• No results found

The effect of ownership concentration on the performance of Europe’s biggest tech companies

N/A
N/A
Protected

Academic year: 2021

Share "The effect of ownership concentration on the performance of Europe’s biggest tech companies"

Copied!
24
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

The effect of ownership concentration on the performance of Europe’s

biggest tech companies

June 25, 2020

Student: Alex Beeren (11111135) Supervisor: Christopher Graser

Programme: Economie en Bedrijfskunde Track: Finance and organization

(2)

Abstract

This paper researches the relation between ownership concentration and performance of Europe’s 200 biggest Tech companies. With the use of data and literature, it shows how the ownership concentration affects the independent variable return on Assets. With the control variables Debt, Market

Capitalisation and the year of establishment, the regression is done via Stata. All necessary data is provided by the database Orbis, where all information from the years 2010- 2019 of the top 200 based on Market Capitalisation is chosen for the regression. Since almost every regression model resulted in an insignificant relation between the created Herfindahl index for ownership concentration on the dependent variable, based on this data, there is not sufficient evidence to concluded that the index on its own have a definitive positive or negative effect within the tech industry in European Union. The insignificance effect can be concluded by the complexity of factors which affect firm performance ranging from country specific effects or the large variety of companies present in the tech industry. More variables have to be added to account for country specific conditions or a more specific industry have to be chosen.

(3)

Statement of originality

Hereby student Alex Beeren with student number 11111135 declares that this article was written by him. The above-mentioned student also states that both the literature and the data come from the sources and database mentioned. This paper is written for the bachelor Thesis Economie en Bedrijfskunde and is intended for the University of Amsterdam.

(4)

Contents

1 Introduction 5 2 Related literature 8 2.1 Literature review 8 2.2 Dataset selection 10 2.3 Hypotheses 12

3 Data description and Methodology 13

3.1 Models 13 3.2 Data 14 3.3 Descriptive statistics 16 4 Results 17 5 Conclusion 20 6 References 21

(5)

1

Introduction

Last year, Ferdinand Piëch passed away at the age of 82. In his book: ‘Auto. Biographie.’ Mr Piëch tells how his vision, way of management and risky projects made him one of the most influential bosses in the car industry. Piëch started his career with Porsche. In the eight years working for Porsche, Mr Piëch was responsible for the company’s successful motorsports operations and was the chief responsible for the iconic Porsche 917 which brought the maker their first win at the Le Mans 24 Hours in 1970. After Porsche, Mr Piëch joined the board of Audi, where he oversaw a branding transformation. Audi cars were seen as solid but unremarkable, Mr Piëch managed to reorganise the company and brand so that it could compete with big German brands like BMW and Mercedes-Benz. After his success as Chairman of the board of Audi, Mr Piëch became CEO at Volkswagen Auto Group from 1993 till 2002 and after that having been the Chairman of the supervisory board of Volkswagen from 2002 till 2015. In the beginning, Volkswagen AG was at a point of almost filing bankruptcy witch a turnover of 40 billion and a market capitalisation of 4 billion, and in the 22 years under Piëch’s supervision, Volkswagen group acquired brands like Bentley, Bugatti, Lamborgini, Ducati, Porsche, MAN and Scania and became Europe’s biggest Automotive Manufacturing Company with a turnover of 202 billion euros and a market capitalisation of 111 billion euro. One of the reasons for the success was bringing together the production of the different car brands of the Volkswagen group. For example, the brands like Volkswagen and Audi had almost 60 per cent of standard parts, and with bringing these production processes together, it resulted in lower production and

development costs. In addition to Piëch’s primary responsibility for, for example, the production of the Bugatti Veyron, he also had unsuccessful projects where he imposed too high demands on the

engineers and set unachievable goals. Except that Piëch was an innovative expert, there has been much criticism on his bullish and authorities’ style of leadership. Even though part of the council often did not support his idea, his persuasion and vision made Volkswagen Auto Group the largest car

manufacturer in Europe.

Volkswagen Auto Group has always tried to find a balance between the input of the board and the input of the shareholders, mainly because of a clear Hierarchy in the German norms and values, but many companies are abandoning this and doing it for the money and relinquishing the majority of shares. With that also the control of the company. As a result, the board has minority control, and that raises the question of what the best ratio is. (Dave, L. 2015)

A Chief Executive Officer is the company’s top decision-maker. The CEO delegates a lot of the tactical responsibilities to other managers and focuses mainly on the strategic issues, like entering new parts of the market, dealing with the competition and with which companies creating a potential partnership. The CEO maintains contact with the supervisory board and is ultimately responsible for the performance of the company. The Chairman is the head of the supervisory board and thus responsible for protecting the interests of shareholders. The shareholders also choose the supervisory

(6)

board, and the supervisory board decides whether the CEO or other members of the board of directors does their job well or should be replaced. The CEO, on his place, also influences the selection of the board of directors, due to his selection of senior executives, which often likely results in a place on the board, given the rules within the company.

Further, the board of directors meets several times a year to discuss long-term visions, assess the results of the company and vote for important strategic movements of, for example, the CEO. The Chairman ultimately has the primary voice in important long-term decisions about the company, but the CEO’s are responsible for day-to-day decisions and major implementations within the company. (Shekshnia, S., & Zagieva, V. 2019) For every company, it differs how much power the supervisory board and the board of directors have. One company gives the CEO much freedom to implement his vision on the company, such as Mr Piëch in his time as a CEO, and the other company has a board of directors that is very involved and has a definite opinion on all decisions of the board of directors, such as Mr Piëch had on his followers during his time as director of the board at Volkswagen AG. In many cases, the above structure goes well within the organisation, but as in the case of Ferdinand Piëch, other well-known visionaries or a skewed distribution within the interested parties, risk aversion, a lack of knowledge of the company or market, or another lack in the supervisory board, it will lead to a hinder in the growth of the company. The supervisory board still represents the interest of all shareholders, thus for example, if we have a company with many shareholders and all with another vision, the company reduce the single-minded approach and thus, reduces the possible inefficient way of managing the company. This sounds very clear, but one should only look at the piece of control that each individual has. If there are a lot of small shareholders, and there is one individual with more than half of the shares, his or her opinion will then weigh most strongly, and the person will retain control. The amount of control within the company, therefore, depends on the relative proportion of shares that an individual hold relative to the rest. The dispersion into shareholders is also referred to as the ownership concentration. You got two extreme ownership structures where on one side has a high ownership concentration with one shareholder and on the other side has a low ownership concentration with an infinite number of shareholders. Both have advantages and disadvantages, and in practice, the amount of shareholders vary between these two extremes.

The interests of the shareholders and managers may differ from each other. In general, Shareholders want to see a high profit and managers have their own preferences like salary and a potential bonus. When there are numerous small shareholders, and thus a low ownership

concentration, they will not actively be engaged by the management activities. Management will then have free rein to implement their strategic plans, as, in practice, A supervisory board for many shareholders are less likely to find the vast majority opposed to the CEO’s decisions. This means that the board of directors both determine the short term and long-term goals which may lead to a moral hazard problem (Pilbeam, K. 2013) if managers take decisions based on their own preferences. An example of this is a CEO who takes many risks to gain short-term profits and earn their bonus, while it

(7)

may be harmful to the firm in the long-term. There are also various ways in which management receives a shared bundle, for example, which ensures that the management also achieves the long-term goals. On the other hand, when you have a firm with only one or a few shareholders, and thus a high ownership concentration, the board of directors have to follow the strategy mentioned by the

shareholders. (Baker, M. P., & Wagonfeld, A. B. 2004) If they want to follow a strategy where they can achieve a high profit for this year, but due to the lack of knowledge within the shareholders is harmful to the long-term, this will also reduce the value of the firm.

These considerations of different structures within an organisation have been subject of discussion for years. Since several studies have been conducted within Europe, where the tech sector in Poland is an exception to the rule in general. Because in most studies, there is a negative effect between ownership concentration and return on assets, for tech firms in Poland, there is a positive correlation between the two. Answers this thesis, whether the deviating result regarding ownership concentration in Poland can be generalised to the tech sector within Europe.

This all leads to the following question, which is the basis of this paper:

(8)

2

Related literature

2.1 Literature review

In this section, the paper discusses various studies in which research has been conducted into the relationship between ownership concentration and performance of the companies and all the advantages and disadvantages of different organizational structures. A number of articles related to this topic will also be discussed. Based on all the named articles, the paper will eventually combine all useful and relevant information to form the basis of the paper.

As mentioned earlier, the basis of all corporate governance considerations is the effect of ownership structure on company performance, and this has been a topic for years now. This is proven by the paper of Jensen and Meckling, written in 1976 and already looking at different types of ownership structure and their performance.

In the year 2000, Thomsen and Pedersen also examined the impact of the ownership structure on the company economic performance. They took a sample of the 435 largest European Companies. The paper looked at the effect of the ownership concentration on the performance of the firm and at the identity of the owners, with the associated interest. Within the regression, it controls for the industry, the capital structure and the nation effects. After performing the regression, they found the following things. First of all, there is a positive effect between ownership concentration and shareholder value. Second, there is a positive effect between ownership concentration and the return on assets, from now on called ROA. The second result state that in Europe, based on the paper with the factors above taken into account, the fewer shareholders, the higher the return on assets, but the effect levels off for high ownership shares, but the effect diminishes as they become larger

shareholders. Furthermore, one can also say that the type of owner makes a difference when

determining the strategy with a vision of the objectives they have set. When a person has a large stake, he may be passively or actively involved in the business. Someone can let management go or suppress it a lot. Unlike other shareholders, financial investors will strive for the highest possible share value and profit, but the turnover growth will not fall within their primary interest. This may differ for family companies, banks, governments or institutional investors.

While the effects of ownership structure on the performance of a company had been studied multiple times, previous analyses neglected the endogeneity issue between ownership

concentration and the firm’s performance as well as between different ownership dimensions. To form a well-grounded conclusion between ownership concentration and performance, it is important to look at this potential endogeneity. In 2007, Sir Groß did an empirical study for German traded companies with this issue as the main subject. In his research, he models the simultaneous relationship between the different dimensions of ownership, the ownership concentration and the return on assets by using the simultaneous equations methodology. ‘Furthermore, there are plenty of factors also assumed to be

(9)

endogenous on the relation of ownership and performance. Other corporate governance mechanisms and mediating factors, such as capital structure could be included as endogenous variables in the simultaneous equations system.’ (Groß, K. 2007) The paper shows that the results are in some terms contradicting some often-accepted ideas regarding ownership concentration and performance. There are a few underlying endogenous relations, but there was no significant effect of the model, and thus additional research needs to be done on the effect of the determinants and the effect on the ROA.

In the case of traditional family businesses, there are usually few shareholders. To take a closer look at the effect of companies with few shareholders on the ROA, the paper examines the results of publications by Andres, C. (2008) and Hamadi, M. (2010). In the first paper, Andres uses data of the 275 largest German exchange-listed firms. Hamadi uses a sample of Belgian listed firms, and both use Tobin’s Q as their main dependent variable. It separates the family effect from general block holder effects to get a possible relation, and the findings in both papers are almost similar to each other. Family businesses have a positive relationship with the performance of the company under certain conditions. If the large controlling shareholders are not organized in voting blocks and if the founding-family is still active in the supervisory board or on the executive position, family firms are not only more profitable, but they also outperform a wide range of shareholder running companies.

A possible interpretation of the findings is that members of families feel more involved with the shareholders when they are still on the board or own a big set of the company. This means that also for family companies with a high ownership concentration in Belgian and Germany. It holds that with a high ownership concentration, they achieve a higher ROA.

In the previous two articles, German and Belgian listed family companies were tested and came to the same results, but demographical differences can strongly influence comparable industries due to, for example, a difference in the tax system, risk aversion, differences in accounting systems or other factors. In the paper of Fredrik, J. (2013), the article explores the relationship between ownership concentration and corporate performance and legal investor protection for both private and public firms on a cross-country study within Europe. Thirty-seven thousand firms including firms from Ireland, Belgium, Finland, Austria and Ukraine. These countries present five different legal families, namely the German, French, Scandinavian, Common Law and the Eurasian law. They took all data from the period 1996-2005 and found out that the legal investor protection, different tax systems and different accounting systems have a significant effect on the performance in particular countries. In Ukraine, with the least advantageous standards, they found out that high ownership concentration is related inversely to the ROA, while the level of ownership concentration for the other countries has a positive relation. Demographical factors thus can have a significant effect on country-specific circumstances on the return on Assets within the research done by Fredrik.

Krivogorsky, V. and Grudnitski, G. (2010) also examined the country-specific institutional effect on ownership concentration and performance of continental European firms. In their literature, they used the performance of firms in Austria, Belgium, Germany, Spain, France, Italy, Portugal and

(10)

the Netherlands. Using a hierarchical moderated multiple regression analysis with variables like dividend payoffs, sales, interest rate, equity growth, dummy variables for industries, potential inflation and the ROA, they found out that there is differential validity established for the relation between ownership concentration and performance measured by return on shareholders’ funds. Country specific effects are thus important for the relevance of cross- country research.

After analyzing the above literature, in general, the results are very similar to each other and give the positive relationship between ownership concentration and performance of the company, ROA. A paper written by Grosfeld, I. (2009), found a deviating result. The paper again explores the relationship between ownership structure and firm value, but now for Polish firms listed on the Warsaw Stock Exchange. Overall, the hypothesis was confirmed by the fact that ‘mature’ firms with higher ownership concentration generally perform better than firms with more dispersed ownership. On the other hand, the opposite appears to be the case for high-tech companies. For high-tech firms with a large share of knowledge related activities, higher ownership concentration is associated with lower firm value. The question which arises is the fact if high-tech firms are different from other sectors, and this will be the main idea in this thesis if this statement can be generalized to the whole European Union.

2.2 Dataset Selection

In the articles that have been reviewed in the previous section, a variety of aspects came to light that emphasises the complexity of the subject. Therefore, it can be concluded that a subject that is influenced by such a large quantity of variables will be hard to research fully in a research paper of this magnitude. Although this might not be a promising outlook, it does indicate what key points will have to be researched in order to create a well-structured and scientifically based paper. The lessons learned from the previously mentioned papers will thus increase the effectivity and efficiency of the research and will benefit the process of selecting the necessary variables. Through the analysation of these articles, multiple variables have come to light, both controlling and independent, that will be defined in the segments to come.

It is starting with the dependent variable. It is shown that the most reoccurring variable used in the papers researched is the Return on Assets (ROA). While the possibilities seem to be very diverse and range from debt levels to earnings after interests and taxes and payout schemes, the ROA appears in the majority of the available literature. The Return on Assets is calculated by first taking the total market value of the specific firm and dividing this by the total amount of assets it owns. This creates a ratio that uses both the firm worth and the firm size according to its balance sheets to create a

(11)

variable will be the sole depending variable in this research because of the effectivity and completeness of its use.

The next variable that will be looked upon will be the independent variable, and the focus of this research paper. The ownership concentration, defined as the amount of leverage the collective shareholders have over the firm, is what is looked upon as a possible impacting variable in this paper. Although the explanation of this variable is fairly simple, the creation of it as a variable is more complicated. Since shareholders can be either an individual, a corporation, a fund, a country or the firm itself, it is difficult to determine the effect of these different kinds of entities. Therefore the difference in the type of owner is not considered as a part of this research. The ownership

concentration, either through different kinds of entities or through similar entities, merely shows the impact the shareholders have on the business itself. Who’s in charge is therefore not of importance. The solution to the creation of the variable is an Index called the Herfindahl Index. This Index takes every owner into account by summing up the square of its ownership percentage, thus taking both the weight as well as the number of shareholders into account. For this, the ownership percentages of the Top 200 tech companies in Europe will be used every year from 2010 until 2019. Publicly available stocks will be accounted for as one single owner, as well as state ownership and non-outstanding stocks to show the amount of control the shareholders together have.

To understand the impact of the independent variable control variables, need to be used in order to not accredit all changes in the Return on Assets purely to the Index itself. In the articles described in the literary review, the specificity of some articles like the one researching the Polish tech industry created the possibility of cultural or political influences. Though these effects are manageable when researching a specific country, this will not be the case in this paper since this paper aims to eliminate country-specific differences through the selection of a continental dataset. Therefore, the entirety of Europe has been chosen as the demographic region. This, however, does impact the selection of control variables since the availability of data and the country-specific manipulation of this data might cause a distortion in the results. Since distortion is preventable through using internationally specified variables, this will be the case for this paper. The first two control variables used will be the

logarithmic value of both the firm debt as well as its market capitalisation. These present the levels of debt a firm has and the value of all of its shares combined. The debt was chosen since more debt is shown to decrease firm growth, while a higher market capitalisation has been linked to faster growth due to higher economic efficiency and other advantages. (Pilbeam, K. 2013) The logarithmic characteristics of these values are due to the impact a change in these values has on the return on assets. An increase of a thousand euros will impact the return on assets of a small company more than it does in a multi-billion-euro corporation. Therefore, relativity is important, thus creating the need for logarithmic values.

The final control variables used are a set of dummy variables for all the major cultural periods in which these companies have been founded to see if their company structure and experience impacts

(12)

the amount of trust and control shareholders have. (Hamadi, M. 2010) By later combining these dummy variables with the Index a combined variable will be created to see if the Index is of greater influence in companies of a certain age, which could be attributed to their need for investment or available knowledge in their respective industries.

2.3 Hypotheses

With all variables selected and having read all the literature, one can articulate a well-understood hypothesis about the relationship between ownership concentration and the ROA of companies in the high-tech sector.

There are three main outcomes from this regression, with combinations possible and justifiable after researching the literature from the previous chapter. In the first case, there will be a significant positive correlation between the two variables. This could be the result when the papers are written by Thomsen and Pedersen, Andres and Hamadi also hold for firms in the European High-Tech industry. The second possible outcome will be an insignificant effect. This means that the data is not sufficient, and we can state that there would neither be a positive nor a negative correlation between the ownership concentration and the ROA. In the last scenario, it will be possible that there will be a significant negative correlation between the two. In this scenario, the paper states that based on the sample used in the paper, the high-tech firm is an exception on the other industries and has a negative correlation between ownership concentration and the ROA.

The tech industry has a lot of firm-specific knowledge. Due to the complexity of many occupations of tech companies, it is often better to let management with expertise in the field decide rather than shareholders who are not involved with the company on a daily basis. When a company has only a few large shareholders, they often have a lot to say, and management must act according to the wishes of these shareholders. When a company has a lot of small shareholders, they are usually not very directly involved in the company and management has more freedom and power to make

decisions which makes it for tech companies with firm-specific knowledge activities more efficient. (Baker, M. P., & Wagonfeld, A. B 2004)

This may be the cause of the negative correlation between ownership concentration and the performance of the company for firms in the tech industry. Since the results and research are so diverse and is largely depended on the type of firm, as well as its specific characteristics, the basis of my hypothesis will be defined as a combination of these results in the form of a bell curve. Resulting in the following hypothesis.:

There is a negative correlation between ownership concentration and the ROA for the outliers of Europe’s 200 biggest tech companies, while the median companies will experience a positive effect.

(13)

3

Dataset description and Methodology

3.1 Models

Multiple models were used to support the research of this paper. All are linear-logarithmic models in which the heteroskedasticity will be tested. If these tests conclude the need for robust standard errors, a robust model will be used to counter the problem of an inconsistent variance. Every model will have the sole purpose of determining the impact of the constructed Herfindahl-Hirschman index on the Return on Assets of the firms in the dataset, but all will take different variables and complications into account.

The use of the Return on Assets (ROA) is a regularly used variable in papers researching the impact of ownership concentrations on firm value (Thomsen, S., & Pedersen, T. 2000). The Control variables Debt (DEBT) and Market Capitalisation (MC) are also regularly seen in these papers but are often implemented in their logarithmic form, as seen in the article of Grosfeld.

Because of the before-mentioned relationship between the ROA, DEBT and MC, this will also be the case in this paper.

The independent variable used in every regression, combined or individually, will be the Herfindahl index constructed from summing and square the fraction of the total stock every

shareholder owns. This index has been widely used in economic theory (Fredrik, J. 2013) and is well applicable in this research since it not only measures the number of shareholders or the total

percentage but combines these traits in a weighted total.

The Dummy variables are included to exclude the effect of different company age-groups as chosen upon the founding during large impactful periods in modern times. Impactful periods like the post-war period, the application of the first computers, and the modern technological revolutions are differentiated upon to exclude the impact of different technological eras, which are most impactful in the tech industry that will be researched.

To check if the impact of these dummy groups can be combined with the impact of the index, a combined variable will be used to research if the interaction between these variables is present. While dummy group 4 will be excluded to prevent multicollinearity. This period was chosen for its significance and its relevance when researching for different time periods.

To conclude this chapter, the following models will be built with the variables explained in the next paragraph:

Regression 1:

(14)

Regression 2:

ROA = β0 + β1∗LogMC + β2∗LogDEBT + β3∗DUMMY1 + β4∗DUMMY2 + β5∗DUMMY3 + β6∗DUMMY5 + β7∗INDEX1 + β8∗INDEX2 + β9∗INDEX3 + β10∗INDEX4 + β11∗INDEX5 + ε0 Time Controlled Regression:

ROA = β0 + β1∗LogMC + β2∗LogDEBT + β3∗ INDEX + β4∗DUMMY1 + ε0 ROA = β0 + β1∗LogMC + β2∗LogDEBT + β3∗ INDEX + β4∗DUMMY2 + ε0 ROA = β0 + β1∗LogMC + β2∗LogDEBT + β3∗ INDEX + β4∗DUMMY3 + ε0 ROA = β0 + β1∗LogMC + β2∗LogDEBT + β3∗ INDEX + β4∗DUMMY4 + ε0 ROA = β0 + β1∗LogMC + β2∗LogDEBT + β3∗ INDEX + β4∗DUMMY5 + ε0

3.2 Data

For collecting the data, this paper uses the internet database Orbis. On Orbis, the selected search steps to get an accurate list with companies were as follows: filter on the European Union to get an

overview of all companies. After that, more specify the search and filter on the BvD sectors: computer software, computer hardware, industrial, electric and electronic machinery, communications,

biotechnology and life sciences and information services to get a broad list of tech companies. To organise this list of the largest 200 companies, the variable Market Capitalisation must be added, and the companies must be sort descending on the basis of this variable. The requested variables for the research in this paper were the following:

- The Market Capitalisation from 2010 to 2019 - The total liabilities and debt from 2010 to 2019 - The total assets from 2010 to 2019

- The operating revenue from 2010 to 2019

- The total and direct percentage of ownership of the top 50 shareholders from 2010 to 2019 - Dummy variable 1 for establishment before 1950

- Dummy variable 2 for establishment between 1950 and 1980 - Dummy variable 3 for establishment between 1980 and 2000 - Dummy variable 4 for establishment between 2000 and 2010 - Dummy variable 5 for establishment between 2010 and 2020

These variables collected in Orbis were exported to Excel to calculate the necessary variables needed for the regression. First, the ROA is calculated with the help of the Assets and Operating Revenue. With the help of calculation rules in Excel, the ROA can be calculated for every year by dividing the Revenues against the total Assets. This ROA is the independent variable in the regression. After that, the total and direct percentages are used to calculate the Herfindahl-Hirschman Index as a measurement of ownership concentration. It is calculated using calculation rules in Excel by squaring

(15)

the percentage of ownership of each shareholder and then summing the resulting numbers. Through mismatches in de database were there any wrong numbers, which indicated a shareholder percentage of more than 100%. Because it is impossible the determine where the error is, for companies with a total number of shareholders above 100%, all shareholders are divided by the sum of these

percentages. Finally, newer companies and those that lacked year are removed from the dataset. the percentage of ownership of each shareholder and then summing the resulting numbers. Through mismatches in de database were there any wrong numbers, which indicated a shareholder percentage of more than 100%. Because it is impossible the determine where the error is, for companies with a total number of shareholders above 100%, all shareholders are divided by the sum of these

percentages. Finally, newer companies and those that lacked year are removed from the dataset.

Standardised variables Description

Market Capitalisation Total market value of the company in Million Euro

Total liabilities and debt Amount of Outstanding Liabilities and debt in Million Euro

Total Assets Amount of Assets in Million Euro Operating revenue Revenue earned in Million Euro Ownership percentage Amount of percentage owned by each

shareholder

Calculated variables Description

ROA Operating revenue divided by Total Assets

LnMC Natural logarithm of Market Capitalisation

LnDEBT Natural logarithm of Liabilities and Debt

INDEX Herfindahl-Hirschman index for ownership

concentration

DUMMY1 Establishment before 1950

DUMMY2 Establishment between 1950 and 1980

DUMMY3 Establishment between 1980 and 2000

DUMMY4 Establishment between 2000 and 2010

DUMMY5 Establishment between 2010 and 2020

INDEX1 Herfindahl INDEX x DUMMY1

INDEX2 Herfindahl INDEX x DUMMY2

INDEX3 Herfindahl INDEX x DUMMY3

INDEX4 Herfindahl INDEX x DUMMY4

INDEX5 Herfindahl INDEX x DUMMY5

(16)

3.3 Descriptive statistics

Variable

Obs

Mean

Std. Dev.

Min

Max

DUMMY1

2000

.32

.4665928

0

1

DUMMY2

2000

.18

.3842835

0

1

DUMMY3

2000

.355

0.478633

0

1

DUMMY4

2000

.11

.312968

0

1

DUMMY5

2000

.035

.1838257

0

1

INDEX1

2000

.0856472

.2927921

0

1

INDEX2

2000

.055715

.1930892

0

1

INDEX3

2000

.0971823

.2534906

0

1

INDEX4

2000

.0302776

.205714

0

1

INDEX5

2000

.0069799

.1343689

0

1

ROA

1837

.8152329

.4065249

.0009419

2.646

INDEX

2000

.2758019

.4386302

0

1

LnMC

1699

3.333601

.6854847

.7764186

5.170

LnDEBT

1829

2.939701

0.882699

-.2742115

5.095

Table 2: Descriptive statistics

In table 2 in the appendix of this paper the descriptive statistics of the variables used in this paper can be found. An average Return on Assets of about 81.5% was found in the 200 companies used for this study. The logarithmic value of the Market Capitalisation reaches from 0.77 to 5.16, which indicates a range between 50.000 and 150 Million. The logarithmic value of Debt has a range between -0.27 and 5.09, indicating a similar total range between 0 and 150 million. Both the average Log of the Debt and the average Log of the Market Capitalisation lay close to the 3 with 2.9 and 3.3 respectively, meaning an average of about 1 million euros. The Herfindahl index covers the complete range from 0 to 1, meaning that both a completely concentrated company, as well as a completely dispersed company are present in the dataset. The Dummy variables naturally range from 0 or 1 and combined with the index again result in a coverage of the complete 0 or 1 scale. The oldest company in the dataset has been staying afloat for about 331 years now, while the youngest corporations are a mere 2 years old. Due to the use of logarithmic values and dummy variables no outliers were found in this dataset.

(17)

4

Results

Table 3: Regression 1and 3 outcomes

ROA ROA (1) (2) INDEX .0081 - (.0229) - LnMC -.1312*** -.1003*** (.0241) .0229 LnDEBT -.0216 -.0656*** (.01907) (.0186) DUMMY1 - .3119*** - (.0447) DUMMY2 - .2740*** - (.0489) DUMMY3 - .0398 - (.0449) DUMMY5 - -.0013 - (.0866) INDEX1 - -0.1710 - (.0315) INDEX2 - .0347 - (.0559) INDEX3 - .0479 - (.0443) INDEX4 - -.0773 - (.0772) INDEX5 - -.0055 - (.0900) Standard errors in parentheses

(18)

Table 4: Time-controlled regression outcomes (regression 2)

ROA ROA ROA ROA ROA

DUMMY1=1 DUMMY2=1 DUMMY3=1 DUMMY4=1 DUMMY5=1 INDEX -.0167 .0248 .0429 -.0757 -.0595 (.0235) (.0626) (.0491) (.0793) (.0826) LnMC -.0815** .0078 -.1772*** .0232 -.1482 (.0349) (.0469) (.0448) (.0806) (.1957) LnDEBT -.1312*** -.2239*** .0051 -.0133 .1437 (.0316) (.0447) (.0336) (.0512) (.1027) Standard errors in parentheses

*p<0.1, **p<0.05, ***p<0.01

The completion of the regression analyses showed that the Index has no significant impact in both the time-controlled regression (table 4) as well as the simple general regressions (Table 3).

In Regression 1, an increase in LnMC results in a significant decrease in the ROA of 13% with a value of 0. The LnDEBT and Index variables are not significantly impacting the ROA due to their p-values of 0.256 and 0.723.

Regression 2 shows that including the dummy variables for the different time frames does impact the other variables included in the regression since LnMC and LnDEBT are now significantly decreasing the ROA with 10% and 6,5% respectively. Both of the p-values of these explanatory variables is 0, thus indicating a significance of <1%. The DUMMY1 variable and DUMMY2 variables show to have a significant positive effect on the ROA of 31% and 27% compared to DUMMY4 with values of 0, while DUMMY3 and DUMMY5 do not imply a significant difference due to their p-values of 0.376 and 0.988.

The indexes created for this regression all appear to be insignificant in their contribution to the ROA with p-values ranging from 0.279 to 0.951. Therefore, their importance and values in this regression are negligible.

The third and final set of regressions shows the lnMC can be found with a different impact depending on the time period tested for. The LnDebt, like the previous variable, has no constant positive or constant negative impact on the ROA. Both will, together with the indexes from their respective time-controlled regressions, be layed out appropriately.

Starting with the oldest companies, the LnMC has been shown to have a significant negative effect of -8.15%, while the LnDEBT shows the negative effect of -13.12%. The Index in this group creates a negative effect of 1.67% but is not significant. (p>|t|=-0.71)

With the next group of companies, the LnMC has been shown to have an insignificant positive effect of -.0715%, while the LnDEBT shows the negative effect of -22.39%, although significant. The Index in this group creates a positive effect of 2.48% and is again not significant. (p>|t|=-0.692)

(19)

For the third group of companies, the LnMC has been shown to have a significant negative effect of -17.72%, while the LnDEBT shows the insignificant negative effect of -0.51%. The Index in this group creates a negative effect of 4.29% but is not significant. (p>|t|=-0.382)

Continuing with the fourth group of companies, the LnMC has been shown to have an insignificant positive effect of 2.32%, while the LnDEBT shows the insignificant negative effect of -1.13%. The Index in this group creates a negative effect of 7.57% but is not significant. (p>|t|=-0.341) Ending with the newest group of companies, the LnMC has been shown to have a significant negative effect of -14.82%, while the LnDEBT shows the insignificant positive effect of 14.37%. The Index in this group creates a negative effect of 5.95% but is not significant. (p>|t|=-0.478)

Both tests for heteroskedasticity show no significant sign of heteroskedasticity. Both the Breusch-Pagan / Cook-Weisberg, as well as Cameron and & Trivedi’s decomposition of IM-test, indicate homoscedastic errors with a p-value of 0.

(20)

5

Conclusion

As a result of the multiple regression analyses that have been done concerning the effect of ownership concentration on the return on assets, a concluding answer to this research can now be formed. Since almost every model resulted in an insignificant effect of the created Herfindahl index on the dependent variable is it fair to conclude that the index on its own has no definitive positive or negative effect on the business within the European continent. The logarithmic value of market capitalisation has a negative effect, and the logarithmic value of debt has a positive effect, as concluded by Pilbeam in 2013. The different age groups had mixed effects with the most notable being the largely positive effect of the eldest two categories.

The insignificance effect of the effect can be attributed to multiple factors, ranging from country-specific effects or the large variety of companies present in the tech industry. While this paper aimed to eliminate the country, specific effects showed in the article by opting for a larger

demographic region it could have been too impactful to be eliminated by just expanding the number of countries to the degree seen in this paper. (Krivogorsky, V., & Grudnitski, G. 2010) Other

contributions to this effect might be the diversity in the portfolio chosen for this paper. It could very well be possible that differences between high tech and telecom, for instance, are well noticeable. The findings of time-controlled regressions with solely the indexes, however, show the combined impact of the dummies and the indexes through the time controlled regressions and does show to have both a negative impact on the older company groups as well as a on the youngest groups, whereas a positive impact can be found on the companies founded between 1950 and 2000. This creates the inversed U-shape expected by combining the studies of Grosfeld and Thomsen and Pedersen. The reason behind this could be contributed to the high amount of knowledge necessary to make decisions in the new high-tech industry (Baker, M. P., & Wagonfeld, A. B. 2004), whereas the companies that have a more explored industry are more effectively governed by outside investors. Together this means that the hypothesis was correct but can’t be proven to be significantly right and is thus left open be

investigated by an alternate researcher.

The intelligence gained by this research is minimal, though a lot can be learned from the implications of the results. Opting for a bigger dataset or choosing for a more specific industry might lead to more conclusive results. More variables will need to be added to account for country-specific results, which increases the complexity of the research, but while the effort needed to increase the validity are probably high, it is worth researching in order to significantly increase the effectiveness and efficiency of governing these large corporations.

(21)

6

Reference list

Andres, C., (2008) Large shareholders and firm performance – and empirical examination of founding-family ownership, Journal of corporate finance.

Baker, M. P., & Wagonfeld, A. B. (2004). Dividend Policy at Lineair Technology. Harvard Business

School, 1–18.

Leggett. D. (2015). Volkswagen’s Ferdinand Piëch got a lot right.

Fredrik, J. (2013). Ownership and Performance in Europe. Review of Business, 33(2), 39–55. Grosfeld, I. (2009). Large shareholders and firm value: Are high-tech firms different? Economic Systems, 33(3), 259–277.

Groß, K. (2007). Equity Ownership and Performance An Empirical Study of German Traded

Companies (1st ed. 2007.).

Hamadi, M. (2010). Ownership concentration, family control and performance of firms. European Management Review, 7(2), 116–131.

Jensen, M., & Meckling, W. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3(4), 305–360.

Krivogorsky, V., & Grudnitski, G. (2010). Country-specific institutional effects on ownership: concentration and performance of continental European firms.

Pilbeam, K. (2013) International Finance, Fourth Edition.

Shekshnia, S., & Zagieva, V. (2019). Leading a Board Chairs’ Practices Across Europe.

Thomsen, S., & Pedersen, T. (2000). Ownership structure and economic performance in the largest european companies. Strategic Management Journal, 21(6), 689–705.

(22)

Appendix

Summary of tables

Table 1: variable description (page 15)

Standardised variables Description

Market Capitalisation Total market value of the company in Million Euro

Total liabilities and debt Amount of Outstanding Liabilities and debt in Million Euro

Total Assets Amount of Assets in Million Euro Operating revenue Revenue earned in Million Euro Ownership percentage Amount of percentage owned by each

shareholder

Calculated variables Description

ROA Operating revenue divided by Total Assets

LnMC Natural logarithm of Market Capitalisation

LnDEBT Natural logarithm of Liabilities and Debt

INDEX Herfindahl-Hirschman index for ownership

concentration

DUMMY1 Establishment before 1950

DUMMY2 Establishment between 1950 and 1980

DUMMY3 Establishment between 1980 and 2000

DUMMY4 Establishment between 2000 and 2010

DUMMY5 Establishment between 2010 and 2020

INDEX1 Herfindahl index x DUMMY1

INDEX2 Herfindahl index x DUMMY2

INDEX3 Herfindahl index x DUMMY3

INDEX4 Herfindahl index x DUMMY4

(23)

Table 2: Descriptive statistics (page 17)

Variable

Obs

Mean

Std. Dev.

Min

Max

DUMMY1

2000

.32

.4665928

0

1

DUMMY2

2000

.18

.3842835

0

1

DUMMY3

2000

.355

0.478633

0

1

DUMMY4

2000

.11

.312968

0

1

DUMMY5

2000

.035

.1838257

0

1

INDEX1

2000

.0856472

.2927921

0

1

INDEX2

2000

.055715

.1930892

0

1

INDEX3

2000

.0971823

.2534906

0

1

INDEX4

2000

.0302776

.205714

0

1

INDEX5

2000

.0069799

.1343689

0

1

ROA

1837

.8152329

.4065249

.0009419

2.646

INDEX

2000

.2758019

.4386302

0

1

LnMC

1699

3.333601

.6854847

.7764186

5.170

LnDEBT

1829

2.939701

0.882699

-.2742115

5.095

Table 3: Regression 1&3 outcomes

ROA ROA (1) (3) INDEX .0081 - (.0229) - LnMC -.1312*** -.1216*** (.0241) .0237 LnDEBT -.0216 -.0324* (.01907) (.01894) INDEX1 - .0955*** - (.0302) INDEX2 - .1681*** - (.0469) INDEX3 - -.1549*** - (.0399) INDEX4 - -.2786*** - (.0650) INDEX5 - -.0925 - (.0853)

(24)

Standard errors in parentheses *p<0.1, **p<0.05, ***p<0.01

Table 4: Time-controlled regression outcomes (regression 2)

ROA ROA ROA ROA ROA

DUMMY1=1 DUMMY2=1 DUMMY3=1 DUMMY4=1 DUMMY5=1 INDEX -.0167 .0248 .0429 -.0757 -.0595 (.0235) (.0626) (.0491) (.0793) (.0826) LnMC -.0815** .0078 -.1772*** .0232 -.1482 (.0349) (.0469) (.0448) (.0806) (.1957) LnDEBT -.1312*** -.2239*** .0051 -.0133 .1437 (.0316) (.0447) (.0336) (.0512) (.1027) Standard errors in parentheses

*p<0.1, **p<0.05, ***p<0.01

Table 5: Test for heteroskedasticity

Referenties

GERELATEERDE DOCUMENTEN

Factors such as container capacity and departure times that are fixed in current routing decisions may impede consolidation opportunities for orders arriving at a later time, such

An exogenous fiscal stimulus on the other hand is successful in increasing the share of extrapolating agents while also increasing the level of their expectations, ensuring

Daarbij zijn elf hypotheses getoetst, waarna we kun- nen concluderen dat het interne sociale netwerk via drie factoren een significante positieve in- vloed heeft gehad op

Matematik eğitiminde yapılan deneysel çalışmaların çokluğu, örneklem olarak en fazla öğretmen adaylarının seçimi, duyuşsal boyuttaki çalışmaların sıklığı,

It is of additional value to investigate the moderating effect of IR, because research has not yet found a significant positive or negative effect of integrated reporting

Dependent variables are ROA defined as EBIT scaled by total assets, ROE defined as earnings after tax scaled by shareholder funds and INST is a dummy variable indicating

This finding is consistent with the efficient contracting perspective, suggesting that the higher the CEO reputation the more likely they are to be associated with lower cost of

However, using a sample of 900 firms and controlling for firm size, capital structure, firm value, industry and nation, my empirical analysis finds no significant