The effect of German codetermination on firm value and short-termism.

(1)

U

NIVERSITY OF

A

MSTERDAM

The effect of German codetermination

on firm value and short-termism

Florian Wohnhas

Faculty of Economics and Finance

Supervisor: Jens Martin

Duration: 1. June 2013 — 20. December 2013

Abstract

The German legal system mandates an allocation of control rights to employees, by em-ployee representation on the supervisory board. The presence of emem-ployee representa-tives on the supervisory board is expected to reduce myopia in corporate strategies. This study shows that employee representation can help to increase asset durability and in-vestments in R&D during periods of crises and in the presence of short-term oriented investors. Furthermore, I give evidence that codetermination reduces stock volatility. In contrast to Fauver (2006), I find that both one-third and parity codetermination have a negative effect on firm value.

(2)

I declare that I have developed and written the enclosed thesis completely by myself, and have not used sources or means without declaration in the text.

Amsterdam, 2013-12-20

. . . . (Florian Wohnhas)

(3)

1. Introduction

The purpose of corporate governance varies across countries. Whereas in the US and UK, corporate governance is seen as a mechanism to protect shareholder interests, in countries such as Germany, France and Japan corporate governance has a much broader mission. Good corporate governance in these countries requires operating in the interest of all stakeholders. In Germany, in particular the broader mandate of corporate governance is quiet clear, since the legal system explicitly requires companies to act also in the interest of their employees due to codetermination. The allocation of control rights to employees, by employee representation on the supervisory board, allows them to influence corporate policies.

Previous research has mainly studied the effect of employee participation in corporate governance on firm value [Gorton and Schmid (2004), Fauver and Fuerst (2006), Renaud (2007), Petry (2009) and Bermig and Frick (2010)]. Neither from a theoretical point of view nor from the empirical evidence these studies present it is clear whether codetermination adds value. This study with its comprehensive data set of approximately 300 listed Ger-man firms over the years 1998-2007 provides further empirical evidence to this question1. Therefore, I hypothesize that employee representation on the supervisory board increases firm value, due to better information exchange processes between management and em-ployees, monitoring capabilities and long-term orientation of employee representatives. The main issue this study focuses on is how employee participation in corporate gover-nance influences the temporal orientation in corporate strategies. I argue that employee participation in the corporate decision-making process reduces myopic management be-havior. Myopic behavior can be caused by (i) stock market pressure to report short-term profits, (ii) compensation schemes that are tied to short-term performance or (iii) man-agers’ fear of not being re-elected for another term. The reasons for fluctuations in earn-ings cannot always be fully understood by investors, due to imperfect information. In the case of information asymmetry between management and investors, those short-term managerial objectives incentivize managers to try to inflate short-term earnings by bor-rowing from future earnings through underinvestment (Stein, 1988) or through earnings management (Brochet, Loumioti, and Serafeim, 2012). Narayanan (1985) argues that a focus on short-term profits can also help to increase the reputation of a manager at the be-ginning of his tenure and consequently boost his compensation. It can also help increase their chance of re-election. Another way to foster reputation and thus another form of

(5)

2

myopia is hyperactive management behavior. It can be characterized by corporate strate-gies that are based on restructuring, financial re-engineering or mergers and acquisitions instead of a focus on the development and improvement of the long-term business fun-damentals (Kay, 2012). Consequently, the effect of codetermination on short-termism is analyzed in this paper by the three characteristics of myopic behavior described above: (1) underinvestment in tangible and intangible assets, (2) earnings management and (3) a hyperactive management behavior. As a result, the following three sub-hypotheses can be formulated. First, long-term orientation and monitoring capabilities of employee representatives on the supervisory board should reduce the underinvestment problem in tangible and intangible assets. Second, long-term orientation and monitoring capabilities of employee representatives on the supervisory board should reduce earnings manage-ment. Third, long-term orientation and monitoring capabilities of employee representa-tives on the supervisory board should reduce hyperactive management behavior.

Empirical evidence for the postulated long-term orientation of employees is provided by a study of Rhein (2010). He examined in 2008 an average tenure of employees in Ger-many of 10.8 years. This exceeds by far the average CEO tenure in GerGer-many of 3.9 years reported by Wulf, Miksche, Roleder, and Stubner (2010) in their study of the 83 largest publicly listed German companies in the period between 1990 and 2007. Consequently, a longer temporal orientation is assumed for employee representatives, following the ar-gument of Jensen and Meckling (1979), who state that the temporal horizon considered in the decision-making process for employees and management, is closely related to their own tenure, since their claims on the firm usually end with the last day of their employ-ment2. Hence, the longer average tenure of employees in comparison to the management should lead to a reduction of myopic management behavior.

The existing studies on the effect of codetermination on corporate governance mainly focused on the influence of employee representation on firm value, while this paper pri-marily analyzes the influence on the temporal orientation in corporate strategies. Refer-ring to the effect of codetermination on firm value, the studies of Gorton and Schmid (2004), Fauver and Fuerst (2006) and Bermig and Frick (2010) are mostly similar. In their study Gorton and Schmid (2004) analyze valuation differences between firms with one-third codetermination and parity codetermination. In a resilience check in section 6 a similar regression analysis is performed and similar results are obtained. In particular, I examine a valuation discount for firms with parity codetermination in relation to firms with one-third codetermination. Fauver and Fuerst (2006) show that codetermination can increase firm value if the level of employee representation stays below the threshold of 50%. Whereas Bermig and Frick (2010) cannot find statistically significant and econom-ically relevant effects of codetermination on firm value. This study, however, provides evidence that both, one-third and parity codetermination reduce firm value.

Regarding the effect of codetermination on short-termism I find that codetermination significantly reduces stock volatility. Albeit, the robustness check suggests as optimal a level of employee representation below the 50% threshold, parity codetermination still reduces volatility. Furthermore, I present evidence that firms, which actively increase the percentage of employee representation, experience lower volatility in the future. The longer temporal orientation of employee representatives is expected to reduce the ten-dency of following short-term market trends and therefore to reduce volatility. Contrary to the expectation the regression analyzes show that parity codetermination significantly increases earnings management. Moreover, firms with parity codetermination have a sig-nificantly higher involvement in earnings management than firms with one-third code-termination.

(6)

3

The vital debate in Germany about the current form of the codetermination legislation between the labor and capital side makes this research particularly relevant. The case of the in 2005 appointed governmental committe on moderninzing German codetermi-nation legislation shows, that there is clearly observable dissension between different interest groups. The committee, consisting of representatives from academia, employees and employers, was not able to agree on a common position regarding the further devel-opment of German codetermination legislation. While representatives of academia and labor examined no need for change, the representatives of the employers demanded a reduction of codetermination for large corporations to one-third codetermination. Con-sequently, the ongoing debates between the different interest groups have not led to any changes. Now, the newly elected German government plans to extend the already exist-ing codetermination legislation. From 2016 onwards 30% of all members of codetermined supervisory boards shall be female. This implies dramatic changes in the structure of su-pervisory boards as the average susu-pervisory board of this dataset comprises of approx-imately 8% female members. Lastly, the expansion of the codetermination legislation in France in June this year makes this research particularly timely.

The rest of the paper is organized as follows. In the first part of section 2 the charac-teristics of the German corporate governance system and the legal framework behind codetermination are discussed. In the second part of section 2 the existing theoretical and empirical literature on codetermination is reviewed and the hypotheses are devel-oped. Section 3 presents the data sources, a summary of the descriptive statistics and a univariate regression on the difference between firms with and without codetermined supervisory board. Section 4 outlines the research design. The results of the multivari-ate regressions are presented in section 5. Section 6 contains a robustness check and an extension to the regression analysis in section 5 and section 7 contains a conclusion.

(7)

2. Economic setting

2.1. Germany’s corporate governance system and legal framework

In this section, I will discuss the characteristics of the German corporate governance system and the legal framework behind employee representation as part of the corporate governance system. The German corporate governance system is characterized by a two-tier board system and the influence of banks, ownership concentration and codetermination in corporate governance. The two-tier board consists of the board of management (BoM) and the supervisory board. The BoM is responsible for the operating business and reports to the supervisory board. The supervisory board appoints and removes the members of the BoM. Moreover, it monitors the BoM and possesses rights of information and audit from the BoM. It is also involved in setting long-term strategies for a company and owns the right to approve business decisions.

The influence of banks originates from their important role as lenders, since in terms of external financing bank credits are far more important than bonds or equity in Germany (Siebert, 2004). In their role as owners banks, exercise control in corporate governance by their proxy voting rights, equity holdings and representation on the supervisory board. The benefits of bank representatives on the supervisory board in terms of corporate gov-ernance include the reduction of managerial agency costs through their monitoring capa-bilities, which can lead to an increase in firm performance (Gorton and Schmid, 2000). Ownership concentration may have the same, or even stronger, influence on corporate governance as bank representation as observed by Stiglitz (1985) and Edwards and Nibler (2000). Ownership concentration can be also seen as a countermeasure of the owners in the supervisory board against the influence of the employee representation (Roe, 2013). Furthermore, ownership concentration over 25% of the voting equity also provides the block holder with a veto right in the shareholders’ general meeting, which enables them to block decisions about changes in capital structure and mergers and acquisitions. Employee representation on the supervisory board in Germany is regulated by law. The current codetermination legislation evolved out of four different laws.

(8)

2.1. Germany’s corporate governance system and legal framework 5

The first law, the so called Montan-Mitbestimmungsgesetz was passed in May 21, 1951. It requires parity employee representation on the supervisory board for all companies of the mining, coal and steel industry with more than 1,000 employees and of the le-gal form Aktiengesellschaft (AG), Gesellschaft mit beschränkter Haftung (GmbH) and bergrechtliche Gewerkschaft. The board composition is also legally prescribed for those companies and has to consist of eleven members. Five from the capital side, five from the labor side and one additional neutral member.

On October 11, 1952 the law Betriebsverfassungsgesetz was passed, which now aimes at all other industries. It mandated one-third employee representation for all companies of the legal form AG and Kommanditgesellschaft auf Aktien (KGaA) without any threshold concerning the number of employees, except for family-owned companies. They are only required to have a one-third codetermined board if they have more than 500 employees. Companies of the legal form GmbH, Versicherungsverein and Genossenschaft with more than 500 employees are also obliged to have a one-third codetermined board.

The law Mitbestimmungsgesetz from May 4, 1976 constitutes an extension to the Be-triebsverfassungsgesetz. It mandated parity employee representation for all companies of the legal form AG, KGaA, GmbH, Versicherungsverein and Genossenschaft with more than 2,000 employees. In the case of stalemate, the double voting right of the supervisory board’s president still allows the capital side to outvote the labor side. The law required the following board composition: (1) the supervisory board of companies with less than 10,000 employees has to consist of 12 members, six from the capital and six from the labor side. (2) Companies with more than 10,000, but less than 20,000 employees, are required to have a supervisory board with 16 members, with eight from each side. (3) The law stipulates that the supervisory board for companies with more than 20.000 employees has to consist of 20 members, ten from the capital and ten from the labor side.

The law Drittelmitbestimmungsgesetz from May 27, 2004 replaced the Betriebsverfas-sungsgesetz. This led to only minor changes in the original legislation. Now companies of the legal form AG and KGaA that were founded after August 10, 1994 are only required to have a one-third codetermined board if they have more than 500 employees.

(9)

2.2. Literature review 6

2.2. Literature review

In this section, I will review the existing theoretical and empirical literature on codetermination. Property rights theory, agency cost theory and participation theory provide the theoret-ical framework that explains, whether employee participation in corporate governance adds value.

Property rights and agency cost theory analyze institutional contracts, which lead to a separation of ownership and control. Property rights theory tries to explain how those contracts determine the use of resources to maximize wealth and the allocation of income among the participants in an organization (Cheung, 1970). Agency cost theory focuses on the costs that arise through the delegation of authority about resources from the owner – the so called principal – to another person, who acts on behalf of the owner, the so called agent (Jensen and Meckling, 1976). In those relationships costs incur to the principal in the form of: (1) costs to monitor the agent and (2) residual losses, which are costs in the form of lost welfare. They occur because the agent might not make decisions that maximize the wealth of the principal. Not only the principal, but also the agent faces costs. Those are mainly bonding costs, as investments in firm-specific human capital (Jensen and Meckling, 1976).

The following arguments against the efficiency of employee participation can derive from the theories mentioned above. Consistent with property rights theory, employee rep-resentatives could favor investments that increase their payroll instead of firm value. Furthermore, they could also avoid investments, for example in new technology, which would increase welfare, but would decrease payrolls or reduce jobs. In addition, they could lobby for an income distribution in favor or the employees instead of the owners. Through the implementation of codetermination, monitoring costs occur due to conflict-ing interests of owners and employee representatives, as agency cost theory suggest. Residual losses may arise through the resistance for restructuring, since restructuring events usually involve layoffs and wage reductions. In addition, they prefer to retain profits rather than paying dividends or share repurchases to minimize the probability of financial distress or default, since both also usually lead to layoffs and salary cuts. Additionally they also might prefer a level of diversification above the optimum for the firm value to decrease the probability of default (Gorton and Schmid, 2004). Other cor-porate policies that reduce the overall risk, such as lower leverage and less risky projects, to avoid events of financial distress or default, might be promoted even if they do not maximize firm value (Berk, Stanton, and Zechner, 2010).

The arguments outlined above imply a direct or indirect wealth transfer from owners to employees. Furthermore, employee representation may also complicate and delay the decision-making process, since it might introduce a voting problem due to conflicting in-terests (Jensen and Meckling, 1979). This is especially true under parity codetermination. Another argument against the efficiency of codetermination is the fact that it is never observed except when it is mandated by law (Jensen and Meckling, 1979). This means that voluntary employee representation has never been introduced.

Participation theory, which analyses the information exchange process, provides the the-oretical support for employee representation. Freeman and Lazear (1994) argue that codetermination leads to a better information transfer from employer to worker. The reduction of information asymmetry between management and workers helps to avoid inefficient economic outcomes. Particularly in bad times, this is very important, since it can help to reduce the probability of work halts and strikes and to agree on concessions (Furubotn and Wiggins, 1984). Moreover, workers can come up with new solutions for existing problems within the firm that management overlooked.

(10)

Codetermination integrates employees in the corporate decision-making process. This creates incentives for workers to share operational knowledge with the management and therefore improves information transfer from workers to management (Freeman and Lazear, 1994). The more information employees set forth, which vary from management, the higher are the possible benefits of employee representation.

Employee representation increases job security for workers. This, in combination with their integration in the decision-making process, helps to align the interest of employers and workers, and as a result fosters loyalty among the employees and can increase mo-tivation (Freeman and Lazear, 1994). It also creates incentives for employees to invest in firm-specific human capital (Furubotn and Wiggins, 1984). The investment of employees in firm-specific human capital can increase productivity and firm value. This is especially true in industries where highly qualified employees are needed due to a complex labor component in production(Fauver and Fuerst, 2006).

Another important benefit of employee representation can be its monitoring capabilities. Acharya, Myers, and Rajan (2011) find that internal governance through non-executive employees can reduce agency problems by preventing managers from shirking and tak-ing perks.

Both theoretical streams also provide insight about the impact of codetermination on tem-poral orientation on corporate decisions. Jensen and Meckling (1979) argue that claims of employees on the incremental earnings stream of the firm do not go beyond their own tenure with the firm. Therefore, the horizon they consider in their decisions is the end of their own employment relationship. In Germany, the average tenure for employees in the 2008 was 10.8 years according to a study of Rhein (2010). Wulf, Miksche, Roleder, and Stubner (2010) report in their study of the 83 of the largest publicly listed German companies in the period between 1990 and 2007 an average CEO tenure of 3.9 years, which is a little bit lower than the average European CEO tenure of 5.7 years examined by Karlsson, Neilson, and Webster (2008). These findings suggest that the time horizon considered in the decision-making process by employees should be longer than the one of the management and therefore can lead to a focus on more long-term decisions. Freeman and Lazear (1994) argue that employees have a more long-term perspective on corporate decisions in a stakeholder governed firm, since employee representation in-creases job security. From the theoretical literature alone it is not possible to conclude, whether codetermination adds value or which temporal perspective in corporate deci-sions is fostered by codetermination.

The following empirical studies analyze the impact of employee representation in cor-porate governance on firm profitability, productivity and firm value. Benelli, Loderer, and Lys (1987) find no evidence that codetermination reduces profitability, affects div-idend payments or company policies. In contrast, Gorton and Schmid (2000) show in their study, with observations on 250 companies from 1989-1993, that parity codetermi-nation relative to one-third codetermicodetermi-nation significantly decreases the return on assets, the market-to-book ratio and the return on equity. They draw the conclusion that under parity codetermination the employee representatives influence the corporate decisions in a way that reduces firm value. Fauver and Fuerst (2006) examine that codetermina-tion can increase firm efficiency and market value, if the level of employee representacodetermina-tion stays below the threshold of 50%. They also point out that those industries, which require more intense coordination, profit more from employee representation. Their dataset con-sists of 786 companies in the year 2002. In his study, Renaud (2007) analyzes, whether there have been differences in productivity and profitability between companies with one-third codetermination and parity codetermination immediately after the Codetermi-nation Act in 1976 and whether those differences have remained or changed in the long

(11)

run. He reports that the introduction of parity codetermination has had a positive ef-fect on productivity and profitability. The long-term impact on productivity remains the same, while the impact on profitability increases. Wagner (2009) observes no significant impact on performance for limited liability companies through one-third codetermina-tion, measured by productivity and profitability. His dataset includes 273 companies between 2005 and 2007. Petry (2009) performs two event studies to measure the wealth effect of codetermination. The first event study analyzes the effect of changes in labor rep-resentation due to the imposition of codetermination in 1976. This event study comprises observations on 140 companies in 1976. He shows that companies, which were forced to increase employee representation from one-third codetermination to parity codeter-mination due to the Codetercodeter-mination Act of 1976, experienced negative announcement returns, but only if five or more additional employee representatives took up board seats. In his second event study, he analyzes company announcements about changes in board size and changes in the percentage of employee representation on the supervisory board between 1998 and 2008. He reports a large negative wealth effect, if employee represen-tatives join a board, which previously was not codetermined. He also observes a positive effect of an almost equal size, if all labor representatives leave the board. Bermig and Frick (2010) presents similar results as Wagner (2009) for 294 publicly traded companies between 1998 and 2007. They conclude that board size and board composition – code-termination – has no statistically significant and economically relevant effects on firm performance.

The theoretical arguments and empirical evidence mentioned above can be summarized in the following hypothesis about the impact of codetermination on firm value:

H1: Codetermination should have a positive impact on firm value.

The following empirical studies provide insights on the impact of employee representa-tion in corporate governance on firm’s temporal orientarepresenta-tion.

In his study about monthly return variances before and after the implementation of the Codetermination Act in 1976, Benelli, Loderer, and Lys (1987) find no evidence that return variances for securities that are subject to codetermination are lower. Gorton and Schmid (2004) examine that in three out of five observed years (1989-1993), codetermination leads to a significantly higher standard deviation of weekly stock returns. They argue that this is the result of risk shifting to capital because employee representatives resist, for example, restructuring events. Fauver and Fuerst (2006) show that codetermination leads to a lower capital expenditure to sales and R&D to sales ratio and explain that this is due to the fact that employee representatives intervene against poor managerial investment decisions. Kraft, Stank, and Dewenter (2011) analyze the impact of the Codetermination Act in 1976 on innovativeness. Their measure for innovativeness is the number of patents granted to a particular company, and find that the imposition of codetermination has no negative effect.

(12)

The theoretical arguments and empirical evidence mentioned above can be summarized in the following hypothesis about the impact of codetermination on temporal orientation in the corporate decisions:

H2: Codetermination should reduce short-termism in corporate decisions. The following table shows the derived sub-hypotheses:

(13)

2.2. Literature review 10 Author (s) Year Title Research question Data Economic model Results

Allen, Franklin and Gale, Douglas

2002

A comparative theory of corporate governance Can stakeholder capitalism be superior when markets are not perfect and complete

-Partial equilibrium models

Stakeholder capitalism can often be superior, when markets are not perfect and complete. An infexible labor market provides incentives for cooperation. They shown that cutting dividends and maintaining wages as well as the employment level can be an optimal response. Codetermination can lead to less dividend payments, but more long-term orientation.

Benelli, Giuseppe and Loderer, Claudio and Lys, Thomas

1987

Labor Participation in Corporate Policy-Making Decisions: West Germany's Experience with Codetermination Effect of codetermination rights on corporate operations and performance 58 companies 1973-1983 Market model regression and F-test No evidence that return variance for security that are subject to codetermination are lower. Evidence that codetermination reduces profitability, affects dividend payments or companies policie is very weak and rarely statistically significant.

Bermig, Andreas and Frick, Bernd

2010

Board size, board composition and firm performance: Empirical evidence from Germany Effects of supervisory board size and composition on firm value and performance 294 companies 1998-2007 2,382 observations Series of fixed-effects regressions to estimate the effect of board size and composition on firm performance Board size and board composition (codetermination) does not have a statistically significant and economically relevant effect on firm performance measued by return on equity and return on invested capital. Only the percentage of former managing board members and the percentage of works council representatives seem to have a significant negative, yet economical weak impact.

Fauver, Larry and Fuerst, Michael E.

2006

Does good corporate governance include employee representation? Evidence from German corporate boards Does employee representation on a corporate board increase firm value? 786 companies 2002 Series of OLS regressions to estimate the effect of codetermination on firm value measured by Tobin's Q and logit regressions to estimate the effect of codetermination on dividend payouts. Robustnes cecks with logit model

Employee representation on corporate boards can increase firm efficiency and market value (optimal representation below 50%). U-shaped relation between codetermination and firm value. Industries that require more intense coordination profit more from employee representation.

Ginglinger, Edith and Megginson, William and Waxin, Timothe

2011

Employee ownership, board representation, and corporate financial policies Analysis of the impact of employee- directors on corporate valuation, payout policy, and internal board organization and performance

201 companies 1998-2008 Series of OLS and firm fixed-effect regressions to estimate the effect of codetermination on firm value measured by Tobin's Q and RoA Directors elected by employee shareholders increase firm valuation and profitability, but do not significantly impact corporate payout policy. Directors elected by employees by right significantly reduce payout ratios, but do not impact firm value or profitability (France).

Figure 2.3.: Overview existing studies on codetermination

(14)

2.2. Literature review 11 Author (s) Year Title Research question Data Economic model Results

Gorton, Gary and Schmid, Frank

2000

Class struggle inside the firm: a study of German codetermination Effect of supervisory board composition on firm performance 250 companies 1989-1993 902 observations Series of OLS regressions and semi-parametric locally weighted regressions to estimate the effect of codetermination on performance measured by the market-to-book ratio (MTB), the return on assets (ROA), and the return on equity (ROE).

Employees on the supervisory board significantly decrease the return on assets, the market-to-book ratio and the return on equity.

Kraft, Kornelius and Stank, Joerg and Dewenter, Ralf

2011

Co-determination and innovation

Effect of the German Codetermination Act of 1976 on innovativness. Innovativness is measured by the number of patents 148 companies 1971-76 and 1981-90 Count data models with number of patents as dependent variable

Codetermination has no negative impact on innovativeness.

Petry, Stefan

2009

The wealth effects of labor representation on the board-evidence from German codetermination legislation Effect of codetermination on shareholder wealth 140 companies 1976 & 59 companies 1998-2008 Event study to analyze the effect of changes in employee represenation after the Codetermination Act in 1976 and event study to analyze company announcements about changes in board size and the number of employee representatives on the board (1998- 2008) Companies which were forced to increase the percentage of employee representatives on the board from 1/3 to 1/2 due to the Codetermination Act of 1976 experienced negative announcement returns but only if 5 or more additional employee representatives took up board seats. There is a large negative wealth effects if employee representatives join a board that previously was not codetermined. Positive effect of almost equal size if employee representatives leave a board that will not be codetermined anymore in the future.

Renaud, Simon

2007

Dynamic efficiency of supervisory board codetermination in Germany Comparison of differences in profitability and productivity between companies with 1/3 employee representation to companies with 1/2 employee representation

1970-2000

OLS regressions

The introduction of parity codetermination has a positive effect on productivity and profitability. There is no long-term effect on productivity through the introduction of 1/2 codetermination, but profitability is positive affected.

Wagner, Joachim

2009

One-third codetermination at company supervisory boards and firm performance in German manufacturing industries: First direct evidence from a new type of enterprise data Effect of codetermination on productivity and profitability. Productivity measured by value added per employee and profitability measured by return on sales

273 companies 2005-2007 Series of OLS regressions to estimate the effect of codetermination on profitability measured by return on sales and productivity measured by value added.

No positive or negative effect of codetermination on firm performance measured by productivity and profitability was observed.

Figure 2.4.: Overview existing studies on codetermination

(15)

3. Data

3.1. Data sources

In this section, I will list the data sources of my empirical study.

This study comprises data from about 300 public German companies that were listed for at least one year during the period 1998-2007. The data on supervisory board compo-sition was obtained from annual reports. The data collected includes the independent variables: supervisory board size, codetermination, 1/3 employee representation, 1/2 employee representation, independent employee representatives, union representatives, work council representatives and bank representatives. Data on accounting performance, number of employees, business and geographic segments was obtained from Thomson Reuters Worldscope. Accounting data, and data on the number of employees that was not available in Worldscope, was collected from Hoppenstedt’s Balance Sheet Database. Data on market capitalization and common shares outstanding was obtained from Datas-tream. The data mentioned above, allowed the construction of the following variables: Sales growth, log. total assets, MV equity, RoS, RoA, capital intensity, Q, leverage, divi-dend payment, industry diversification, geographical diversification and the dependent variables Tobin’s Q, annual stock volatility, asset durability, investment rate, earnings management indicator, R& D intensity, restructuring indicator and layoff indicator. Data on the annual trading volume of common shares was drawn from Yahoo Finance. This data, together with the data on common shares outstanding, were used to derive the control variable short investor horizon. The availability of the data for all variables men-tioned above is for the period 1998-2007. Data on ownership concentration is only avail-able from 2001-2007 and was obtained from the BAFIN database. This implies the control variables ownership 0-10%, ownership 10-30% and ownership >30%.

(16)

3.2. Summary of univariate results 13

3.2. Summary of univariate results

In this section, I will present a summary of the descriptive statistics and the results of a t-test on mean differences between firms with and without employee representation on the supervisory board.

Table 3.1 presents a summary of the descriptive statistics for the variables mentioned above. The mean (median) supervisory board consists of 9.94 (9.00) members and the mean (median) percentage of employee representation is 30% (33%). Thereof, the mean (median) percentage of union representatives is 7% (0%) and 9% (0%) for work coun-cil representatives. This means that 14% (8%) of the board members are independent employee representatives. In this sample, approximately 7% of the firms have bank rep-resentatives on the supervisory board.

Table 3.1.:Descriptive statistics

Firm characteristics N Mean Median Std. Dev. Min. Max.

Log. total assets 2354 13.57 13.22 2.45 3.83 21.42

MV equity 2363 3516.22 298.08 10453.75 0.00 135635.30 Sales growth 2342 1.23 0.06 47.37 -6.65 2290.55 RoS 2224 0.12 0.11 0.69 -20.47 11.99 RoA 2227 0.15 0.12 2.14 -5.78 100.70 Capital intensity 2276 0.10 0.04 0.89 -1.23 40.46 Q 1968 1.60 1.25 1.25 0.20 23.78 BV/MV 2336 0.83 0.85 0.31 0.02 5.05 Leverage 2345 0.27 0.24 0.21 0.00 1.71 Dividend payment 2316 0.71 1.00 0.45 0.00 1.00 Investor horizon 1174 0.57 0.20 0.96 0.00 13.93 Industry diversification 2157 0.82 1.00 0.39 0.00 1.00 Geographical diversification 1719 0.87 1.00 0.33 0.00 1.00 Ownership 0-10% 939 0.09 0.10 0.02 0.00 0.10 Ownership 10-30% 939 0.16 0.20 0.07 0.00 0.20 Ownership >30% 939 0.26 0.23 0.30 0.00 6.40 Tobin’s Q 2336 1.52 1.17 1.48 0.20 42.67 Investment rate 2279 0.06 0.04 0.07 0.00 1.00 Asset durability 2155 219.16 6.03 4772.40 0.00 163868.45

Asset durability (win 95%) 2155 7.47 6.03 6.13 0.95 35.59

RaD intensity 915 0.03 0.02 0.04 0.00 0.27

Layoff indicator 2326 0.37 0.00 0.48 0.00 1.00

Earnings management 2156 0.07 0.00 0.26 0.00 1.00

Annual stock volatility 1784 29.17 26.76 10.80 7.43 78.95

Restructuring indicator 1239 0.31 0.00 0.46 0.00 1.00

Size of supervisory board 2381 9.94 9.00 5.94 3.00 21.00

Employee repr. 2381 0.30 0.33 0.22 0.00 0.50

Bank representatives 2381 0.07 0.00 0.11 0.00 1.00

Union representatives 2381 0.07 0.00 0.08 0.00 0.30

Work council repr. 2381 0.09 0.00 0.12 0.00 0.35

Independent employee repr. 2381 0.14 0.08 0.14 0.00 0.50

The following figures illustrate the trending development of employee representation during the period under observation. From the 2,381 firm-year observations, 1,630 (68%) are codetermined. Thereof 560 (24%) have one-third employee representation and 1,070 (45%) have parity employee representation. On the other hand there are 504 (21%)

(17)

firm-3.2. Summary of univariate results 14

year observations with more than 500 and less than 2,000 employees and 1,414 (59%) firm-year observations with more than 2,000 employees. Regarding the one-third em-ployee representation, this is in line with the expectations, since the number of firm-year observations with more than 500 and less than 2,000 employees represents only the min-imum of all possible outcomes. In this case the fact is neglected that non-family-owned companies with less than 500 employees, which were founded before August 10, 1994 are still required having a one-third codetermined supervisory board. In the case of parity codetermination the number of observed cases with parity employee representation is lower than mandated by law. The special thresholds for companies in the mining, coal and steel industry can be neglected, since only two firms of the entire sample belong to these industries. The mean percentage of employee representation for firms with more than 500, and less than 2,000 employees, in this sample is 21%, and 42% for companies with more than 2,000 employees. Both numbers are lower than the corresponding per-centages – 33% and 50% – mandated by law. Figure 3.1 also shows that there is a trend to less employee representation. The percentage of firms with employee representation decreased from 77% in 1998 to 66% in 2007. While the percentage of firms with one-third codetermination remained rather constant, the percentage of firms with parity codeter-mination decreased from 52% in 1998 to 42% in 2007. The big drop occured in the period from 1998 to 2000. In contrast to this decrease in relative terms, is an increase in abso-lute numbers. From 1998 to 2001, the number of firms with employee representation increased from 157 to 179, which means, that those firms that were listed in this period, have a lower percentage of employee representation. Another interesting observation is that the implementation of the Codetermination Act in 2004 did not lead to significant decrease in the percentage of firms with one-third codetermination.

Figure 3.1.:Trend employee representation

Statistics on whether firms without employee representation differ from firms with em-ployee representation are presented in Table 3.2. The first column of the table contains the means and medians of the different firm characteristics for the firms without employee representation and the second the ones for the firms with employee representation. The differences in the means in absolute numbers and the results of a Student t-test on the sig-nificance of these differences are reported in the third column. The last column presents the means and medians for the entire sample.

(18)

As seen in Table 3.2, firms with employee representation are significantly larger than firms without employee representation in terms of log. total assets. There is no signifi-cant difference in profitability, measured by RoS, between companies with and without codetermination. The average capital intensity for companies with employee representa-tion is lower than the one for companies without employee representarepresenta-tion. This can be explained by the fact that through the monitoring capabilities, and the additional infor-mation from employee representatives, poor investment decisions by the management can be avoided. In contrast to Roe (2013), in this sample the ownership concentration of firms with employee representation is significantly lower than the one of firms without employee representation. He argues that ownership concentration is a countermeasure against the influence of employee representatives on the supervisory board and there-fore expects higher ownership concentration on codetermined firms. The mean lever-age ratio of companies with codetermination is lower, than the one for companies with-out codetermination. Firms with employee representation are more diversified, both in terms of industries and geographically. These findings are consistent with the idea that employee representatives favor less leverage, and a higher diversification, to lower the probability of financial distress or default. They provide support for the idea of Gorton and Schmid (2004), who argue that codetermination leads to higher diversification. Fur-thermore, the annual stock volatility of codetermined companies is significantly lower, than the one of companies without codetermination. Table 3.2 also shows that companies with employee representation pay significantly more dividends than companies with-out employee representation. This contradicts to property rights theory, which suggests that employee representatives would prefer to retain profits instead of paying dividends. Table 3.2 also presents evidence that companies with employee representatives on the su-pervisory board invest more in R&D. Surprising are the findings regarding layoffs, since the univariate analysis shows that firms with codetermined supervisory boards experi-ence significantly more layoffs than firms without codetermined supervisory boards. The result that firms with employee representation undergo more frequently restructurings is also unexpected. With regards to earnings management codetermined firms involve more often in earnings management, yet the difference does not reach the level of sta-tistical significance. In addition Tobin’s Q of firms with codetermination is significantly lower, than for firms without codetermination.

To conclude, firms with employee representation seem to be larger, less capital intense, more diversified, have lower leverage, lower volatility, invest more in R&D, experience more layoffs and restructuring, pay more dividends and have a lower Tobin’s Q.

(19)

Table 3.2.:Descriptive statistics by codetermination

Firm characteristics No employee Employee Difference Total

repr. mean repr. mean mean mean

(median) (median) (median)

Log. total assets 11.79 14.37 -2.58∗∗∗ 13.57

(11.72) (13.95) (13.22) MV equity 402.32 4933.20 -4530.90∗∗∗ 3516.22 (81.13) (535.81) (298.08) Sales growth 3.77 0.09 3.68∗ 1.23 (0.08) (0.05) (0.06) RoS 0.10 0.13 -0.03 0.12 (0.13) (0.11) (0.11) RoA 0.21 0.13 0.08 0.15 (0.11) (0.12) (0.12) Capital intensity 0.16 0.06 0.10∗∗ 0.10 (0.03) (0.04) (0.04) Q 1.92 1.49 0.43∗∗∗ 1.60 (1.44) (1.22) (1.25) Leverage 0.31 0.25 0.06∗∗∗ 0.27 (0.28) (0.23) (0.24) Dividend payment 0.51 0.80 -0.29∗∗∗ 0.71 (1.00) (1.00) (1.00) Investor horizon 0.36 0.66 -0.31∗∗∗ 0.57 (0.17) (0.22) (0.20) Industry diversification 0.68 0.87 -0.19∗∗∗ 0.82 (1.00) (1.00) (1.00) Geographical diversification 0.77 0.90 -0.13∗∗∗ 0.87 (1.00) (1.00) (1.00) Ownership 0-10% 0.10 0.09 0.01∗∗∗ 0.09 (0.10) (0.10) (0.10) Ownership 10-30% 0.19 0.16 0.03∗∗∗ 0.16 (0.20) (0.20) (0.20) Ownership >30% 0.31 0.24 0.08∗∗∗ 0.26 (0.27) (0.22) (0.23) Tobin’s Q 1.87 1.37 0.50∗∗∗ 1.52 (1.30) (1.13) (1.17) Investment rate 0.06 0.06 0.00 0.06 (0.03) (0.05) (0.04)

Asset durability (win 95%) 7.80 7.32 0.47∗ 7.47

(4.97) (6.38) (6.03) R& D intensity 0.03 0.04 -0.01∗∗ 0.03 (0.01) (0.02) (0.02) Layoff indicator 0.30 0.40 0.10∗∗∗ 0.37 (0.00) (0.00) (0.00) Earnings management 0.06 0.08 -0.02 0.07 (0.00) (0.00) (0.00)

Annual stock volatility 0.37 0.27 0.10∗∗∗ 0.29

(0.35) (0.25) (0.27)

Restructuring indicator 0.16 0.34 -0.18∗∗∗ 0.31

(0.00) (0.00) (0.00)

(20)

4. Economic model

In this section, I will explain the choice of the variables used in the different economic models as well as the econometric models themselves.

The empirical model is based on the panel data set explained in section 3. A fixed-effect model is used to study the effect of employee representation on firm value and short-termism, since a correlation of firm-specific effects with the independent variables is as-sumed; management skills influence firm size, and therefore, employee representation. As regressors for board size and composition are rather constant over the years a least square dummy variable model (LSDV) with year dummies to examine time-specific ef-fects is applied rather than a within estimator model. The model also includes industry dummies to control for industry-specific effects. Moreover, it is augmented with control variables to capture firm-specific effects.

The choice of the dependent variables as measures for firm value and short-termism are explained in the following.

Firm value is measured by Tobin’s Q following Fauver and Fuerst (2006). Tobin’s Q is defined as market value of equity plus book value of total assets minus book value of eq-uity all divided by book value of total assets. To measure the effect of codetermination on firm value a regression similar to the one of Fauver and Fuerst (2006) is employed. In ad-dition to supervisory board size and composition it also allows to control for size (natural logarithm of total assets), profitability (RoS), capital intensity, capital structure (leverage), diversification (industrial and geographical), dividend payments, periods of crises, short investor horizon and ownership concentration. Furthermore, a series of interaction terms is included.

It can be summarized in the following general economic model:

Tobin0s Qi,t=α0+β0Board sizei,t+β1Codeterminationi,t+β2Bank rep.i,t

+β3Log. total assetsi,t+β4RoSi,t+β5Capital intensityi,t+β6Leveragei,t

+β7Industr. div.i,t+β8Geograph. div.i,t+β9Dividend paymenti,t

+β10Period o f crisesi,t+β11Employee rep. × crisesi,t

+β12Own <10%i,t+β13Own 10−30%i,t+β14Own >30%i,t

+β15Employee rep. × Own <10%i,t+β16Employee rep. × Own 10−30%i,t

(21)

18

Short-termism in corporate strategies can be characterized by three phenomena. The first one is the tendency to underinvestment in both tangible and intangible assets in the case of information asymmetry between management and investors. The second one is earn-ings management, which can also be observed in the case of imperfect information. The third one is hyperactive management behavior. This is characterized by a management approach that tries to create value by restructuring, financial re-engineering or merg-ers and acquisitions instead of developing and improving business fundamentals (Kay 2012). To ensure the tenacity of the empirical analysis several variables are always used to test underinvestment in tangible assets, intangible assets, earnings management and hyperactive management behavior.

Underinvestment in physical assets can be analyzed on the basis of the investment rate as demonstrated by Harford and Kolasinski (2012). The investment rate is defined as the ratio of CAPEX divided by total assets. A low ratio is associated with underinvestment. For this test, an investment-cash flow sensitivity regression similar to the one of Fazzari, Hubbard, and Petersen (1988) is used. This multivariate model allows to control for size (natural logarithm of total assets), current and future growth opportunities (sales growth and Q), cash flow sensitivity (RoA), periods of crises, short investor horizon and own-ership concentration. In addition the model is augmented with a series of interaction terms.

This investment-cash flow sensitivity regression can be summarized in the following gen-eral economic model:

Investment rate_i,t =α0+β0Board sizei,t+β1Codeterminationi,t+β2Bank rep.i,t

+β3Log. total assetsi,t+β4Sales growthi,t+β5RoAi,t+β6Qi,t

+β7Period o f crisesi,t+β8Employee rep. × crisesi,t

+β9Short investor horizoni,t+β10Employee rep. × short horizoni,t

+β11Own <10%i,t+β12Own 10−30%i,t

+β13Own >30%+i,t+β14Employee rep. × Own <10%i,t

+β15Employee rep. × Own 10−30%i,t

+β16Employee rep. × Own >30%i,t+ei,t

Another way to measure underinvestment in tangible assets is the durability of the ac-quired property, plant and equipment assets (PPE) as suggested by Souder and Bromiley (2012). They point out that those investment decisions reflect the management’s temporal orientation, since these activities lead to immediate costs in exchange for future profits. Firms with a shorter temporal orientation are therefore expected to undertake projects that create profits in the near-term and consequently acquire PPE with a shorter expected useful life even if this might not be the optimal investment decision based on net present value method (NPV) or internal rate of return (IRR). The dependent variable asset dura-bility is defined as PPE divided by depreciation. For this analysis the same economic model as for the investment rate is used.

Investments in R&D are used to test for underinvestment in intangible assets following Bushee (1998). He observes that companies with a shorter investment horizon are more likely to cut R&D expenditures to increase short-term earnings. The findings of Brochet, Loumioti, and Serafeim (2012) also support the idea that R&D expenditures can be used as an indicator for temporal orientation. They show that short-term oriented firms report lower R&D expenditures in years when they report small positive earnings or just meet the analysts’ forecasts. In contrast, Laverty (1996) argues that R&D expenditures are a

(22)

19

questionable indicator for long-term investments because they also include many short-term projects. A series of investment-cash flow sensitivity regressions identical to the one mentioned above with R&D intensity – R&D expenditures divided by sales – as a dependent variable are employed for this analysis.

Investments in intangible assets also comprises investments in human capital, as em-ployee skills. One form of investments in human capital is thus the protection of firm-specific human capital. Long-term oriented firms are assumed to maintain their work-force even in economic shocks and not to lose the firm-specific human capital of their workers. Consequently a layoff indicator can be used as a measure for underinvestment in intangible assets. Layoffs are defined as negative changes in the number of person-nel from one year to another. In the case of layoffs the indicator assumes a value of one and zero otherwise. To estimate the effect of employee representation on layoffs a similar investment-cash flow sensitivity model as the one described above is used.

The second phenomenon of short-termism examined in this study is earnings ment. Brochet, Loumioti, and Serafeim (2012) present evidence that earnings manage-ment is associated with short-termism. Consequently a proxy for earnings managemanage-ment is included in this study, which tries to capture earnings management activities associ-ated with attaining the previous year’s earnings. Degeorge, Patel, and Zeckhauser (1999) show in their study about behavior thresholds for earnings management that attaining previous year’s earnings is the second most important phenomenon, when it comes to earnings management after reporting positive earnings and even before meeting ana-lysts’ expectations. Similar to Brochet, Loumioti, and Serafeim (2012), earnings manage-ment is assumed if the ratio EBITDAtminus EBITDAt−1divided by market capitalization

falls in the range from zero to 0.01. Furthermore, the following histogram with the em-pirical distribution of this dataset around the threshold of meeting or beating last year’s earnings also presents a large jump in this interval. Hence, the proxy for earnings man-agement takes a value of one if the ration falls into this interval and zero otherwise.

(23)

20 19 22 22 24 35 37 45 57 76 100 174 168170 124 102 72 65 40 40 43 0 50 100 150 200 Frequency −.1 −.05 0 .05 .1 Sustainable earnings

Figure 4.1.:Histogram of earnings management

Note: Figure 4.1 shows the histogram around the threshold of earnings sustainability. Earnings sustainability is defined as EBITDAtminus EBITDAt−1divided by market

cap-italization. The distribution interval widths are 0.01 and the vertical axis shows the fre-quency of distributions in each interval.

To test the effect of employee representation on earnings management a regression model similar to the one of Yu (2008) and Jones (1991) is used. In addition to the characteristics of the supervisory board, it consists of the following control variables: size (MV equity), current growth (sales growth), future growth opportunities (BV/MV), earnings sensitiv-ity (RoA), proxy for volatilsensitiv-ity of earnings (annual stock volatilsensitiv-ity), ownership concentra-tion, periods of crises and short investor horizon. Again, a series of interaction terms is included in the model.

This economic model can be summarized by the following formula:

Earnings managementi,t =α0+β0Board sizei,t+β1Codeterminationi,t+β2Bank rep.i,t

+β3MV equityi,t+β4Sales growthi,t+β5RoAi,t+β6BV/MVi,t

+β7Annual stock volatilityi,t+β8Period o f crisesi,t

+β9Employee rep. × crisesi,t+β10Short investor horizoni,t

+β11Employee rep. × short horizoni,t+β12Own <10%i,t

+β13Own 10−30%i,t+β14Own >30%+i,t

+β15Employee rep. × Own <10%i,t

+β16Employee rep. × Own 10−30%i,t

(24)

21

As already mentioned one form of hyperactive management behavior is frequent restruc-turing. Hence, an indicator for restructuring is used to test for short-termism. The re-structuring indicator assumes a value of one if a company reports rere-structuring expenses and zero otherwise. Again, the investment-cash flow sensitivity model is used for this analysis.

Hyperactive management behavior is also assumed to increase the volatility of business. Brochet, Loumioti, and Serafeim (2012) also provide theoretical support that stock price volatility can be used as an indicator of short-termism. They argue that short-term ori-ented firms show a higher volatility, since the shorter time horizon considered in their corporate decisions makes them follow short-term market fluctuations, and therefore, they experience a higher volatility in their stock returns. Consequently, annual stock volatility is used as an indicator for short-termism in a similar economic model as the one for Tobin’s Q.

Based on the characteristics of the German corporate governance system as discussed in section 2, independent variables for the size and composition of the supervisory board are employed in all economic models. The following binary variables are used to measure the effect of employee representatives on firm value and short-termism: codetermination, 1/3 employee representation, 1/2 employee representation, independent employee rep-resentatives, union representatives and work council representatives. The independent variable supervisory size is incorporated, since Yermack (1996) presents evidence that smaller boards lead to a higher firm value. Banks and their representatives on the super-visory board are considered long-term investors (Porter, 1992). Gorton and Schmid (2000) also show that they improve firm value due to their monitoring capabilities and their in-fluence on the cost of external financing. To measure their effect on temporal orientation and firm value the binary variable bank representatives is used.

The following control variables are employed in the three different economic models, since previous studies have documented their importance.

Troch (2009) shows that in Germany the existence of a supervisory board significantly depends on the number of employees. In order to be able to distinguish the impact of codetermination from that of firm size on firm value and short-termism the control vari-ables natural logarithm of total assets and market value of equity are used. To control for the disciplinary effect of debt, the leverage ratio is used (Jensen, 1986). Another firm char-acteristic that can influence firm value as well as temporal orientation is capital intensity, measured by the ratio CAPEX over sales. Berger and Ofek (1995) finds that overinvest-ment, measured by capital intensity, leads to a decrease in firm value. The variables sales growth and Q allow to control for current and future growth opportunities. Sales growth is defined as the percentage change of sales from one year to another. Q is defined as market value of equity plus total liabilities divided by total assets minus goodwill. In the context of earnings management future growth opportunities are controlled for via the ratio BV/MV, book value divided by market value, following Yu (2008). RoS is used in the regression on firm value and volatility to control for different levels of profitability. RoS is defined as EBITDA divided by sales. RoA is defined as EBITDA divided by total assets. The variable RoA is used in the investment-cash flow sensitivity regressions as a proxy for cash flow sensitivity.

To control for industrial and geographical diversification, two binary variables were con-structed based on the definition of Fauver and Fuerst (2006), since they report both, a significant industrial and geographical diversification discount for German firms. A com-pany is defined as industrial diversified if less than 90% of the comcom-pany’s sales stem from one four-digit SIC segment. The industrial diversification indicator assumes a value of one if the company is diversified and zero otherwise. A company is defined geographical

(25)

22

diversified if less than 90% of the company’s sales stem from one geographical segment. The geographical diversification indicator assumes a value of one if the company is geo-graphically diversified and zero otherwise.

As already discussed in section 2, ownership concentration is an important feature of the German corporate governance system. Gorton and Schmid (2000) find evidence that the performance of German firms improve due to ownership concentration. Fauver and Fuerst (2006) also report that equity ownership above 30% increases firm value. Owner-ship concentration is also assumed to affect the temporal orientation since block holders are expected to be long-term investors. Following Fauver and Fuerst (2006) three bins of continuous variables to control for ownership concentration were constructed. The percentages of equity ownership are allocated to the three variables ownership 0-10%, ownership 10-30% and ownership >30% as follows: (1) ownership 0-10%, contains equity ownership between zero and 10%, and therefore, assumes values between zero and 10%, (2) ownership 10-30%, contains equity ownership between 10% and 30%, and therefore, assumes values between zero and 20% and (3) ownership >30%, contains equity owner-ship between 30% and 100%, and therefore, assumes values between zero and 70%. Ownership concentration imposes agency problems, in particular the problem of expro-priation of outside shareholders by controlling shareholders (Faccio, Lang, and Young, 2001). A good example thus would be pyramiding. Dividend payments can be seen as a signaling effect to investors against insider expropriation, since they limit the potential of expropriation. Consequently dividend payments can help to avoid an undervaluation of the company and to reduce agency costs (Lopez de Silanes, Vishny, and Shleifer, 2000). The set of control variables, therefore, also includes an indicator for dividend payments. The dividend indicator assumes a value of one if in the current year a dividend has been paid and zero otherwise.

Derrien, Thesmar, and Kecskes (2011) document that management’s investment deci-sions, are influenced by the temporal orientation of their investors, in the case that the firm is mispriced. They show that CAPEX increases as firms are undervalued and the amount of long-term oriented investors increases. The variable investor horizon is, there-fore, included in the model to control the influence of the investor’s temporal orientation. Following Polk and Sapienza (2009), investor horizon is measured by the share turnover. Share turnover is defined as the annual trading volume of the common shares divided by common shares outstanding. A short investor horizon is assumed for all observa-tions above the median share turnover of 0.20. Consequently, the control variable short-investor horizon assumes a value of one if the share turnover is higher than 0.20 and zero otherwise.

A crisis indicator is also added to the set of control variables. It assumes a value of one for the years 2000-2003, the period of the dotcom crisis, and zero otherwise.

(26)

5. Discussion

The empirical analysis begins with an investigation of the impact of employee representation on firm value. The second part constitutes of the empirical analysis of the effect of employee represen-tation on short-termism. Additional graphs and tabulations to the discussion below can be found in the appendix.

5.1. Regression results of employee representation on firm value

The univariate results presented in section 3 suggest a negative impact of codetermina-tion on firm value, measured by Tobin’s Q. As discussed in seccodetermina-tion 4, the factors firm size, profitability, capital intensity, capital structure, industrial and geographical diversi-fication, dividend payments, crises and ownership structure are assumed to affect firm value. To be able to differentiate between the impact of those factors and codetermination on firm value a multivariate regression model is needed. Table 5.1 presents the results of three multivariate regression models on Tobin’s Q and codetermination. The first model of Table 5.1 includes, in addition to the measures for board size and board composition, the control variable firm size (natural logarithm of total assets), profitability (RoS), capi-tal intensity, capicapi-tal structure (leverage ratio) and industrial and geographical diversifica-tion. In this model a small negative, but statistically insignificant effect of the supervisory board size and the codetermination indicator on firm value can be observed. The second model is augmented by a dividend and crises indicator as well as an interaction term of the codetermination indicator and the crises indicator. The results for both the supervi-sory board size and the codetermination indicator are almost identical to the ones of the first model. In the third model the influence of ownership concentration is tested. Besides measures of ownership concentration, a series of interaction terms between measures of ownership concentration and the codetermination indicator are introduced, since own-ership concentration is assumed to reduce the influence of employee representatives on the supervisory board. Again, the coefficients of both regressors are negative, but now the codetermination indicator reaches the level of statistical significance. Summarizing the results of the three models, supervisory board size has a consistent negative, yet sta-tistically insignificant effect on firm value. The negative impact on firm value is in line with the findings of Yermack (1996). The regression analysis shows that codetermination is negatively related with firm value. However, this effect reaches only in the last model the level of statistical significance. This applies also for the one-third and parity employee representation indicator. Whereas the coefficient of the one-third codetermination indica-tor is consistently negative, the effect of parity codetermination is not clear, since the sign

(27)

5.1. Regression results of employee representation on firm value 24

of the coefficient varies throughout the different models. This contradicts the findings of Fauver and Fuerst (2006), who report an increase in firm value, if employee repre-sentation stays under the threshold of 50%. The observation that parity codetermination reduces firm value is in line with the findings of Fauver and Fuerst (2006) and Gorton and Schmid (2004). Regarding the effect of the different types of employee representatives a consistent effect can only be observed for the independent employee representative in-dicator. In the third model there is evidence that independent employee representatives increase firm value. This is not surprising since independent employee representatives usually work in the firm and therefore are able to improve the information exchange process between management and workers. In contrast to Gorton and Schmid (2000), who show that bank representatives improve firm value, this analysis shows that bank representatives on the supervisory board have a negative, yet statistically insignificant effect on Tobin’s Q. In all three models one can examine a weak evidence that a higher profitability, measured by RoS, increases firm value. In addition, there is also some weak evidence that capital intensity decreases firm value in the first two models. In contrast to Jensen (1986), who argues that the disciplinary effect of debt should increase firm value, I find that leverage decreases firm value. Furthermore, no economically relevant indus-trial or geographical diversification discount as reported by Fauver and Fuerst (2006) can be observed. The positive, yet statistical insignificant coefficient of the crises indicator is striking. When the crises indicator interacts with the employee representation indicator a consistently negative effect on firm value can be observed. In the last model this effect be-comes significant. This effect can be explained by the fact that employee representatives may engage in risk shifting towards capital during crises. Combining the results of the three models, the codetermination and the one-third codetermination indicator seem to have consistent negative, yet economically weak impact on firm value. Moreover, Tobin’s Q seems to be higher for firms with a higher margin, lower leverage and lower capital intensity.

Table 5.1:Dependent variable: Tobin’s Q

Note: Table 5.1 presents the results of three fixed-effect regression models of Tobin’s Q on several board, firm and ownership characteristics as well as several other indicator variables. Those models also include year and industry dummies. Tobin’s Q is defined as market value of equity plus book value of total assets minus book value of equity all divided by book value of total assets. The board characteristics include the size of the supervisory board, an indicator for codetermination and an indicator for bank rep-resentation on the supervisory board. The indicator codetermination assumes a value of one if the supervisory board is codetermined and zero otherwise. The indicator for bank representation assumes a value of one if there is at least one bank representative on the supervisory board and zero otherwise. The firm characteristics controlled for include the natural logarithm of total assets, RoS, capital intensity and the leverage ratio. RoS is defined as EBITDA divided by sales and capital intensity is defined as CAPEX divided by sales. The leverage ratio is defined as debt over total assets. Moreover, indicators for industrial and geographical diversification are included. Both indicators assume a value of one if the firm is diversified and zero otherwise. A firm is considered industrial di-versified if less than 90% of the company’s sales stem from one four-digit SIC-segment. A firm is considered geographically diversified if less than 90% of the company’s sales stem from one geographical segment. The dividend indicator assumes a value of one if a dividend has been paid in the current year and zero otherwise. The crises indica-tor assumes a value of one for the years 2000-2003, the period of the dotcom crisis, and zero otherwise. The ownership characteristics controlled for are the three bins of con-tinuous variables ownership 0-10%, ownership 10-30% and ownership >30%. They are constructed in the following way: (1) ownership 0-10%, contains equity ownership be-tween zero and 10% and therefore assumes values bebe-tween zero and 10%, (2) ownership

(28)

5.1. Regression results of employee representation on firm value 25

10-30%, contains equity ownership between 10% and 30% and therefore assumes values between zero and 20% and (3) ownership >30%, contains equity ownership between 30% and 100% and therefore assumes values between zero and 70%. The sample includes about 300 public German companies that were listed for at least one year during the pe-riod 1998-2007. Heteroskedasticity-consistent White (1980) t-values are in parentheses. *, **, and *** imply statistical significance at the 0.10, 0.05, and 0.01 levels, respectively.

(1) (2) (3)

Size of supervisory board -0.015 -0.014 -0.022

(-0.59) (-0.53) (-0.88)

Codetermination -0.092 -0.053 -0.859∗∗∗

(-0.45) (-0.32) (-4.41)

Bank representatives -0.556 -0.562 -1.035

(-0.85) (-0.84) (-1.11)

Log. total assets -0.052 -0.058 0.038

(-0.85) (-0.83) (0.56) RoS 1.284∗ 1.285∗ 2.533∗ (2.10) (2.06) (2.28) Capital intensity -0.218∗∗ -0.215∗ -0.319 (-2.32) (-2.25) (-1.38) Leverage -1.664∗∗ -1.647∗∗ -1.280∗∗ (-2.98) (-3.33) (-2.55) Industry diversification 0.111 0.123 0.014 (1.32) (1.36) (0.37) Geographical diversification -0.088 -0.102 0.201 (-0.31) (-0.37) (0.59) Dividend payment 0.079 -0.058 (0.47) (-0.59) Period of crises 0.059 0.101 (0.22) (1.29)

Employee repr. x crises -0.236 -0.192∗∗

(-0.94) (-2.66) Ownership 0-10% -5.341 (-1.65) Ownership 10-30% -2.391 (-1.33) Ownership >30% -0.138 (-0.48)

Employee repr. x ownership 0-10% 5.326

(1.39)

Employee repr. x ownership 10-30% 3.390

(1.17)

Employee repr. x ownership >30% -0.284

(-0.39)

Constant 2.989∗∗∗ 2.950∗∗∗ 2.774∗∗∗

(4.79) (4.69) (4.68)

Year dummies Yes Yes Yes

Industry dummies Yes Yes Yes

R2 _0.225 _0.229 _0.342

Observations 1573 1550 692

t statistics in parentheses