2014-05-11
Subject terms: Family-controlled, ROA, Return on assets, Tobin’s Q, founding-family, CEO, Swedish, listed, performance, firm-specific
knowledge
Abstract
This thesis investigates the performance of Swedish listed family-controlled firms using re-turn on assets (ROA) and Tobin’s Q as performance measures. Results show that found-ing-family firms perform 6.1 % better than other firms for ROA. Firm-specific knowledge of the founder-CEO is the main cause of the enhanced performance. The results are not robust to residual testing which suggests caution when drawing conclusions from these re-sults. The main contribution of this thesis is an empirical analysis of family insider repre-sentation and the relation to ROA and Tobin’s Q in a sample consisting entirely of Swedish firms.
Table of Contents
1
Introduction ... 1
1.1 Problem discussion ... 1 1.2 Research question ... 1 1.3 Purpose ... 12
Theoretical framework ... 2
2.1 Dual-class share system ... 2
2.2 Agency theory ... 2
2.3 Swedish code of corporate governance ... 3
3
Institutional framework ... 4
3.1 Ownership-types and their features ... 4
3.1.1 Family-controlled firms ... 4
3.1.2 Other ownership-types ... 4
4
Literature review ... 6
5
Method ... 9
5.1 Data and sample ... 9
5.2 Performance measure ... 9 5.3 Descriptive statistics ... 10 5.4 Research design ... 12
6
Results ... 15
6.1 Regression results ... 15 6.1.1 Regression: Founding-families ... 156.1.2 Regression: Other ownership types ... 17
6.1.3 Regression: Family characteristics ... 18
6.2 Residual tests ... 20
7
Discussion ... 23
7.1 Main results ... 23 7.2 Further results ... 24 7.3 Endogeneity ... 24 7.4 Social implications ... 258
Conclusion ... 26
8.1 Future research ... 26List of References ... 27
Tables
Table 4-1 Summary of related empirical work ... 7
Table 5-1 Industry classification from Nasdaq OMX ... 9
Table 5-2 Ownership information and insider representation ... 11
Table 5-3 Data characteristics continuous variables ... 11
Table 5-4 Correlation matrix ... 12
Table 5-5 List of variables included in the analysis ... 14
Table 6-1 Founder vs non-founder ... 16
Table 6-2 Founder vs non-founder with effects of ownership and control ... 16
Table 6-3 Ownership-types ... 18
Table 6-4 ROA and family characteristics ... 19
Table 6-5 Tobin’s Q and family characteristics ... 20
Figures
Figure 5.1 Return on assets (ROA) ... 11Figure 5.2 Tobin's Q ... 12
Figure 5.3 Tobin's Q and size ... 13
Figure 5.4 ROA and size ... 13
Figure 6.1 ROA robustness check plot ... 21
1 Introduction
1.1 Problem discussion
Firms controlled by families are a common feature around the world. Villalonga and Amit (2006) report that 37 % of the firms in Fortune 500, which is a list of US firms ranked by revenue, are controlled by families. Similar numbers have been reported by Faccio and Lang (2002) for western European firms where 44 % of the firms were family-controlled. For East Asia numbers reported range between 50 – 70 % (Claessens, Djankov, Fan & Lang, 2002). In Sweden, 65.4 % are family-controlled which is derived from the empirical sample of this thesis. Family-control is in previous work generally defined as a family being the largest shareholder and in control of 10-25 % or more of the votes.
When one single form of ownership controls such a vast number of firms that itself moti-vates further investigation. Small differences in performance between families and other ownership-types would have huge impact on the economy as a whole due to the large number of firms controlled by families. If differences in performance are observed, posi-tive or negaposi-tive, laws and rules can be designed to expand on what families are doing right and to redirect what they are doing wrong, benefiting society as a whole.
Families exhibit some salient characteristics with respect to their ownership. From the sample of this thesis it is evident that families have a larger ownership stake on average than other owners. They also have insider representation as CEO or on the board of direc-tors to a larger extent. In addition to this families are more likely to have founded the firm. The founder has more knowledge in general about the firm which is expected to enhance firm performance. Also, the founder may be too psychologically attached to the firm and resist favorable changes (Pierce, Kostova & Dirks, 2001), which reduce firm performance. Altogether there are many forces at work regarding families that are different from other owners which make families and firm performance very interesting to investigate further. This thesis focuses on Sweden and the main contribution is an empirical analysis of family insider representation and the relation to performance which previous studies with samples consisting entirely of Swedish firms have neglected.
1.2 Research question
The research questions investigated in this paper are:
1. How do Swedish listed family-controlled firms perform compared to other firms? 2. What is causing the results in the previous question?
1.3 Purpose
The purpose of this thesis is to evaluate Swedish listed family-controlled firms regarding performance and to investigate what the causes of this performance are. The results of this thesis can be used by anyone interested in how family-controlled firms perform and are particularly directed to legislators and decision-makers in corporate governance1. The
pur-pose of this thesis is also to discuss ethical implications of the results.
1 Corporate governance refers to laws and regulations about how the firm is ”governed” on one side and to
2 Theoretical framework
2.1 Dual-class share system
Dual-class shares are important for this thesis as family-controlled firms are the most fre-quent users. Dual-class shares means that different shares have different voting power in respect to cash flow rights. Dual-class shares are regulated by the law Aktiebolagslagen (ABL 2005:551) and in ABL 4:2 it is stated that it is possible to deviate from one share-on vote principle. If the firm uses shares with different voting rights they have to disclose it in their article of association. The information disclosed in the article of association should be which letter that represents each share and what voting power each share has. In Sweden it is most common to use two share classes and name them A and B where one of the shares (A mostly) has 10:1 voting power over cash flow rights. More than two share classes with each having different voting structures are possible as well. The maximum deviation of vot-ing power over cash flow rights are 10:1, which is stated in ABL 4:5.
One practical implication of dual-class shares is that the percentage of cash flow rights and voting rights possessed by the family-firm need not to be equal. For example, the Wallen-berg family’s ownership in Atlas Copco is 22 % of the votes and 17 % of the cash flow rights as of 2012 (Atlas Copco 2013).
2.2 Agency theory
Agency theory provides possible explanations to the behavior of owners and managers of the firm. Agency theory is important as it might help interpret the results of this thesis. The core of agency theory can be described by the following quote:
“We define an agency relationship as a contract under which one or more persons (the principal(s)) engage another person (the agent) to perform some service on their behalf which involves delegating some decision making authority to the agent. If both parties to the relationship are utility maximizers there is good reason to believe that the agent will not always act in the best interest of the principal.”
(Jensen & Meckling 1976, p. 308) The agency relationship highlighted in this thesis is between controlling families and minor-ity shareholders. Minorminor-ity shareholders are concerned about whether families are conduct-ing activities related to moral hazard. Moral hazard is behaviors that hurt minority share-holders. Moral hazard occurs when the agent does not bear the full consequence of his ac-tions and changes his behavior because of this (Demarzo & Berk, 2011). Incentives to conduct moral hazard activities occurs because controlling families face a trade-off between using vote control to enjoy the full value of private benefits while only losing a fraction of the share value (Cronqvist and Nilsson, 2003).
Large family possessions of cash flow rights reduce incentives to conduct moral hazard ac-tivities for two main reasons. First the trade-off from consuming private benefits is smaller as the family bears most of the loss themselves. Secondly, large cash flow rights gives in-centives to monitor because the time and effort spent on monitoring benefits large share-holders the most. Monitoring means that some form of surveillance is conducted by the large shareholder. Tirole (2006) distinguishes between active and passive monitoring. Ac-tive monitoring means interacting and intervening with managers to increase firm perfor-mance. Active monitoring is prospective in terms of information. This means that large
shareholders gathers information about an event and exercises its control to influence managers to increase firm performance. Passive monitoring is for example the typical stock market analysts that analyzes past performance. Passive monitoring does not include any intervention with management and passive monitors sell their shares if unhappy with man-agement.
Large cash flow rights also make it easier to exercise control over the firm and conduct moral hazard activities. Although incentives to monitor increase with cash flow rights it is still easier to conduct moral hazard activities when in control (because of large cash flow rights), which is a trade-off in itself.
Incentives to conduct moral hazard activities increase when there is a wedge between vot-ing rights and cash flow rights. The trade-off from private benefits extraction described above is larger and less cash flow rights decrease the monitoring incentives. However, these are just incentives. Families might be superior monitories or have other competencies supe-rior to other shareholders leading to a better performance despite the larger wedge.
2.3 Swedish code of corporate governance
Swedish Code of Corporate Governance, hereafter called the code, is a set of regulations directed to Swedish listed firms. The code is important for this thesis as the results and conclusions might suggest additional regulation. This quote explains what the purpose of the Code is and the general function:
“The aim of the Swedish Corporate Governance Code is to improve confidence in Swedish listed companies by promoting positive development of corporate governance in these companies. The Code acts as a complement to legislation and other regulations by specifying a norm for good corporate governance at a higher level of ambition than the statutory regulation. However, this norm is not mandatory.”
(Corporategovernanceboard 2010, p. 3) The target group is firms on regulated markets in Sweden (Corporategovernanceboard 2010, p. 3). All firms in this thesis are from regulated markets and the Code naturally ap-plies to them.
3 Institutional framework
3.1 Ownership-types and their features
This section is introduced to give an institutional approach to why firms with different ownership-types may differ in performance. Highlighting these differences is important be-cause it will help explaining why the results in this thesis regarding families may differ. The four most frequent occurring ownership-types are discussed in this section. These are fami-lies, funds, dispersed ownership and the government.
3.1.1 Family-controlled firms
One positive feature of family-controlled firms proposed by Cronqvist and Nilsson (2003, p. 700) is that a founding-family as the largest owner may contribute with vital firm specific know-how. The entrepreneur who has been involved from the beginning has more knowledge about the firm than the typical fund manager or government representative. Another positive aspect of founding-family ownership is that the firm may carry the family name or that the family is deeply associated with the firm. The latter is the case of the Wal-lenberg family. They have generally not created the firm and instead have been the largest owners a long time and are heavily associated with the firm. Bad firm performance will than affect the reputation of the families and thus incentives are given to work hard and in-crease performance.
One negative aspect of family ownership is that the founding-families may be too deeply connected to the business and resist changes (Pierce, Kostova & Dirks, 2001). It may be ef-ficient in some cases to diverge, change or close the business and the deeply connected founder that is resisting such actions could lower firm performance.
Nepotism is a possible force at work regarding family-controlled firms. Nepotism means that families hire family members in favor of non-family people. As discussed by Dyer (2006, p. 261-262) nepotism leads to ineffective monitoring and thus lowering firm per-formance. This strengthens the potential agency problem between families and minority shareholders.
3.1.2 Other ownership-types
Funds are created to generate income for their customers. This is the primary motive for funds. However, there are other forces at work here that might affect the firm’s perfor-mance. Funds have superior knowledge about finance in general which could prove useful for the firms they are invested in. For example funds can give advice on loan terms, equity issue terms, general market conditions and many other aspects of the financial environ-ment. When the fund is invested in a firm there exist incentives to give well researched fi-nancial advice. Well researched advices increase the performance of the firm.
However, there are behaviors of funds that have negative effects on firm performance as well. David and Kochhar (1996, p. 460-463) address this by suggesting “barriers to effec-tive governance”. Funds are generally invested in many firms and it is hard to process in-formation effectively about all firms. This indicates that some firms receive less monitoring which decrease firm performance.
The government as largest owner is a special case in the sense that it is chosen by the peo-ple and is likely to be used as a tool to achieve political motives. The political motives are
not always aligned with profit-maximizing. For example, political motives could be envi-ronmental or national security.
Dispersed ownership is often defined in the literature on ownership and performance as no single shareholder holding 10 % or more of total shares. This percentage can be adjusted back and forth and the point is that dispersed owners are considered too small to have in-centives to monitor managers. No monitoring means agency problems which reduce firm performance. As dispersed owners would not intervene with managers the only significant effect at work here is the agency problem between managers and shareholders.
4 Literature review
Family firm performance has been studied empirically around the world. Table 4.1 provides a summary of important empirical work from United States, East Asia, Western Europe and Sweden. The results in table 4.1 generally seem to support the notion that family firms outperform other firms.
An important concept that emerges from previous empirical work is the need to separate ownership from control. This is due to the fact that large cash flow ownership increases in-centives to monitor and simultaneously having a wedge between voting rights and cash flow rights increase the incentives to conduct moral hazard activities and extract private benefits. The problem that occurs in this situation is to find a way to include both effects and avoid correlation. This is discussed in more detail in section 5.4.
Previous empirical work, for example Villalonga and Amit (2006), has highlighted the im-portance of including family characteristics in the analysis. One typical characteristics ac-counted for are family insiders, meaning family representation on the board of directors, as chairman of the board or as CEO of the firm. Insider representation is accounted for in this thesis as well. Another characteristic of family firms are whether the family has found-ed the firm or not. Families that are the original creator of the firm or have been large owners for a long time have had the possibility to shape the firm and should be separated from families investing short term. The founding family issue is accounted for in this thesis as well. One further characteristic of family firms noticed by previous empirical work is the extensive use of dual-class shares among families. Families use dual-class shares to a large extent in this thesis as well. This aspect of family ownership is also accounted for in the analysis of this thesis. Isakov and Weisskopf (2014) include which generation of the family that is currently controlling the firm and find some differences in performance with respect to this. However, in this thesis founding-families are defined as being either the creator of the firm or have been the largest owner in 10 consecutive years backwards from 2011, which is discussed in more detail in section 6.1.1. With this definition the term descendants do not have the same meaning as in Isakov and Weisskopf (2014). Descendants are there-fore not investigated in this study.
Previous empirical work also differentiates in the choice of performance measure. Tobin’s Q and return on assets (ROA) are most commonly used and Book-to-Market (B/M) ratio and Marginal q also occur. Tobin’s Q is a measure that relates market value of equity and debt to book value of equity and debt. ROA measures accounting performance where profits are related to total assets. Book-to-market ratio is similar to Tobin’s Q which relates only market value of equity to book value of equity. Marginal q relates the change in To-bin’s Q to the firm’s investments. For a full discussion of performance measures see sec-tion 5.2.
There has been some research on Sweden and an early example is Cronqvist and Nilsson (2003) who study how ownership-types differ in performance. The ownership-types under study are financial institutions, families and other firms. Their choice of performance measures are Tobin’s Q and ROA. The first issue with this is study is the inclusion of ex-cess voting rights and voting rights in the same regression. These measures are generally correlated and it is better to include excess voting rights and cash flow rights instead. This is what subsequent work has done. For example, Isakov and Weisskopf (2014), Villalonga and Amit (2006) and Barontini and Caprio (2006). The second issue is that some family characteristics are excluded from the analysis, for example family CEOs. Previous empirical
works mentioned above have showed that insider representation impacts performance. This thesis contributes further on the aspects discussed in this paragraph.
Another more recent study on Sweden is Bjuggren and Palmberg (2010) who study family firm performance with focus on excess voting rights. Performance measure in this study is marginal q. Their study neglects aspects of insider representation that have proved to be a common family feature, as discussed above. However, their study will at least serve as a benchmark to this thesis regarding family firms in general and the effect of excess voting rights
Table 4-1 Summary of related empirical work
Study Description of study and main results
(Isakov & Weisskopf, 2014) 185 firms in Switzerland between 2003 and 2010 are investigated. Dependent variables are ROA and To-bin’s Q. Family controlled firms perform better than other firms for ROA, not for Tobin’s Q. Performance is better when family members are active in the firm for ROA, not for Tobin’s Q.
(Cronqvist & Nilsson, 2003) 309 Swedish firms between 1991 and 1997. Dependent variables are ROA and Tobin’s Q. Family firms per-form worse than other firms for both ROA and To-bin’s Q.
(Bjuggren & Palmberg, 2010) 110 Swedish firms between 1999 and 2005. Dependent variable is marginal q. Family firms perform better than other firms. Excess voting rights have negative effect on performance.
(Reeb & Andersson, 2003) 403 firms in the United States between 1992 and 1999. Dependent variable is Tobin’s Q and ROA. Families firms perform better than other firms for both ROA and Tobin’s Q.
(Villalonga & Amit, 2006) 508 firms in the United States between 1994 and 2000. Dependent variable is Tobin’s Q. Families outperform only when a family member serves as CEO or as chairman of the board.
(Barontini & Caprio, 2006) 675 firms. Cross national European study. The de-pendent variables are Tobin’s Q and ROA. Family firms outperform other firms for both ROA and To-bins’s Q.
(Chu, 2011) 786 firms in Taiwan between 2002 and 2007. Families firm perform better than other firms. Results are stronger when family members serve as CEO or on the board of directors.
(Claessens et al, 2002) 1301 firms year 1996. Cross national study in East Asia. Dependent variable is Tobin’s Q. Negative rela-tion between control rights and Tobin’s Q.
5 Method
5.1 Data and sample
There are a total of 253 firms listed on OMX Stockholm Small, Mid and Large Cap. These firms where identified using the Nasdaq OMX Nordic website. The industry classifications provided in table 5.1 were also collected from this website.
Ownership data, vote percentage, cash flow percentage, board representation and CEO representation was collected from annual reports found on each firm’s website. Sometimes the largest owner’s identities were not disclosed in the annual report and the search had to be extended to specific owner’s websites, newspaper articles and press releases. Balance sheet and income statement information was collected from business retriever. Market val-ue and beta valval-ues were collected from Amadeus database.
14 investment firms invested in other firms in the sample were deleted to avoid double-counting. It is common practice in empirical work to exclude banks in the sample when in-vestigating performance. The reason for exclusion is that banks can lend more money than they have thus increasing leverage that does not have the same meaning in non-financial firms (Fama & French 1992, p. 429). 8 banks are excluded from the sample. Another 17 firms of foreign origin with headquarters abroad were deleted due to differences in regula-tory environments. Four firms listed as of spring 2014 had their IPOs after 2011 and was therefore not listed, was removed from the sample. It was not possible to retrieve infor-mation about who the largest owner was in ten cases and these were removed as well. Fi-nally 16 firms deviating from calendar year as financial year were excluded. The final sam-ple size is 184 firms.
Table 5-1 Industry classification from Nasdaq OMX
Number of firms Industry
5 Energy (Oil, gas etc.)
11 Materials (Mining, lumber etc.)
62 Industrials
16 Consumer goods
14 Consumer services
24 Health care
4 Telecom
20 Financials (Mainly real estate firms )
29 Technology
5.2 Performance measure
ROA and Tobin’s Q are the most frequently used performance measure in previous empir-ical work. The advantages of using return on assets (ROA) as a performance measure is that it captures pure operating performance. As nothing about market value of equity is in-cluded in this measure it becomes a performance measure of what the firm does operation-ally during the year. An example is if the firm is involved in a lawsuit that is expected to be finished 3 years from now. Investors and traders might speculate about the outcome now already which is not really relevant for the performance this year. Another example is that market value of equity based measures includes information about future profits of the
firm while ROA measures what the profits actually have been. This feature has been called “Backward looking” by Isakov and Weisskopf (2014, p. 6). The weakness of ROA is that market value might contain important information that ROA would miss.
Equation for return on assets:
( )
Tobin’s Q relates market value of equity to total assets. The advantage of Tobin’s Q is that information about market value of equity is included. This basically means that future events are valued by investors now which show up in the market value of equity. This fea-ture has been described as fufea-ture-looking by Isakov and Weisskopf (2014, p. 6). The disad-vantage of Tobin’s Q is that the firm might not have any revenue now (typical health care firm in the research phase) which lowers ROA while having promising future outlooks which raises Tobin’s Q.
Equation for Tobin’s Q:
Marginal q is a measure that relates the change in market value of a firm to the investments made during the same period. The advantage of marginal q is that the results are interpret-ed as a direct measure of managerial discretion (Bjuggren, Wiberg & Eklund, 2007) as in-vestments are mainly management activities. The disadvantage of marginal q is that the change in market value is likely to be determined by other factors than investments (Ber-glund, 2011). This is due to the fact that “most investments are markedly different from the assets that the firm has accumulated over time” (Berglund 2011, p. 1491). Marginal q then cannot separate which part of the change in market value that is due to new investments and which part that is due to change in the value of existing assets (Berglund 2011, p. 1490). Hence, marginal q is not used in this thesis.
5.3 Descriptive statistics
In this thesis’ sample 121 of the 184 firms had families represented as the largest owner with 10 % or more of the votes, which equals 65.4 %. Of the 121 family-owned firms 60 have either founded the firm or have had the largest ownership stake consistently 10 years back from 2011. 61 of the 121 firms have non-founder families or individuals owning more than 10 %. How the rest of the firms are allocated across the sample is seen in table 5.2. From table 5.2 it is also seen that founding families have the largest vote and cash flow stakes on average. They also have family members or hires connected to the family as in-siders to a larger extent than other owners.
47 of the 60 founding families apply dual class shares which equals 78.3 %. 26 out of the 61 non-founding family firms apply dual-class shares which equals 42.6 %.
Statistics of continuous variables in the sample are presented in table 5.3. A plot of ROA and Tobin’s Q are provided in figure 5.1 and 5.2 respectively. Table 5.3, figure 5.1 and 5.2 are introduced to give the reader an idea how the data behave. The correlation matrix is provided in table 5.4 to show how the variables are related. Generally, independent
varia-bles with high correlation should not be included simultaneously. The high correlations that appear in table 5.2 are discussed further in section 5.4
Table 5-2 Ownership information and insider representation
Ownership type Number
of firms CEO rep. Board rep. Chairman of the board rep. Both board and CEO rep. Votes% /cash% avg. Founding family 60 13 31 35 13 44.4/28.2 Non-founding family 61 7 27 24 5 28.8/22.9 Government 5 0 3 0 0 24.4/21.4 Funds 12 0 0 0 0 12.6/12.5 Dispersed 15 - - - - - Other 31 - - - - - Total 184 20 61 59 18 -
Table 5-3 Data characteristics continuous variables
Variable Mean Std dev Max Min Observations
Cash flow rights 0.230 0.145 0.754 0.018 184 Excess voting rights 0.168 0.132 0.486 -0.104 90 Control variable risk 0.49 0.21 0.99 0.035 184 Control variable size 7.56 2.02 12.77 3.06 184 Control variable beta 0.59 0.31 1.46 0.03 184
ROA 0.017 0.21 0.83 -1.34 184
Tobin’s Q 1.78 1.46 9.8 0.57 184
Figure 5.1 Return on assets (ROA) -1,5 -1 -0,5 0 0,5 1 0 50 100 150 200 ROA Firm
Return on assets
Figure 5.2 Tobin's Q Table 5-4 Correlation matrix
Tob. Q ROA Exc. v. r. Cash f. r. Voting r. Tobin's Q 1 ROA 0.123 1 Excess v. r. 0.096 0.200 1 Cash f. r. -0.149 -0.013 0.079 1 Voting r. -0.049 0.117 0.686 0.779 1
5.4 Research design
The analysis is divided into a set of regressions and each regression has its own section be-low where details are explained. All regressions were conducted in Excel or Eviews and the method used was Ordinary least squares (OLS) regression. The summary of all variables are found in table 5.5.
Control variables and industry dummies are included in every regression. Control variables used are natural logarithm of total assets (size), solidity and beta. Size is included as ROA and Tobin’s Q are expected to be affected by size. Natural logarithm is used on size as the relationship between size and performance is not expected to be linear as seen in figure 5.3 and 5.4. Tobin’s Q in figure 5.3 might be approximated better with 1/x, which was tried and the linearity induced by ln x and 1/x are basically the same. Besides size, firm perfor-mance is also related to risk. Therefore two different risk measures are introduced. Solidity captures the increase in risk that follows from increased leverage. Beta captures the firm’s risk in relation to the stock market. Industry dummies are introduced to capture differences in characteristics of industries. For example, the IT-sector might exhibit higher perfor-mance on average than the health care sector because of the nature of the industry, margins on products etc. The opportunities for moral hazard activities and extraction of private benefits may also differ between industries. An example could be media-industries (TV-channels, radio, newspapers etc.) where the information flow can be controlled by an own-er. This is likely to open up for more possibilities to conduct moral hazard activities than for example a mining firm. This is because mining is concentrated to single areas and few products whereas media reach many people and are much more dynamic in nature.
0 2 4 6 8 10 12 0 50 100 150 200 Tobin's Q Firm
Tobin's Q
Figure 5.3 Tobin's Q and size
Figure 5.4 ROA and size
The regressions of founding families and other ownership types (section 6.1.1 and 6.1.2) are also alternated between including ownership variables and including only family dummy variables. Dummy variables are variables that take on the value 1 or 0 and can be used when a category needs to be separated in the analysis. For example 1 if the is a founding-family firm and 0 otherwise. Conducting a regression without ownership variables means that the dummy variable is an average measure of all effects. Including ownership and trol makes the family dummy variable a measure of families “net” of ownership and con-trol. In the regressions on family characteristics the focus is on these characteristics and therefore only one type of regression where ownership variables are included is conducted. The ownership stake variable captures the incentives to monitor and excess voting rights captures the incentives to moral hazard activities and extraction of private benefits. Two al-ternatives to represent ownership and control are voting rights and excess voting or cash flow rights and voting rights. These combinations are highly correlated and thus have a high risk of multicolinearity. Voting rights and excess voting rights have a correlation-coefficient of 0.685 and voting rights and cash flow rights 0.779. Excess voting rights and cash flow rights have a coefficient of 0.079. The correlation matrix is presented in table 5.4. The general form of the regression is:
( ) ( ) ( ) ( ) ( ) 0 2 4 6 8 10 12 0 100000 200000 300000 400000 Tobin's Q Assests MSEK
Tobin's Q and size
-1,5 -1 -0,5 0 0,5 1 0 100000 200000 300000 400000 ROA Assets MSEK
Table 5-5 List of variables included in the analysis
Variable Description
Cash flow rights The amount of cash flow rights possessed by the largest share-holder in percentage points.
Excess voting rights Excess voting rights calculated as voting rights minus cash flow rights
Family Dummy variable equal to one if the family is in possession of 10 % or more of total votes (Founder and non-founder)
Founding family Dummy variable equal to one if the family is in possession of 10 % or more of the votes and also has founded the firm Non-founding family Dummy variable equal to one if the family is in possession of
10 % or more of the votes and has not founded the firm Board representation Dummy variable equal to one if the founding family is
repre-sented on the board of directors (except Chairman of the board)
Chairman of the board Dummy variable equal to one if the founding family is repre-sented as chairman of the board.
Family management Dummy variable equal to one if the founding family is repre-sented as CEO.
Strong control Strong family control measured as both family CEO and board representation in founding family firms
IND 8 industry dummies.
Control variable risk Control variable firm risk. Measured as the firm’s solidity. Control variable size Control variable firm size. Measured as the natural logarithm of
total assets.
Control variable beta Control variable market risk. Measures as the firm’s beta
Fund Dummy equal to one if the firm has a fund as largest owner owning more than 10 %
Government Dummy equal to one if the firm has the government as largest owner owning more than 10 %
Dispersed Dummy equal to one if the firm has no single shareholder own-ing more than 10 %
ROA (Dependent) Measured as net profit divided by average total assets.
Tobin’s Q (Dependent) Measured as market value of equity plus book value of debt di-vided by total assets
6 Results
6.1 Regression results
6.1.1 Regression: Founding-families
The regression is constructed in order to investigate if there are differences between found-ing families and non-foundfound-ing families. A foundfound-ing-family is defined as befound-ing either the creator or has had the largest ownership stake last 10 years as from 2011. Being the largest shareholder for a long period means that the family has had the possibility to shape the firm significantly and have psychological attachments to the firm comparable to the found-er, which was discussed in the institutional framework. The same psychological attachment is not expected in non-founding families.
To compare founder and non-founder families the variables family, founding family and non-founding family are introduced. Family is a dummy variable equal to one when the firm has a family as an owner in possession of 10 % or more of the votes (Both founder and non-founder families). Founder family has the same structure except that the family has also founded the firm. Non-founding family is equal to one if the family has not founded the firm. These three variables are tested separately in (1),(2) and (3) for ROA in Table 6.1 and in (4),(5) and (6) for Tobin’s Q. Table 6.2 (7)-(12) show the same variables and include cash flow rights and excess voting rights.
In table 6.1 (2) and (5) it is shown that founding families perform better than non-founding families as evident by the significant coefficients of founding families at the 5 % level for both ROA and Tobin’s Q. Non-founding family is not significant in any regression. The coefficient for ROA in (2) tells us that founding family firms have 7.5 % higher return on assets on average. The coefficient for Tobin’s Q is harder to concretely interpret. The founding family effect is so strong for ROA that it results in a significant coefficient for the variable family, where both founding families and non-founding families are included. This means that founding families outperform other ownership-types on average while non-founding families do not.
Table 6.2 (8) and (11) shows the coefficients of founding-families with cash flow rights and excess voting rights included. Significance is obtained for Tobin’s Q at the 10 % level. Ex-cess voting rights are positive and significant for all regressions in table 6.2 except for (11). Cash flow rights are negatively related to all regressions and are significant only for Tobin’s Q (10)-(12). The coefficient on excess voting rights (8) for ROA are interpreted as when the distance between votes and cash flow rights increase 10 %, return on assets increase 2.79 % on average.
Table 6-1 Founder vs non-founder
Variable ROA ROA ROA Tobin’s Q Tobin’s Q Tobin’s Q
(1) (2) (3) (4) (5) (6) Constant -0.226 -0.181 -0.174 3.481*** 3.513*** 3.757*** (0.141) (0.139) (0.142) (0.979) (0.954) (0.967) Family firm 0.079** - - 0.171 - - (0.034) (0.234) Founding family - 0.075** - - 0.459** - (0.034) (0.229) Non-founding fam. - - 0.018 - - -0.259 (0.033) (0.226) Control v. risk 0.128 0.104 0.097 0.141 0.139 -0.019 (0.085) (0.084) (0.086 (0.589) (0.575) (0.582) Control v. size 0.023** 0.023** 0.026** -0.318*** -0.332*** -0.313*** (0.011) (0.011) (0.011) (0.077) (0.076) (0.076) Control v. beta -0.059 -0.076 -0.079 1.425*** 1.414*** 1.308*** (0.068) (0.068) (0.069) (0.474) (0.465) (0.471) IND Energy -0.032 -0.053 -0.065 -0.637 -0.614 -0.765 (0.143) (0.143) (0.145) (0.989) (0.976) (0.983) IND Materials -0.005 0.016 -0.065 -0.791 -0.667 -0.736 (0.119) (0.119) (0.121) (0.823) (0.817) (0.823) IND Industrials -0.004 0.009 0.005 -0.548 -0.515 -0.485 (0.105) (0.105) (0.106) (0.723) (0.715) (0.723) IND Cons. goods -0.000 0.009 0.005 -0.375 -0.352 -0.324
(0.113) (0.113) (0.115) (0.781) (0.772) (0.779) IND Cons. services 0.031 0.014 0.029 -0.077 -0.157 -0.146
(0.111) (0.115) (0.117) (0.796) (0.788) (0.795) IND Health Care -0.152 -0.140 -0.161 0.378 0.483 0.417
(0.111) (0.110) (0.112) (0.766) (0.759) (0.765) IND Financials 0.029 0.039 0.012 -0.269 -0.145 -0.259 (0.109) (0.110) (0.111) (0.758) (0.753) (0.756) IND Technology 0.020 0.034 0.024 -0.379 0.320) -0.316 (0.109) (0.109) (0.111) (0.755) (0.747) (0.754) R2 0.15 0.14 0.12 0.17 0.19 0.18
*,**,*** Stand for significance at 10, 5 and 1 % respectively. Standard error is within parenthesis.
Table 6-2 Founder vs non-founder with effects of ownership and control
Variable ROA ROA ROA Tobin’s Q Tobin’s Q Tobin’s Q
(7) (8) (9) (10) (11) (12) Constant -0.179 -0.142 -0.164 4.026*** 3.5017*** 4.223*** (0.145) (0.096) (0.146) (1.002) (0.655) (1.001) Family firm 0.067** - - 0.147 - - (0.035) (0.239) Founding-f. - 0.051 - - 0.479* - (0.038) (0.259) Non-f. family - - 0.035 - - -0.194 (0.033) (0.227) Cash flow r. -0.142 -0.142 -0.112 -1.583** -1.692** -1.517**
(0.109) (0.109) (0.109) (0.761) (0.752) (0.752) Excess voting rights 0.306 ** (0.126) 0.279** (0.136) 0.384*** (0.125) 1.559* (0.869) 0.871 (0.931) 1.551* (0.859) Control v. risk 0.143* 0.119 0.129 0.229 0.171 0.101 (0.084) (0.083) (0.146) (0.582) (0.568) (0.579) Control v. size 0.025** 0.025** 0.028** -0.303*** -0.307*** -0.299*** (0.011) (0.011) (0.011) (0.076) (0.074) (0.075) Control v. beta -0.095 -0.110 -0.104 1.123** 1.076** 1.04** (0.070) (0.069) (0.069) (0.482) (0.472) (0.480) IND Energy -0.520 -0.067 -0.063 -0.877 -0.337 -0.967 (0.142) (0.106) (0.143) (0.984) (0.726) (0.982) IND Materials -0.033 -0.014 -0.036 -1.076 -0.391 -1.023 (0.118) (0.071) (0.119) (0.821) (0.485) (0.822) IND Industria. -0.041 -0.024 -0.038 -0.835 -0.204 -0.776 (0.104) (0.044) (0.105) (0.721) (0.299) (0.721) IND C. goods -0.029 -0.016 -0.027 -0.619 -0.019 -0.571 (0.111) (0.061) (0.113) (0.775) (0.420) (0.775) IND C. serv. 0.026 0.018 0.031 -0.153 0.316 -0.202 (0.113) (0.066) (0.115) (0.785) (0.448) (0.786) IND H. Care -0.190* -0.171* -0.197* 0.061 0.752** 0.101 (0.111) (0.054) (0.109) (0.766) (0.366) (0.766) IND Financials -0.017 0.011 -0.013 -0.525 0.143 -0.511 (0.109) (0.059) (0.109) (0.754) (0.406) (0.754) IND Technol. -0.017 -0.015 -0.017 -0.719 -0.715 -0.659 (0.109) (0.108) (0.110) (0.757) (0.789) (0.757) R2 0.18 0.17 0.17 0.20 0.21 0.21
*,**,*** Stand for significance at 10, 5 and 1 % respectively. Standard error is within parenthesis.
6.1.2 Regression: Other ownership types
The purpose of this regression is to investigate founding-families in relations to other fre-quent occurring ownership types in the regression.
Three more dummy-variables are introduced for each ownership category, the government, funds and dispersed ownership.
Table 6.3 (14) and (15) show that founding-families have positive and significant coeffi-cients for ROA and Tobin’s Q when excess voting rights and cash flow rights are not in-cluded. None of the other ownership types showed significance except for the negative co-efficient of dispersed ownership for ROA (13).
With the effects of excess voting rights and cash flow rights included, (13) and (14), found-ing families are significant only for Tobin’s Q (14). Excess votfound-ing rights are positively relat-ed and significant for ROA. Cash flow rights are negatively relatrelat-ed and significant for both ROA and Tobin’s Q, (13) and (14).
Table 6-3 Ownership-types
Variable ROA Tobin’s Q ROA Tobin’s Q
(13) (14) (15) (16) Constant -0.095 3.302*** -0.180 3.501*** (0.098) (0.679) (0.140) (0.960) Founding-family dummy 0.044 0.499* 0.061* 0.547** (0.038) (0.264) (0.035) (0.240) Dispersed dummy -0.109* 0.394 -0.089 0.545 (0.059) (0.409) (0.058) (0.394) Government dummy 0.029 -0.259 0.046 -0.309 (0.093) (0.639) (0.096) (0.658) Fund dummy -0.078 0.344 -0.065 0.458 (0.063) (0.434) (0.063) (0.428)
Cash flow rights -0.222* -1.385* - -
(0.116) (0.798)
Excess voting rights 0.236* 1.047 - -
(0.137) (0.945)
Control variable risk 0.097 0.269 0.085 0.262 (0.084) (0.572) (0.084) (0.579) Control variable size 0.025** -0.302*** 0.022** -0.328***
(0.011) (0.075) (0.011) (0.076) Control variable beta -0.102 1.045** -0.062 1.328***
(0.069) (0.477) (0.069) (0.469) IND Energy -0.047 -0.425 -0.062 -0.899 (0.106) (0.732) (0.145) (0.995) IND Materials -0.009 -0.416 0.041 -0.836 (0.071) (0.488) (0.122) (0.835) IND Industrials -0.031 -0.175 0.019 -0.589 (0.044) (0.302) (0.107) (0.730) IND Consumer goods -0.015 -0.027 0.028 -0.480
(0.061) (0.423) (0.115) (0.789)
IND Consumer ser. 0.019 0.319 0.033 -0.277
(0.065) (0.452) (0.116) (0.795) IND Health care -0.171*** 0.745** -0.118 0.337
(0.054) (0.369) (0.112) (0.769) IND Financials 0.025 0.087 0.070 -0.347 (0.059) (0.369) (0.113) (0.774) IND Technology 0.027 -0.412 0.058 -0.474 (0.110) (0.731) (0.111) (0.765) R2 0.19 0.22 0.16 0.20
*,**,*** Stand for significance at 10, 5 and 1 % respectively. Standard error is within parenthesis.
6.1.3 Regression: Family characteristics
As families tend to have higher insider representation than other owners a regression is conducted in order to investigate the effects of this representation. Dummy variables of board representation, CEO representation, chairman of the board and strong control (both board and CEO) are introduced. The results from ROA are seen in table 6.4 (17)-(20). All
characteristics are positive and significant at the 5 % level except for chairman of the board. When families are represented as CEOs firms have 14.9 % higher return on assets (18). When families are represented on the board of directors firms have 8.8 % higher re-turn on assets (17). Strong control (20) does not increase rere-turn on assets compared to when the sole CEO-representation. The results for Tobin’s Q in table 6.5 confirm the find-ings for ROA. For Tobin’s Q the coefficient is even negative, although not significant.
Table 6-4 ROA and family characteristics
Variable ROA ROA ROA ROA
(17) (18) (19) (20) Constant -0.105 -0.133 -0.143 -0.133 (0.145) (0.143) (0.145) (0.142) Board representation 0.088*** (0.044) Family CEO - 0.149** - - (0.062) Chairman of the board - - 0.000 (0.042) - Strong control - - - 0.149** (0.062) Cash flow rights -0.165 -0.149 -0.113 -0.149
(0.112) (0.109) (0.114) (0.109) Excess voting rights 0.275** 0.256** 0.359*** 0.255**
(0.129) (0.129) (0.126) (0.129) Control variable risk 0.105 0.106 0.116 0.106
(0.083) (0.083) (0.084) (0.083) Control variable size 0.024** 0.029*** 0.027** 0.029***
(0.011) (0.011) (0.011) (0.011) Control variable beta -0.110 -0.118* -0.112 -0.118
(0.069) (0.069) (0.069) (0.069) IND Energy -0.090 -0.121 -0.072 -0.121 (0.142) (0.142) (0.144) (0.142) IND Materials -0.033 -0.031 -0.028 -0.031 (0.118) (0.118) (0.119) (0.117) IND Industrials -0.035 -0.035 -0.031 -0.035 (0.104) (0.103) (0.105) (0.103) IND Consumer goods -0.020 -0.037 -0.021 -0.037
(0.112) (0.111) (0.113) (0.111) IND Consumer ser. 0.016 0.012 0.023 -0.012
(0.113) (0.112) (0.115) (0.113) IND Health care -0.188* -0.194* -0.189* -0.194* (0.111) (0.109) (0.111) (0.109) IND Financials -0.005 -0.033 -0.007 -0.033 (0.109) (0.109) (0.110) (0.109) IND Technology -0.031 -0.011 -0.009 -0.011 (0.109) (0.109) (0.111) (0.109) R2 0.18 0.19 0.16 0.19
Table 6-5 Tobin’s Q and family characteristics
Variable Tobin’s Q Tobin’s Q Tobin’s Q Tobin’s Q
(21) (22) (23) (24) Constant 4.471*** 4.177*** 4.092*** 4.176*** (0.981) (0.979) (0.995) (0.979) Board representation 0.840*** - - - (0.299) Family management - 1.002** - - (0.424) Chairman of the board - - -0.102 (0.286) - Strong control - - - 1.002** (0.424) Cash flow rights -2.019*** -1.765** -1.446* -1.765**
(0.758) (0.749) (0.780) (0.749) Excess voting rights 0.872 0.977 1.743** 0.977
(0.876) (0.885) (0.865) (0.885)
Control variable risk 0.057 0.100 0.149 0.100
(0.563) (0.566) (0.577) (0.566) Control variable size -0.333*** -0.287*** -0.297*** -0.287***
(0.075) (0.074) (0.075) (0.074) Control variable beta 1.104** 1.047** 1.091** 1.047**
(0.468) (0.471) (0.479) (0.471) IND Energy -1.096 -1.245 -0.902 -1.245 (0.962) (0.976) (0.984) (0.976) IND Materials -1.113 -1.081 -1.058 -1.081 (0.803) (0.809) (0.822) (0.808) IND Industrials -0.848 -0.839 -0.802 -0.839 (0.704) (0.709) (0.721) (0.709) IND Consumer goods -0.593 -0.709 -0.607 -0.709
(0.757) (0.764) (0.775) (0.764)
IND Consumer ser. -0.226 -0.237 -0.146 -0.237
(0.768) (0.774) (0.787) (0.774)
IND Health care 0.078 0.035 0.066 -0.035
(0.749) (0.754) (0.766) (0.754) IND Financials -0.648 -0.717 -0.559 -0.717 (0.740) (0.746) (0.755) (0.745) IND Technology -0.645 -0.718 -0.691 -0.718 (0.740) (0.745) (0.757) (0.745) R2 0.23 0.23 0.20 0.23
*,**,*** Stand for significance at 10, 5 and 1 % respectively. Standard error is within parenthesis.
6.2 Residual tests
Normality and homogenous variance are assumption of OLS regression and if they are vio-lated significance of the coefficients may not be valid (Gujarati & Porter, 2009). Non-normality and heterogeneous variance in residuals are present in all 24 regressions.
In order to investigate further what the reason for the violation of assumptions is a plot of ROA and Tobin’s Q against cash flow rights are presented in figure 6.1 and 6.2 respective-ly. The observations circled are significantly higher or lower than average. These observa-tions can alone affect the results significantly and are often a cause of non-normality and heterogeneous variance.
Figure 6.1 ROA robustness check plot
Figure 6.2 Tobin's Q robustness check plot
In order to check the robustness of the regressions all 24 regressions were conducted again without these observations. For ROA all observations above 50 % and below -50 % were removed. For Tobin’s Q all observations above 6 were removed.
When the deviating observations are removed the coefficients have the same signs as be-fore and are no longer significant. This is true for all regressions. This shows that the ro-bustness of the regression is very sensitive to these observations. These observations can-not be removed unless there is a good reason. When going through the firms’ annual re-ports and news nothing extraordinary was found that could motivate the removal of these observations.
It should also be mentioned that heterogeneous variance is still present after the removal of the deviating observations. This can be seen in the figure 6.1 and 6.2 where the variance is higher in the lower values of cash flow rights and lower in the higher values of cash flow rights. This is illustrated by the blue lines in figure 6.1 and 6.2.
7 Discussion
7.1 Main results
Table 6.1 – 6.3 regression (1)-(16) show interesting results regarding excess voting rights and founding-families. When excess voting rights are not included the coefficient of found-ing families is positive and significant as showed in regression (2),(5),(15) and (16). When excess voting rights are included, the significance of founding-families weakens or disap-pear as showed in regression (8),(11),(13) and (14). As mentioned in the descriptive statis-tics section 78.3 % of the founding families use dual-class shares. Altogether this means that founding-families using dual-class shares is the main contributor to the enhanced per-formance. Money is generally scarce and families might have used dual-class shares simply to invest in more firms and then enhanced performance in those firms.
In the theoretical framework it was discussed that although incentives to conduct moral hazard activities increase with the use of dual-class shares families might as well use the du-al-class shares to perform well because they are confident that they will perform better than other owners. The latter supports the results discussed in the previous paragraph.
In the study by Cronqvist and Nilsson (2003), which focuses on Sweden, 92.5 % of the founding families used dual-class shares and they also find a positive coefficient for excess voting rights for Tobin’s Q, although not significant. In Europe the situation is different. Both Isakov and Weisskopf (2014) and Barontini and Caprio (2006) find that excess voting rights are significantly negatively related to performance while at the same time finding that families perform better, for Switzerland and Western Europe respectively. This indicates that the purpose of using dual-class shares leans more to extraction of private benefits and moral hazard activities in Europe than it does for Sweden.
Table 6.4 and 6.5 regression (17)-(24) show the results when family insider representation are included. The strongest result is obtained for family CEO (18) and (22). Firm-specific knowledge, discussed in the institutional framework, is an explanation of this result. Previ-ous empirical work by Villalonga and Amit (2006) and Chu (2011) show similar findings for United States and Taiwan respectively.
Regression (20) and (24) show that board representation along with CEO does not further enhance the performance, which indicates that it is the CEOs that drive the enhanced per-formance. Regression (17) and (21) show that representation as a single board-member has lower performance than CEO representation, which further strengthens the view that it is CEO and firm-specific knowledge that drives the results.
Representation as chairman of the board does not increase performance which is showed in regression (19) and (23), for Tobin’s Q in regression (23) the performance is even nega-tive, although not significant. This is quite surprising as the coefficient for board represen-tation was positive and significant. It is logical to expect that the coefficient for chairman to be positive as well then. The reason for the difference in results could be that the chairman of the board representation is a sign of extreme control with the purpose of conducting moral hazard activities. The underlying assumption for this is that the chairman of the board can more easily conduct moral hazard activities than a regular board member be-cause of the more powerful position.
As discussed in section 6.2 the regressions are not robust to residual testing. There is non-normality and heterogeneous variance present and a few observations are determining
whether the coefficients are significant or not. This means that decision-makers and legisla-tors and other parties interested in the results should be careful when drawing conclusions from the results of this thesis.
7.2 Further results
The control variable size is significant in every regression indicating that it was the right approach to use natural logarithm of total assets. This decision was based on the observed non-linearity between size and performance in figure 5.3 and 5.4.
Regression (13) in table 6.3 show a negative coefficient for dispersed ownership and signif-icance at the 10 % level. As discussed in the theoretical framework, incentives to actively monitor management increase with large cash flow stakes. This is because owners receive more in return on their possession in relation to their time spent on monitoring when the cash flow stake is large. In absence of active monitoring managers are not “watched” and can enjoy private corporate benefits instead of working which reduces performance of the firm. Results indicate that agency problems are present. The results were only significant at the 10 % level of confidence and Tobin’s Q did not show any significance of dispersed ownership.
Table 6.2 show significant coefficients for the health care sector. For ROA the coefficients are negative and for Tobin’s Q they are positive. The reason for this could be that the sec-tor contains firms in the “research stage” that plan to launch a new medicine or a medical method in the future. These firms have large research and development costs (affecting ROA negative this year) while possibly having promised future outlooks which make them attractive to investors, thus raising Tobin’s Q.
The constant is always positive and significant for Tobin’s Q in all regressions. The signifi-cance of the constant reveals that average Tobin’s Q value is estimated to be positive. This is not surprising as all Tobin’s Q values are positive per definition. It is necessary to include a constant as the regression otherwise are forced to go through the zero point which might lead to inaccurate results. This also then explains why the constants for ROA are not sig-nificant as ROA varies above and below zero with no obvious inclination in either direc-tion.
7.3 Endogeneity
Endogeneity refers to the issue of ownership structure being determined by performance. If endogeneity is present in this thesis it would indicate that families decide to hold a spe-cific amount of votes and cash flow rights based on firm performance. Choosing cash flow rights, excess voting rights or using family-dummies as independent variables would then cause biased results.
In order to investigate if endogeneity is a problem in this thesis a review of the literature on the subject is conducted. Demsetz and Lehn (1985) provide evidence of ownership con-centration being determined by variation in profits, thus indicating endogeneity. In a study on ownership structure and performance Demsetz and Villalonga (2001) present evidence that ownership structure does not affect firm performance.
The previously mentioned studies use the five largest shareholders in the analysis. The in-clusion of the top five largest shareholders indicates a problem. It is not likely that all, es-pecially not shareholder ranked 4-5 in size, would have a big influence on the firm. It is more likely that the largest shareholder is the one with most influence on the firm. A more
correct way to address the endogeneity issue is then to investigate the largest shareholder and its relation to performance.
Gugler and Weigand (2003) include only the largest shareholder when investigating if own-ership affects performance. Their findings show that the largest shareholder is affecting the performance and that the results were significant at the 1 % level, this was true for both USA and Germany.
The definition of a family-controlled firm in this thesis is that a family is the largest owner with vote ownership of 10 % or more. With respect to the results of Gugler and Weigand (2003) it is indicated that endogeneity is not a problem in this thesis.
7.4 Social implications
Results showed that founding-family firms perform 6.1 % better than other firms for ROA. These results can be used by individual investors, funds and other stakeholders to enhance performance of their own portfolios. For example, consider two similar firms where one was created by a family and the other was created by the government. When de-ciding which of the similar firms to invest in the results of this thesis suggest that the founding-family firm is the best choice. With the interest-on-interest effect of accumulated profits the difference in the long run would be substantial.
The above discussion has implications for society as a whole. Individuals are taxed on their profits from investing in stocks and the corporations are taxed on their accounting based profits. As founding-family firms perform better, the tax-revenue is higher with resepct to these firms which is of benefit to the whole society.
8 Conclusion
How do Swedish listed family-controlled firms perform compared to other firms?
Results show that founding-family firms perform better than other firms. The coefficient of the variable excess voting rights (the result of using dual-class shares) was positive in all regressions and significant in a majority of the regressions. This suggests that the maximum deviation from one-vote-one-share (10:1) could be raised further, thus facilitating for the competent founding-families and others to control more firms and enhance performance. The regressions above are not robust to residual testing. Carefulness is therefore advised when implementing the conclusions from these results.
What is causing the results in the previous question?
Results show that firm-specific knowledge of the CEO is the main cause of the enhanced performance of founding-family firms. Representation on the board of directors is also a source to enhanced performance regarding founding-family firms, although less than for CEO. The code could state a recommendation that firm-specific knowledge of insiders is preferable.
The regressions regarding family-CEO are not robust to residual testing. Carefulness is therefore advised when implementing the conclusions from these results.
8.1 Future research
Executing this thesis again next year is suggested in order to compare the robustness of the regressions in the two studies.
List of References
Atlas Copco (2013) Annual report 2012. Nacka: Atlas Copco AB. http://viewer.atlascopco.com/Atlas_Copco_Arsredovisning_2012/Atlas_Copco_Arsredo visning_2012.pdf
Barontini, R & Caprio, L (2006) The effect of family control on firm value and perfor-mance: Evidence from continental Europe. European financial management. 12(5): 689-723 Berglund, T. (2011) The elusive marginal q. Applied financial economics. 21(20): 1489-1493 Bjuggren, P-O. & Palmberg, J. (2010) The impact of vote differentiation on investment performance in listed family firms. Family business review. 23(4): 327-340
Bjuggren, P-O., Wiberg, D. & Eklund, J. (2007) Ownership structure, control and firm per-formance: the effects of vote-differentiated shares. Applied financial economics. 17(16): 1323-1334
Chu, W. (2011) Family ownership and firm performance: Influence of family management, family control and firm size. Asia pacific journal of management. 28(4): 833-851
Claessens, S., Djankov, S., Fan, J. & Lang, L. (2002) Disentangling the incentives and en-trenchment effects of large shareholdings. The journal of finance. 57(6): 2741-2771
Corporategovernanceboard (2010) The Swedish corporate governance code.
http://www.corporategovernanceboard.se/media/45322/svenskkodbolagsstyrn_2010_eng
_korrigerad20110321.pdf 2014-05-10
Cronqvist, H. & Nilsson M. (2003) Agency costs of controlling minority shareholders. Jour-nal of financial and quantitative aJour-nalysis. 38(4): 695-719
David, P & Kochhar, R (1996) Barriers to effective corporate governance by institutional investors: Implications for theory and practice. European management journal. 14(5): 457-466 Demarzo, P. & Berk, J. (2011) Corporate finance 2nd edition. Edinburgh: Pearson education
Demsetz, H. & Lehn, K. (1985) The structure of corporate ownership: Causes and conse-quenses. Journal of political economy. 93(6): 1155-1177
Demsetz, H. & Villalonga, B (2001) Ownership structure and corporate performance. Jour-nal of corporate finance. 7(3): 209-233
Dyer, W (2006) Examining the “family effect” on firm performance. Family business review. 19(4): 253-273
Faccio, M & Lang, L (2002) The ultimate ownership in Western European corporations. Journal of financial economics. 65(3): 365-395
Fama, E & French, K (1992) The cross-section of expected stock returns. The journal of fi-nance. 47(2): 427-465
Gujarati, D. & Porter, C. (2009) Basic econometrics 5th edition. New York: McGraw Hill
Gugler, K. & Weigand, J. (2003) Is ownership really endogenous? Applied economics letters. 10(8): 483-486
Isakov, D. & Weisskopf, J-P. (2014) Are founding families special blockholders? An inves-tigation of controlling shareholder influence on firm performance. Journal of banking & fi-nance. 41(C): 1-16
Jensen, M & Meckling, W (1976) Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of financial economics. 3(4): 305-360
NASDAQ (No date). Share prices for all companies listed on Nasdaq Omx Nordic.
http://www.nasdaqomxnordic.com/aktier 2014-03-05
Pierce, J., Kostova, T. & Dirks, T. (2001) Toward a theory of psychological ownership in organizations. Academy of management review. 26(2): 298-310
Reeb, D. & Andersson, R. (2003) Founding-Family ownership and firm performance: Evi-dence from S&P 500. The journal of finance. 58(3) : 1301-1328
SFS 2005:551 Aktiebolagslagen. Stockholm: Department of justice.
Tirole, J. (2006) The theory of corporate finance. Princeton: Princeton university press
Villalonga, B. & Amit, R. (2006) How do family ownership, control and management affect firm value ? Journal of financial economics. 80(2): 385-417
