Does size matter? : an empirical analysis of the influence of the firm-size on the wages it pays its employees, and the effect industry-characteristics have upon this relationship

(1)

University of Amsterdam

Amsterdam Business School

Does Size Matter?

An Empirical Analysis of the Influence of the Firm-Size on the Wages it pays its

Employees, and the Effect Industry-Characteristics have upon this Relationship

Master Thesis

Supervisor: Dr. Ilir Haxhi

2

nd

_{Reader: Dr. Niccollò Pisani}

Torben Krauss

10825045

(2)

Statement of originality

This document is written by Student Torben Krauss who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

1

Abstract

Despite the broad variety of literature on the positive effect the firm-size has on the wages paid, the research conducted fails to provide consistent results on the existence of the effect. In addition to that, the studies cannot be compared internationally due to non-consistent datasets and -types. We argue that the implications of a firm’s size as e.g. increasing monitoring costs, churn rates, etc. positively influence an employee’s wage and that the firm-size concentration, productivity, and capital intensity of an industry positively moderate this relationship, due to the known direct impact each factor has on the size of a company and the wages paid. In this thesis we examine the firm-size wage effect on a multinational and national level for Belgium, Czech Republic, Finland, Germany, the Netherlands, and Sweden. For our cross-sectional analysis of the firm-size wage effect, we employ the WageIndicator dataset with matched industry data from the

Eurostat and STAN OECD databases, which allows us to compare the considered effects across

borders and to identify the influence of firm-size concentration, productivity, and capital intensity onto the firm-size wage effect. We find that the firm-size positively impacts the wages paid to a firm’s employees and that the size of the effect differs between the considered countries. The productivity of an industry positively moderates the firm-size wage effect as opposed to the average capital intensity. An industry’s capital intensity negatively moderates the effect. The firm-size concentration in an industry facilitates the relationship between firm-size and wages depending on the size of the share of big companies in an industry. Additionally to contributing to the academic discussion concerning the firm-size wage effect, we provide valuable information for employees and employers to understand the factors influencing their wages to improve their bargaining situation.

(4)

2

Table of

Table of Figures

Figure 1 - Framework ... 30

Table 1 - Number of datasets ... 31

Table 2 - Model Configuration ... 35

Table 4 - Wage Means of Groups ... 39

Table 5 - Correlation Matrix ... 40

Table 6 - VIF ... 42

Table 7 - Regression Analysis ... 48

Table 8 - Moderator Regression Analysis ... 49

The WageIndicator data used in this master thesis have been made available by the WageIndicator Foundation, and have been used with their permission. The Foundation bears no responsibility for the analyses or interpretation of the data reported here.

(6)

4

1 Introduction

In the standard labor market model a worker’s marginal productivity determines the equilibrium wage. Accordingly, two individuals with identical characteristics earn the same wages (Du Caju et al., 2011). The differences in wages between employees result from differences in their characteristics (Dickens & Katz, 1986). Contrary to these theoretical assumptions, various studies identify significant deviations of the wage level of workers that exhibit similar experiences, education levels, etc. (Du Caju et al., 2011; Garciano & Rossi-Hansberg, 2006; Amiti & Davis, 2011; Pedace, 2010; Brown et al., 1990). In addition to a worker’s characteristics, factors as the industry he works in (Du Caju et al., 2011) or employer characteristics influence the wage level (Scoppa, 2014). One of the observed firm characteristics influencing an individual’s wage is its size (Pedace, 2010; Smyth & Gao, 2011). The so-called ‘firm-size wage effect’ describes the influence a firm’s size has on the wages it pays its employees (Ferrer & Lluis, 2008; Pedace, 2010; Scoppa, 2014). This factor cannot exist if the theory of perfect competition applies (Benito, 2000); accordingly, it is a phenomenon of great interest to study for understanding the mechanisms influencing the wage bargaining between employee and employer.

Despite the broad variety of literature concerning the firm-size wage effect, the quantitative studies performed do not have consistent outcomes. Smyth and Gao (2011) find that the impact of the firm-size on the wage level is negative in China; in Italy, Brunello and Colussi (1998) do not find evidence for a significant relation, and in the USA the relation is positive (Pedace, 2010). Additionally, previous studies usually have been unable to compare the differences of the effect between countries due to a lack of accessible data. They also did not analyze the influence industry-characteristics have on the relationship. The existing literature regarding industry-industry-characteristics suggests a positive impact of the concentration, productivity, and capital intensity of an industry

(7)

5

onto the size of a firm within an industry and the wages it pays (Abowd et al., 1999; Long & Link, 1983; Van Biesebroeck, 2005; Du Caju et al., 2011; Hamermesh, 1980; Idson and Oi, 1999), but has not investigated the effect the characteristics have onto the relationship between firm-size and wages paid yet.

After the extensive examination by researchers the questions remain unanswered how the firm-size of a company influences the wages it pays to its employees, how the effect differs between different countries, and how the relationship between firm-size and wages is influenced by an industry’s characteristics.

To answer these questions we analyze the firm-size effect on a multinational level and on a national level. This allows us to determine the existence of the effect and to compare the results between countries. We hypothesize a positive impact of the firm-size onto the wages paid based on the size implications regarding monitoring costs, churn rates, etc.. Bigger companies have to pay higher wages to compensate for increasing monitoring costs and/or reduce the churn rate of their employees. We also argue that the industry-characteristics introduced above moderate the firm-size wage effect positively due to the positive direct effect they have on the firm-size and wages paid within an industry. With this analysis we provide a quantitative foundation to derive potential drivers of the firm-size wage effect from it.

We design a framework in this master thesis, to conduct the analysis and to test the hypotheses we formulate. We employ the firm-size as the independent variable, the wage of the individual employee of a company as dependent variable, and the average capital intensity, productivity, and firm concentration of the firm’s industry as moderating variable. In addition, we compare the effects depending on the country of work of the employee for Belgium, Czech Republic, Finland, Germany, the Netherlands and Sweden. To conduct the analysis, we use

(8)

cross-6

sectional data from WageIndicator on the level of an individual working in the countries mentioned above, containing information on the wage, employer size, and worker’s characteristics. This data was matched with industry data from the Eurostat and the STAN OECD database.

The results of our analysis show a significant firm-size wage effect on a multinational level and in each country individually. The effect size varies between the observed countries, suggesting that country characteristics are important drivers of its occurrence. The moderating effects of the industry-characteristics vary. The average productivity of an industry positively moderates the influence of the firm-size on the individual’s wage. In addition, a high concentration of big companies within an industry facilitates the effect. The average capital intensity in an industry negatively moderates the firm-size wage effect. The higher the capital intensity, the smaller the firm-size wage effect.

The results contribute to the existing academic research by quantitatively validating the existence of the firm-size wage effect. In addition to that, it shows that the country of work and the industry of a company influence the effect, potentially explaining differences in the results of previous researchers. We also provide valuable information for employees and employers due to the fact that they need to consider the firm-size when bargaining wages.

This thesis provides an overview of the status quo of the literature in section two, followed by an introduction to the framework employed in section 3. The data and methodology we use to test the hypotheses are presented in chapter 4. Chapter 5 provides the results we receive from the analysis. We discuss the results in section 6, followed by the conclusion in chapter 7.

(9)

7

2 Literature Review

In the following chapter, we provide the theoretical background to understand the framework employed in our study. In the first part, we present general factors that influence the wages paid and why the firm-size wage effect is considered a phenomenon. In the second part, we introduce the literature regarding the relationship between a firm’s size and its effect on the wages the firm pays to its employees. This includes the status quo of the quantitative analyses conducted and of the explanatory approaches. In the third part, we provide an overview of the industry’s influence on the wages and the firm.

2.1 Background

According to the labor market’s standard Walrasian model the marginal productivity of a worker determines the equilibrium wage; therefore, two agents exhibiting similar characteristics receive identical wages (Du Caju et al., 2011). Following this argumentation, the wage differences between workers result from differences between workers’ characteristics (Dickens & Katz 1986). High skilled workers earn higher wages, compared to lower skilled workers (Troske 1999). In addition to the skill level of employees gender, ethnicity, age, or productivity may have a significant influence upon the wage of an individual (Ours & Stoeldraijer, 2011; Nopo et al., 2010; Troske, 1999).

Opposing to the theoretical Walrasian framework and in addition to the wage differences resulting from different characteristics of the individuals, many studies conducted find significant differences between wages of workers having similar levels of education, experience etc. (Amiti and Davis, 2011; Du Caju et al., 2011; Pedace, 2010; Brown et al., 1990; Garciano & Rossi-Hansberg, 2006). The results of the existing research suggest that there are other drivers of wage

(10)

8

differentials than characteristics of the individual or the job (Troske 1999). One observed factor is the size of the firm that drives wage differences. The standard theoretical frameworks cannot explain this factor. According to Troske (1999), unobserved characteristics of big firms in comparison to smaller firms account for this effect. Furthermore, there are observed wage differences for the same jobs between industries, which cannot exist if the competitive theory would apply (Benito 2000). The literature studies a broad variety of factors potentially accounting for these phenomena.

2.2 The Firm-Size Wage Effect

The size of an enterprise is positively related with the wages it pays. The literature refers to this phenomenon as firm-size wage effect, size-wage effect or size-wage premium, which is established by a variance of studies as a significant positive relationship between the size of the labor-force and workers’ wages. Brown et al. (1990) estimate that employees of companies employing above 500 workers pay 35% higher wages compared to companies employing less than 500 workers. Therefore, its impact exceeds the effects of union status and race, and is comparable to the gender-wage-gap. Bhorat and Lundall (2004) distinguish between the impact of the size-wage effect depending on the occupation of the employees. For a study conducted about this effect in South-Africa, they find that managers and professionals in big firms earn 25% higher wages on average compared to their colleagues in smaller firms. When including blue collar employees the total effect decreases to 15% on average. Opposing to the findings by the studies introduced above, Brunello and Colussi (1998) do not find a firm-size premium at Italian firms that significantly differs from zero after controlling for a potential selectivity bias. Contrary to other studies, the premium concerning employees’ wages results from variations in selection effects and observed characteristics. Depending on the selection model employed, the significance of the effect varies.

(11)

9

In his study published in 2014, Scoppa obtains results that are contrary to Brunello and Colussi (1998). He determines a significant size-wage effect that has about the same impact as unmeasured individual abilities of the workers in Italy. Scoppa’s (2014) analysis differs from Brunello and Colussi (1998) in the sample size and the type of data. The data sample he uses for his analysis of Italian companies is bigger, and is longitudinal as opposed to the cross-sectional analysis by Brunello and Colussi.

Pedace’s study (2010) shows an average of approximately nine percent as earnings premium in large firms. Other effects as total revenue of a firm or capital per worker do not correlate significantly with it. However, the firm-size wage effect decreases if including non-wage benefits and training. The research conducted by Söderbom et al. (2005) shows a substantial effect after controlling for effects by individuals. They determine a difference of the impact between developed and developing countries. According to them, companies from countries that are less developed exhibit a stronger size-wage effect than companies from more developed countries.

The value of the size-wage premium differs depending on the study conducted; the variables and the structure within these studies may account for the observed variations of the premium. Albaek et al. (1998) were the first researchers using a continuous measure when considering the establishment size. They considered the exact number of employees of each company as opposed to the discrete measures of previous studies. Therefore, the results obtained are of high interest, especially since they found significant size-wage effects after controlling for characteristics of the individual and the job.

Smyth and Gao (2011) find a contrary effect in China. The relation between the size of Chinese companies and the wages they pay is negative. The bigger the company, the lower the average wages paid. They explain this phenomenon by the increasing share of blue collar workers

(12)

10

within bigger companies compared to small companies. As opposed to Albaek et al. (1998), they divided the company samples provided within two discrete groups: large companies above 85 and small companies below 85 employees. Fox’s (2009) findings are ins support of this explanatory approach. The analysis of Swedish firms shows an increase of wage differences between smaller and bigger firms depending on the position and the responsibility an employee has. Fox finds a significant difference of the wage gap when distinguishing between blue and white collar workers.

The characteristics of employers and employees that drive the wages are correlated with the firm-size the literature argues. These characteristics include the sorting and job matching, capital-labor complementarities, unionization, efficiency wages, internal labor markets, rent sharing, and market power.

2.2.1 Sorting and Job Matching

Garicano and Rossi-Hansberg (2006) determine the way companies with a higher number of employees organize the work as a factor that attracts workers that are more productive. These may have a higher set of skills or acquire knowledge more quickly, which allows them to perform a broader variety of tasks. Due to their characteristics, they tend to have a higher degree of motivation and orientation within a work-organization containing a low degree of hierarchy. The results of the research conducted by Söderbom et al. (2005) contradict these assumptions. Opposing to Garicano and Rossi-Hansberg (2006), they find that the employment of high-ability individuals does not explain the size-wage effect. Garciano and Rossi-Hansberg (2006) also distinguish between two types of information technology (IT) that impact the wage distribution within companies. The first aspect is costs of communication. If the costs of communication decrease, the employees tend to increasingly rely upon “problem solver” for non-routine challenges; therefore, many problems are solved on the top of the hierarchy. This effect decreases the wage differences

(13)

11

between workers, but increases the wage differences between top management and workers. The second aspect of IT is the decrease of accessing-costs for knowledge. A reduction in knowledge accessing-costs allows more low level employees to retrieve information necessary to solve problems. This may lead to an increase in wage inequality within certain layers of organization and the economy as a whole. In the goods-producing sector the difference in wages may arise from the organization of work processes and the increased efficiency due to that. The work is organized around a team, high-productive workers are recruited, retained and trained, and higher effort standards are established. Troske (1999) also identifies improved monitoring mechanisms and screening techniques of large firms compared to smaller firms as a reason for their ability to offer higher fringe benefits, better education and training programs, and higher wages. Accordingly, employers of higher quality are matched with employees of higher quality. The research conducted by Evans and Leighton (1989) determines the difficulty to monitor their employees as a reason for higher wages only in the largest firms. Rebitzer (1993) points out the difficulty of empirically testing the impact of monitoring on the size-wage effect, but interprets the differences between the correlation of the size-wage effect and jobs from primary and the secondary sector as potential indicator for it. A study conducted by Ferrer and Lluis (2008) about companies in Canada shows that large companies do not compensate high performance of employees as well as medium-sized companies do due to the problem of monitoring the productivity of their employees. It is less costly for medium-sized companies to monitor their employees compared to large companies. Therefore, the ability-sorting share of the size-wage effect mainly accounts for the differences between small and medium sized companies, but not between small and large. As Rebitzer (1993) points out, it is complicated to measure the monitoring costs of companies; therefore, Brown and Medoff (1989) observe wage differences between easy to monitor workers: piece-rate workers. Their efficiency

(14)

12

can be measured by the number of products worked on. Brown and Medoff find that the size-wage effect is still significant; therefore, they are skeptical for the monitoring explanatory approach.

In their paper published in 1991 Mayo and Murray determine employment risk as potential factor. Small firms have a higher failure probability, which leads to the attraction of less stable employees. They argue that the size of a company works as a proxy for heterogeneity in firms and workers, which is unobservable otherwise. It reflects a firm’s employee commitment and its risk of failure.

2.2.2 Capital Labor Complementarities

An additional approach to explain the size-wage effect is provided by Hamermesh (1980). He argues that labor and capital are production complements. Bigger companies tend to be more capital-intensive compared to smaller companies, and therefore may be more productive with similar workers. Not the divergence in human capital, but differences in equipment result in productivity differences. More capital allows the implementation of high-quality equipment and a more efficient work-flow. The size of the company may also be favorable for the acquisition of more capital. The bigger a company is the more favorable the credit terms (Idson & Oi, 1999). Troske (1999) determines the degree of the effect as accounting for approximately 45% of the premium concerning size of the firm and wages. A study conducted by Arai (2003) indicates the total impact on wages by the capital-labor structure of bigger companies as approximately 25%.

2.2.3 Efficiency Wages

Shapiro and Stiglitz (1984) acknowledge the increasing difficulty to observe the productivity of an individual within a firm in regard of the company size as important factor for the size-wage effect. Monitoring employees within large firms requires a high capital intensity; to reduce turnover and shirking, they offer above-market-average-wages. The intention behind this

(15)

13

measure is to increase the unemployment opportunity costs. The reduced churn of employees due to the higher wages also decreases the costs for training and hiring and therefore offsets the higher costs for wages. Pedace (2010) adds that higher wages increase the demand for the jobs offered, enabling the employer to demand more effort from the employees. Shapiro and Stiglitz (1984) acknowledge that in addition to efficiency wages, the danger of unemployment can be seen as a device to discipline the employees. It motivates the workers to perform on a productivity level at all times that minimizes the danger of losing the job.

2.2.4 Unionization

Concerning unionization, Voos (1983) emphasizes the intuition of large companies to offer higher wages to reduce the unions’ influence. Earning above average wages decreases the demand of the workforce to participate in unionization activities. Since bigger companies suffer more from a unionized workforce due to their size, they are more likely to increase the wages to decrease the incentives for the workers to join a union. This may result in a size-wage effect for non-unionized workers, but no effect on unionized workers. However, Brown and Medoff (1989) find that size is also related to the wages of employees that are organized within a union. They conclude that union avoidance efforts may impact many actions taken by companies, but not the volume of the wages. Zappalà (1994) acknowledges spillover effects by unions in economies with a large manufacturing sector, but denies the effect for economies with large service-sectors. He also points out that testing the impact of the ‘union threat’ on companies’ wages is difficult.

2.2.5 Internal labor markets

Large companies require employees to acquire certain job specializations and job-specific skills. To stay successful, a company needs to keep these employees for long time periods. The internal labor markets of big companies are designed to evaluate and reward productive workers

(16)

14

(Doeringer & Piore, 1971). The higher an employee is evaluated, the more money a company is willing to spend to keep him. Due to the existence of these internal labor practices, valued employees within bigger companies may earn higher wages than employees of smaller companies. Bigger companies tend to provide more training for their employees compared to smaller companies; therefore, their skill-level increases and the mechanisms described above increase the wages (Pedace, 2010). Internal labor markets also allow the employees to switch between occupations within companies. The larger a company, the more possibilities exist to switch to if an employee is unhappy with his assigned tasks. This leads to a decrease in the employee churn rate, and an increase in years spend at a company. Companies increase wages depending on the maturity of an employee at the company. Therefore, the lower the churn rate through internal labor markets, the higher the wages paid (Brown & Medoff, 1989)

2.2.6 Rent Sharing and Market Power

A different effect is discussed by Gimble (1991). He argues that firms dominating certain product markets may also exhibit monopsic labor market characteristics providing them with wage setting power. Concerning this effect, firms would tend to pay smaller wages, the bigger their size. The study conducted by Idson and Oi (1999) does not support this effect. They determine a higher productivity as determinant for higher wages. Egger and Kreickemeier (2009) find a strong correlation between the perception by the employees concerning the fairness of their wages and a firm’s productivity.

The strength of the effects varies depending on the type of industry considered. In the hospitality industry (García-Pozo et al., 2012) productivity measures and working conditions account for a significant part of the size-wage effect. On the other hand, labor quality factors and internal labor markets account only for a minor part; differences in levels of educational degrees

(17)

15

of the staff do not impact the wage-size premium. A large part of this effect remains unexplained for. Garcia-Pozo et al. (2012) argue that the data provided was fragmentary concerning variables that cover the organizational and financial diversion between small and large companies. Additional unobservable factors may result from large companies employing workers that exhibit characteristics as work attitude or innate capacity among others increasing their unmeasured individual productivity. Söderbom et al. (2005) also identify the existence of not yet observed factors that account for a large fraction of the size-wage premium.

2.3 Inter-Industry Effect

Wage-structures differ between industries. Du Caju et al. (2011) find significant differences in wages among employees who exhibit similar observed characteristics, but are employed in different sectors. The highest wages are paid within the refined petroleum, coke and the nuclear fuel industry, the financial inter-mediation sector, and the chemical industry. On the other side of the wage distribution are jobs within the textile and furniture manufacturing, water industries, and retail sector.

Abowd et al. (1999) define the industry of a firm as one of its characteristics. According to this definition the industry effect upon the wage of an individual are the “correct aggregation of

the pure firm effect within the industry” (Abowd et al., 1999: 257). In support of this argumentation,

many explanations for differences in the payment structure among industries resemble the arguments for the firm-size wage effect. But there are also other approaches explaining wage differentials between industries e.g. higher wages as compensation for job-related mortality (Leigh, 1995). As illustrated by Thaler (1989) the most common models to explain the wage anomaly are: shirking models, turnover models, adverse selection models, and fair wage models.

(18)

16

2.3.1 Shirking Models

As discussed above, monitoring employees to control their productivity is impractical due to a variety of reasons. According to Thaler (1989), companies within industries where monitoring workers is costly, an increase in wages might lead to a reduction of shirking. If a company engages in minor monitoring, pays above market wages, and lays off employees that have been shirking, the employees can lose rents and therefore have a reduced incentive to shirk. Summarizing, the higher the costs for monitoring employees within an industry, the higher the wages the companies pay to reduce shirking.

2.3.2 Turnover models

Stiglitz (1974) introduces a model that explains the volume of wages as influenced by the turnover costs. The more expensive the churn of employees is within an industry due to factors as training costs, the learning curve, etc., the higher the need for measures to reduce this trend. Above average wages may be interpreted as employee-turnover-reduction approach. The higher the wages, the higher the potential loss of rents for the employees when changing the company. According to Thaler (1989), the turnover model predicts that the industries paying the highest wages are the industries with the highest turnover rate otherwise. However, the existing research shows that industries with the highest average wages have the lowest turnover rates. This suggests that the higher wages overcompensate the potential higher turnover rate, allowing the workers to earn rents compared to workers in other industries. This opposes any efficiency wage models. According to them, rents are not possible due to the ability of workers to switch between industries. In this case, the turnover rates would be the same across industries (Gibbons & Katz, 1992).

(19)

17

2.3.3 Adverse selection model

The adverse selection model (Weiss, 1980) predicts that in a heterogeneous labor market exhibiting a large variety of abilities concerning employees companies offer high wages to attract high-performing workers. This leads to an increase in demand by potential employees, but instead of lowering the offered wages the companies can retrieve the employees with the highest potential from the pool of applicants. Applicants with a lower potential cannot increase their hiring probability by decreasing their acceptance wages. Thaler (1989: 188) argues that “industries which

are more sensitive to quality differences or have higher costs of measuring quality will offer high

wages”.

2.3.4 Fair wage models

According to fair wage models (Aklerof, 1984), the wage level impacts the fairness perception of the employees. If wages are low, workers will feel treated unfairly. This decreases the productivity of the employee and negatively affects the profitability of the firm. The opposite is the case if wages are high. This may increase the productivity of the employee, and is of high importance within industries with a high degree of worker cooperation and teamwork (Thaler, 1989). Gibbons and Katz’s (1992) research does not support this approach. According to them, it fails to explain the similarities observed in their study of the industry wage structure across borders and in countries with very different market systems. On the other hand, in his study concerning inter-industry wage differentials Benito (2000) determines the profitability of an industry as positively related to the wage premium, supporting the findings of the fair wage model by Aklerof (1984).

(20)

18

2.3.5 Trade liberalization

As found by Amiti and Davis (2011), the level of trade liberalization impacts the wages significantly. Decreases of input tariffs increase wages, while increases of input tariffs decrease the average wages paid. Therefore, an industry that exhibits a high degree of trade liberalization will exhibit a different wage structure compared to an industry with lower trade liberalization. Harrigan and Davis’ research (2011) argues that in economies where different jobs pay different wages to identical workers, trade liberalization may lead to an increase in competition for those ‘good jobs’. Employees receiving above average compensation for their work fear to be replaced by an employee demanding lower wages. In a simulation conducted by Harrigan and Davis (2011) about one quarter of the jobs paying above average wages will be destroyed by realizing the trade liberalization of an industry.

3 Framework

The literature broadly analyzes the effect of the firm-size upon the wages paid within a company and identifies potential drivers of it. Based upon this, we argue that the implications of the size of a company as e.g. monitoring costs, churn rates, and internal labor markets positively affect the wages paid by companies. As shown, these implications differ between the country of work and the industry. However, the existing literature does not agree upon the existence and the extent of the effect. The results of the studies vary depending on the countries studied and the data available. Therefore, our study closes the existing research gap by using a standardized sample of data that includes information of employees from various countries. The research also provides indications, which factors drive the firm-size wage effect. Additionally, we analyze the effect industry-characteristics have on the firm-size wage relationship. The literature has not investigated the moderating effect of the concentration, productivity, and capital intensity of an industry onto

(21)

19

the firm-size wage effect yet. However, the existing research indicates a positive direct relationship between the concentration, productivity, and capital intensity of an industry and the size of a firm and the wages paid. Accordingly, these factors may influence the firm-size wage relationship positively. To test the firm-size wage effect and to determine the moderating effect of the industry-characteristics, we propose a framework in this chapter.

The relationship between firm-size and wages has been researched and discussed by a broad variety of scholars. The so-called firm-size wage effect refers to the positive influence the size of a firm has on the level of wages. The literature considers it as a phenomenon, because it cannot be explained with the classic theoretical frameworks (Troske, 1999). It was discovered as an additional influence onto the wage level of a worker on top of individual skills, age, gender, education, etc. (Ours & Stoeldraijer, 2011; Nopo et al., 2010; Troske, 1999; Pedace, 2010; Amiti & Davis, 2011).

The drivers behind this effect are explained with sorting and job matching, capital-labor complementarities, unionization, efficiency wages, internal labor markets, and rent sharing and market power. These factors are supposed to be characteristics that differentiate bigger from smaller companies and that impact the wage level.

Referring to the sorting and job matching argument (Garicano & Rossi-Hansberg, 2006), bigger companies organize their work in ways that attract workers with a higher productivity compared to smaller firms. Additionally, bigger firms need to create monetary incentives for workers to work productively, because they cannot monitor them as cost efficient as smaller firms can. Based on this argumentation, bigger firms pay higher wages to achieve a higher productivity compared to average wages. Hamermesh’s (1980) argumentation refers to the higher capital intensity of bigger companies on average compared to smaller companies. This can lead to a higher

(22)

20

average productivity of the employees due to better equipment. More capital allows the implementation of higher-quality equipment and a more efficient work-flow. Therefore, the bigger the company, the better the equipment it can provide to improve its productivity. The higher wages reflect the increased productivity. Another argument for higher wages of employees of bigger companies is the intention of the firms to decrease the influence of unions. If workers earn above average wages, the demand to participate in unionization activities decreases. This excludes the direct effect union wages have on the wage level within a company, but refers to a so-called spill-over effect (Voos, 1983). Due to their size, larger firms suffer more from a unionized workforce compared to small firms; therefore, they are more likely to implement incentives that decrease the interest of workers to get unionized. Hence, bigger companies pay higher wages compared to smaller firms to avoid the influence of unions on their labor practices (Pedace, 2010). Shapiro and Stiglitz (1984) propose that bigger firms offer higher wages to reduce worker’s turnover and shirking. Above average wages increase the unemployment opportunity costs; therefore, they decrease the incentives for the employees to exhibit damaging behavior towards the firm. As Garicano and Rossi-Hansberg (2006) argue that this is due to the inability of big companies to monitor their workers effectively. Following this argumentation, the bigger the company the higher the wages paid.

The internal labor markets of bigger companies provide a broader variety and therefore a more specific fit of measures and activities to train certain job specializations and job-specific skills. These educational programs are investments into their workforce, which are of high value for the company. Due these investments, it is of high interest for the companies to reduce potential turnover of the trained and productive workforce. The internal labor markets of big companies evaluate and reward productive workers. Therefore, the rewarded workers have a smaller incentive to leave the company. The internal labor markets of big companies include reward systems that

(23)

21

increase the wages of high productive workers, which, in turn, decrease their turnover. Hence, smaller companies that do not have such an internal labor market do not reward the productivity of workers as bigger companies do (Brown & Medoff, 1989). According to Gimble (1991), bigger companies exhibit monopsic wage setting power due to their size and importance in the job market. According to this argumentation, employees have less bargaining power when applying for jobs at bigger companies. Therefore, the bigger the company, the smaller the bargaining power, which should result in lower wages.

The literature provides various attributes of firms that raise wages for the employees if the number of workers increases. The size of firms drives characteristics as sorting and job matching, capital-labor complementarities, unionization, efficiency wages, internal labor markets, and rent sharing. Accordingly, they differ between firms of different sizes. Following these assumptions, the firm-size drives the necessity to pay higher wages to compensate employees’ productivity, capabilities, or to avoid their churn. The firm-size positively influences the wages paid to the employees.

Despite the broad existing literature regarding the size-wage effect that largely agree on the predicted effect of the size of the company on the wages, the results of the quantitative studies are not consistent. The size-wage effect varies across countries. According to the study of Smyth and Gao (2011) the relation between firm-size and wage level is negative in China, in Italy Brunello and Colussi (1998) do not find evidence for a significant relation, and in the USA the relation is positive (Pedace, 2010). The differences of the measured effects may either originate from country factors or from differences in the data and analysis employed. The data sets differ concerning the level of observation (firm level (Pedace, 2010)/ individual level (Scoppa, 2014; Gao & Smyth,

(24)

22

2011), the time (Scoppa, 2014; Lallemand et al., 2007; Brunello & Colussi, 1998), or longitudinal or cross-sectional data (Scoppa, 2014; Lallemand et al., 2007).

Relating to the theory and evidence introduced above, we formulate the following hypothesis:

H1: The size of a company positively influences its wage level.

The firm-size-wage effect has been studied extensively on the level of one individual country (e.g. Pedace, 2010; Brown et al., 1990; etc.). Only few studies examine the size-wage effect across borders (Söderbom et al., 2005; Lallemand et al., 2007). Additionally, the existing studies did not analyze the wage-size effect across the borders of a sample of EU countries due to the lack of harmonized data. Some of the explanatory factors introduced above that cause the positive firm-size wage relationship might deviate in different countries. One of these effects is the capital intensity. The capital structures of companies vary across countries according to Rajan and Zingales (1995). If the impact described by Hamermesh (1980) is statistically significant, its influence onto the firm-size wage effect should differ between countries due to the differences in the capital structure.

The impact of the unionization on the firm-size wage effect should also differ between different countries. As Visser (2006) illustrates, the unionization ratio between the 14 examined countries differs between 6% of workers in Spain and 78% of workers in Denmark. The differences in the ratios of unionization lead to differences of the level of influence unions have in these countries. Therefore, following the argumentation of Voos (1983) that the threat of unions raises the wage level of employees of big companies to decrease their potential influence, the differences in the unionization ratio across the borders manifest in differences of the firm-size wage effect.

(25)

23

People’s perception of their risk to lose their job differs between countries according to Andersson and Pontusson (2007). This has an impact on the unemployment opportunity costs of the workers. Shapiro and Stiglitz (1984) argue that the wages of companies influence the opportunity costs of unemployment for employees; big companies increase their wages and therewith the opportunity costs of unemployment to decrease the probability of shirking and turnover. The perceived probability of employees to lose their job also influences the opportunity costs. Therefore, differences between countries in this area cause differences in the wages companies need to pay to increase the opportunity costs. Therefore, the firm-size wage effect will be influenced by these country specific characteristics.

We test the effect on a multi-national level, and on a country level. As shown above, the literature does not agree upon the existence and the impact the firm-size has on the wage level. By testing for the effect on a country level, we illuminate the origin of the divergence of the results of the existing literature; whether it results from different analysis types or the data studied, or factors on the country level influence it. The existence or non-existence of differences of the firm-size wage effect between countries provides valuable information on the driver of the effect. As the literature shows, the researchers do not see the size as the main driver of the higher wages, but certain yet unobserved effects that get amplified by firm-size. If differences exist between countries, the influence of macro-economic factors onto the effect can be derived.

We test the effect the country of work has on the firm-size wage relationship for Belgium, Czech Republic, Germany, Finland, the Netherlands, and Sweden. The countries’ firm-size wage effect has not been subject in the established literature and requires further investigation. Additionally, the countries are located in Central Europe and are members of the European Union and of the OECD. The World Bank database considers them as high income countries. They have

(26)

24

a life expectancy above 78 years, and a school enrollment in primary schools of 100%. This exhibits a high degree of conformity concerning the life quality across these countries. The Heritage Foundation in their 2015 Index of Economic Freedom characterizes the degree of economic freedom as moderately free for all of these countries. The similarity of the economic situation between these countries allows an analysis of the effect size differences based on structural differences besides the level of development. Söderbom et al. (2005) find differences between developing and developed countries; Smyth and Gao (2011) find a negative firm-size wage effect in China. The effect seems to be affected by country and/or environmental characteristics. A potentially large amount of impacts can dilute the analytical results of a comparison of countries that exhibit a high degree of differentials. Accordingly, the selection of Belgium, Czech Republic, Finland, Germany, the Netherlands, and Sweden allows a more narrow analysis of potential influences accounting for the firm-size wage effect.

Countries in the European Union exhibit various similarities in their culture, economic structure, etc., but also a large number of differences in their traditions and history (Biava et al., 2011). This results in differences in their economic systems. The Netherlands e.g. have a more developed capital market compared to Germany. Forty-five percent of the shares of Dutch companies are held by private owners as opposed to Germany, where the banks own a large share of the industry. In the Netherlands, bigger companies employ a larger share of the working population compared to big companies in Belgium and Germany. On the other hand, the Dutch medium-sized companies tend to be smaller in comparison to their neighbor countries (van Iterson & Olie, 1991). The countries also differ concerning their unionization rate. In the literature review, we introduced the unionization as an explanation for the firm-size effect by the existing literature. Studies show the existence of the firm-size effect even after controlling for unionization, but it is argued that it might have indirect effects as illustrated in chapter 2. According to Visser (2006),

(27)

25

the unionization in 2003 varies between 22.3% of the workforce in the Netherlands and 78% of the workforce in Sweden. The German share of unionized workers of the workforce amounts to 22.6%, in Czech Republic 27%, Belgium 55.4%, and 74.1% in Finland. Additionally, the countries also have different developments concerning the unionization rate. Finland increased the share of unionized workers in the years between 1993 and 2003, and so did Belgium. Czech Republic, Germany, the Netherlands, and Sweden have a declining share. Accordingly, the economic systems differ significantly concerning the unionization of the workforce.

The countries also differ concerning their degree of corporatism. This refers to the “type of

organized or coordinated capitalism where power to make important economic policies is

transferred from the parliament and government to semi-private organizations; these are based on

economic function or industrial sector and include a strong representation of labor interest”

(Siaroff, 1999: 176). The degree of corporatism influences the wage bargaining position of employees in their system; accordingly, it influences the wages set within these environments. Germany and Belgium have the lowest degrees of corporatism according to the ranking by Schmitter (1981). Finland, the Netherlands, and Sweden have a higher ranking compared to Germany and Belgium. The scale designed by Siaroff (1999) considers Sweden as a highly corporatist system, Germany and the Netherlands as moderate to strong corporatist countries, and Finland and Belgium as moderate corporatist countries.

Another approach proposed by the literature is the effect of rent-sharing onto the wages paid by companies. They argue that bigger firms tend to be more profitable, share their profits with their employees, and therefore pay higher wages. Tojerow (2008), Guertzgen (2010), and Arai et al. (2011) show that the rent sharing depends on a number of factors as the type of industry or characteristics of the individual and workforce. For instance, Guertzgen (2010) shows the

(28)

26

dependence of the degree of rent sharing on the existence of centralized wage agreements in a company or industry. Accordingly, the country characteristics as commonality of centralized wage agreements, sizes of certain industries, or educational standards of employees influence the rent shared with the employees.

Following the literature, the country characteristics influence the effect the firm-size has upon the wages paid. Besides many similarities the countries exhibit structural differences that influence the wages paid by companies. As discussed above, the unionization rate of a country influences the bargaining positions of employees and employers. Accordingly, differences in the commonality of unionized workers influence the threat of unionization for firms. In addition to that, the degree of corporatism differs between the countries. As the unionization rate of a country, the corporatization of a country’s system influences the bargaining position of employers and employees. In addition to that, differences in centralized wage agreements, country specific size of industries, and educational standards also influences it. Hence, country characteristics influence the bargaining positions of firms and individuals. We argue that these changes in the bargaining positions influence the impact of the effect between countries. The country of origin influences the relationship between firm-size and the wages a company pays due to country specific characteristics.

In general, the existing literature analyzes the firm-size wage effect in a single country. Smyth and Gao (2011) find a negative relation between firm-size and wage level, Brunello and Colussi (1998) analyze Italy and do not find evidence for a significant relation, and Pedace (2010) finds a positive relationship. The results differ across countries. The reasons can either arise from differences in the data or result from country specifics. Based on the results of the previous research and the argumentation above, we formulate the following hypothesis:

(29)

27

H2: The relationship between the size of a company and its wage level depends on the

country of origin.

As shown above, there exists a broad variety of literature about the influence of the type of industry on the wage level. Being one of a firm’s characteristics (Abowd et al., 1999), the industry influences its organizational structure and practices. Therefore, it also influences its human resource practices and the wages in particular.

In Fairris and Jonasson’s study (2008) of intra-industry differences they do not find a correlation between the direct effects of firm-size and industry type upon wages, but they identify a variance of the firm-size wage effect depending on the type of industry examined. Therefore, the characteristics of an industry have to impact the size wage relationship. One characteristic of an industry is the average size of its firms. The firm structure, according to Long and Link (1983) and Kwoka (1983), influences the average wages and benefits of employees in an industry significantly. Pugel (1980) determines market concentration as one aspect of product market power, and therefore as an influence on the bargaining situation of employees in an industry. Accordingly, the wages paid within an industry are influenced by the concentration.

The aggregated characteristics of companies within an industry establish an industry’s characteristics. If the concentration of large company within an industry is high, the size of the company is likely to be bigger than in industries with a minor concentration of big companies. In addition, the structure influences the wage level within an industry and an individual company’s bargaining of wages. If an industry has a large share of big companies that tend to pay higher wages, a company has to increase its wage level the bigger it is to compete with its competitors on the hiring of employees. Accordingly, the higher the concentration of big companies within an industry, the stronger the effect of the size of a company onto the wages it pays its employees.

(30)

28

Previous studies have examined the direct effect of the firm concentration on the wages paid within an industry (Long & Link, 1983; Kwoka, 1983), but there have not been studies regarding the moderating effect on the relationship between firm-size and wages.

H3: The share of large firms within an industry positively moderates the relationship

between a firm’s size and its wage level.

As Dickens and Katz (1986) determine, the capital intensity of a company positively influences its productivity. The increased productivity of companies with a higher capital intensity may result from a decreased downtime of the machinery or an increased level of safety at the workplace, reducing illnesses and injuries of the workforce (Idson & Oi, 1999). Additionally, capital intense firms adopt productivity enhancing technology earlier compared to less capital intense companies, and may spend more money on recruiting, training, and retaining workers that are more productive. A part of the efforts to hire and retain more productive employees includes the payment of higher wages according to Idson and Oi (1999). Troske (1999) shows a relationship between capital intensity of a firm and skill of the workers; Idson and Oi (1999: 107) argue “technology, preferences, markets, and regulations inﬂuence the size distribution”. Dickens and Katz (1987) acknowledge that the capital intensity of an industry influences the average wages paid. Concerning the firm-size, Hamermesh (1980) states that labor and capital intensity are complements. The more capital provided the more labor can be employed to increase the productivity. A high capital intensity within the industry accompanies a higher firm-size following that.

According to Abowd et al. (1999) the characteristics of the environment impact a firm. Therefore, the industry-characteristics influence the properties and behavior of a firm. In addition

(31)

29

to its own capital structure, the average capital intensity of a firm’s industry should impact its size and the value of the wages paid.

H4: The average capital intensity of an industry positively moderates the relationship of a

firm’s size and its wage level.

The literature agrees upon the impact the productivity has upon an employee’s wage. The higher the productivity, the higher the wage a firm pays to the employee (Idson & Oi, 1999; Pedace, 2010; Garcia-Porzo et al., 2012). Wage differentials between industries may result from differences in labor productivity (Genre et al., 2011) that can result from a better access to capital as computers etc. (Troske, 1999), intrinsic motivation of the employees (Becchetti et al., 2013), or efficient internal labor markets (Doeringer & Piore, 1971). Improvements in technology and therefore in productivity allow companies to realize scale effects (Tovar et al., 2011). Accordingly, the literature agrees on the higher productivity of employees as potential explanation for the observed wage differentials between industries (Du Caju et al., 2011). Van Biesebroeck (2005) finds that large firms tend to exhibit a higher productivity. The capital markets provide credit to more productive firms and the labor markets transfer employees from less to more productive firms, which facilitates the growth of bigger and more productive companies. Large companies gather resources more easily compared to small companies. Summarizing, firm and industry productivity influences the company sizes and wage bargaining.

As shown above, the average productivity of an industry positively affects the size of the companies in it and the wages paid. Accordingly, we argue that the relationship between the size of a firm and the wages it pays to its employees is affected positively.

The conclusion described above is in accordance with Van Biesebroeck (2005) who conducted a study showing the positive impact of productivity on the firm-size and Du Caju et al.

(32)

30

(2011) who determine a positive effect of an industry’s productivity onto the wages paid to employees.

H5: The average productivity of an industry positively moderates the relationship of a

firm’s size and the wages it pays.

(33)

31

4 Data & Methodology

4.1 Sample & Data Collection

The sample we employ to analyze the relation of a firm’s size and the wage level consists of the datasets from 2011 of employees from

Belgium, Czech Republic, Finland, Germany, Netherlands, and Sweden and adds up to a total number of 25,606 datasets. As shown in Table 1, Germany and the Netherlands

provide the biggest number of datasets, which make up to 42% each. The number of datasets from Belgium, Czech Republic, Finland, and Sweden is smaller, but still big enough to test for differences between countries concerning the wage levels. Each dataset consists of the matched data from the WageIndicator, the STAN database from the OECD, and the Eurostat database. The data is cross-sectional and we analyze it based on a linear-regression.

The WageIndicator is a dataset collected by the Amsterdam Institute for Advanced Labour

Studies. The data was retrieved from a worldwide, continuous, multilingual web-survey on work

and wages with paper supplements from 2004 until 2013. It contains individual information of employees from 47 countries worldwide as their wages and benefits, their working hours, employment status and contract information, occupation, industry and place of work, their employment history, and personal information. We use the data on wages and firm-size to determine the relationship between the dependent and the independent variable. The additional information will be used to control for effects of individual characteristics as age, education, etc. have upon the wage level.

N % Country Belgium 1922 8% Czech Republic 619 2% Finland 629 2% Germany 10838 42% Netherlands 10837 42% Sweden 761 3%

(34)

32

Using the WageIndicator’s data allows us to observe the firm-size wage effect across national borders, which has not been able to be examined yet due to a lack of conformity of the data between different countries. The standardized questionnaires that have been translated into several languages allow a multinational comparison. However, it is important to recognize a weakness of this dataset. The questionnaires have been responded to anonymously in regard to the individual and the firm. Therefore, it is not possible for us to add additional control variables as e.g. the firm performance. An omitted variable bias could be a possible result from this shortcoming if the performance of a firm is correlated with its size. Additionally, it is not possible to determine whether two employees work for the same firm.

For the moderating effects of the average productivity and concentration of an industry we need a different dataset. We retrieved this information from the Eurostat database provided by the European Commission. The data Eurostat contains is aggregated information upon the country and industry level and can be matched to the data from WageIndicator due to the industries’ NACE codes. The dataset from Eurostat provides information about the average wage within an industry, the average value added by each employee, the average turnover and profit per employee within an industry. We use this data to measure the moderating effect the productivity and concentration of a company’s industry have upon the relationship between dependent and independent variable.

The advantage of using the Eurostat database is that the data collection has been standardized across the EU-member states.

For the industry data on the average capital intensity employed, we need a third database. The information is provided by the STAN Indicators Rev. 4 database of the OECD. The database’s data is the average investment intensity in an industry in percent. The ratio is calculated by dividing an industry’s gross fixed capital formation by the value added. We used the data of the capital

(35)

33

intensity to measure the moderating effect of an industry’s capital intensity upon the relationship between the dependent and independent variable. The advantage of using the data from the STAN Indicator database by the OECD is that it contains standardized data across all relevant countries.

4.2 Variables

4.2.1 Dependent Variable

The dependent variable of this quantitative research is the natural logarithm of an individual worker’s net hourly wage. This excludes bonuses and other employee benefits provided by the employer as health insurances, overtime allowances, job trainings, etc. This is due to a better comparability between different occupations. The WageIndicator database provides this data. The usage of the logarithm of the hourly wages of employees for analyzing the firm-size wage effect is in accordance with previous studies (Lallemand, 2007; Pedace, 2010; Brunello & Colussi, 1998). Scoppa (2014) uses the average wages received by an employee per working day, but this is due to a lack of information available. Measuring wages on a daily basis does not consider different length of working days. To avoid outliers, the analyzed data excludes wages below the 0.1 percentile and above the 99.9 percentile.

4.2.2 Independent Variable

The independent variable of the model is the firm-size measured by the number of employees. The data provided by the WageIndicator includes the number of employees divided into 7 groups (1-9, 10-19, 20-49, 50-99, 100-199, 200-499, 500+). For the regression analysis of the data we value the groups with their middle value divided by 100. Accordingly, the values assigned respectively are: 0.05; 0.15; 0.35; 0.75; 1.5; 3.5; 7.5. Albaek et al. (1998) and Brown & Medoff (1989) employ the same technique to use a regression analysis for ordinal data of continuous variables. The data by the WageIndicator only allows clustering, which is the less

(36)

34

appreciated approach compared to using the accurate number of employees of a firm (Pedace, 2010; Lallemand, 2007). Clustering the firms regarding to their employee size into groups can result in a “misspecification and firm-size wage identification issues, especially if the relationship is

non-linear” (Pedace, 2010: 169).

4.2.3 Moderating Variable

We use the average productivity, average capital intensity, and concentration of an industry as moderating variables. The average productivity is measured by the average value added per employee in an industry of a country as proposed by Genre et al. (2011). The value added per employee is calculated by using the data of total value added in an industry of each country and divided by the total number of employees in this country’s industry. The Eurostat database provides this data for the European Union.

The concentration of an industry is determined by the share companies with sizes of 0-9 employees (referred to as small), 10-49 employees (referred to as medium-small), 50-249 employees (referred to as medium), and above 250 employees (referred to as big) have in the industry. We analyze the moderating effect of each group onto the relationship between dependent and independent variable.

We retrieve the value of the capital intensity parameter from the STAN Indicator database by the OECD. The capital intensity is calculated by dividing an industry’s gross fixed capital formation by the value added.

4.2.4 Control Variables

The control variables used in this analysis are individual characteristics to allow the determination of the pure firm-size effect. The first control variables is the occupation of the

(37)

35

employee as the wages are on average higher for white collar employees compared to blue collar employees (Fox, 2009). Additionally, we control for gender, since this also determines the level of wages, and the age of the employees, because wages get raised with the level of experience of an employee. We use age as indicator for the experience in this analysis. The control variables we use are in line with previous studies as performed by Pedace (2010), Smyth & Gao (2011), Scoppa (2014), and Lallemand (2007).

To test the hypotheses proposed in chapter three we design fourteen models. In each model created, we test for different effects and employ a different configuration of variables e.g. other moderators or control variables. The table below shows the variables used in each model (table 2).

To determine the relationship between firm-size and wages, we conduct a linear regression analysis to estimate the model based on the cross-sectional data collected. We use two frameworks.

Dependent Variable

Independent Variable

Control Variable Moderator

Model Firm-Size Ln Wage Age Gender Occupation Country

1 x - x x x x - 2 x x x x x x - 3 x x x x - - Belgium 4 x x x x x - Czech Rep 5 x x x x x - Finland 6 x x x x x - Germany 7 x x x x - - Netherlands 8 x x x x - - Sweden 9 x x x x x x Capital Intensity 10 x x x x x x Value Added

11 x x x x x x Share of Small Sized Firms 12 x x x x x x Share of Medium-Small Sized Firms 13 x x x x x x Share of Medium Sized Firms 14 x x x x x x Share of Large Firms

(38)

36

The first framework tests the models 1 to 8. The framework implemented is based upon previous studies (Pedace et al., 2010):

ln(𝑤_𝑖𝑡) = 𝛼 + 𝑠_𝑖𝑡𝛿 + 𝑥_𝑖𝑡𝛽 + 𝜀_𝑖𝑡

In this framework the subscripts 𝑖 and 𝑡 represent an individual and the time, respectively. The 𝑤 stands for the wage of the individual employee and the 𝑠 for the size of the firm the individual employee works for. The framework considers control variables by the vector 𝑥. The effect of the firm-size on wages is considered by 𝛿. The control variables are primarily individual characteristics of the employee as age, occupation (blue collar or white collar work), gender, but also the country of employment in the models 1 and 2. In the models 3 to 8 the country of work determines which data is selected.

For the analysis of the moderation effects we use a different formula that includes the moderating effect of the industry.

ln(𝑤_𝑖𝑡) = 𝛼 + 𝑠𝑖𝑡𝛿 + 𝑥𝑖𝑡𝛽 + 𝑑𝑖𝑡𝜃 + 𝑑𝑖𝑡𝑠𝑖𝑡𝜇 + 𝜀𝑖𝑡

In this framework the variables have the same meaning as in the first framework, but the direct effect of the industry-characteristics is considered by 𝑑. The moderating effect of the industry on the relationship between the dependent variable and 𝑑𝑖𝑡𝑠𝑖𝑡 considers the independent variable. This

framework applies to the models 9-14.

5 Data Analysis and Results

In the following chapter, we present descriptive statistics of the data employed, analyze the correlations and the collinearity of the variables, and provide the results from the regression analysis we performed to test the hypotheses.

(39)

37

5.1 Descriptive Statistics Analysis

Table 3 shows the descriptive statistics of the control variables and the independent and dependent variables. The mean ln net wage per hour of the datasets is 2.317. This amounts to a total mean of 12.24 € hourly net wage as shown in Table 4. 79% of the respondents answering this question (N=11,990) of the WageIndicator

questionnaires are white collar employees. The Oxford Dictionary defines “white collar” as “a

person engaged in non-manual work, esp. in office work of an administrative, managerial, or

clerical nature; an office worker. Opposed to blue-collar”. The average net hourly wage of an

white collar employee equals 12.95 €. This significantly differs with a significance above a 99% level from the mean wage of blue collar employees, who earn 10.04 €. The Oxford Dictionary defines blue collar workers as “manual work or workers, esp. in industry”.

The share of women in the sample is 42%. This amounts to 10,721 women as opposed to 14,885 male respondents. With a mean net hourly wage of 11.03 €, women earn 2.08 € less compared to the other gender. This difference is also significant at a 99% level (F=124.661).

Comparing the mean wages of the considered countries shows a wide spread of incomes. Employees in Sweden earn, on average, the highest wages (15.04 €/hour). The lowest net wages are paid to workers in the Czech Republic with a mean net hourly wage of 5.40 €. Belgium (11.49 €/hour), Finland (11.61 €/hour), Germany (12.65 €/hour), and the Netherlands (12.19 €/hour) are in the center of this interval. The ANOVA analysis shows that the national means differ at a significance level of over 99 % (F=35.374).

Mean Standard Deviation Ln Wage 2.317 0.542 White Collar 0.7972 0.4021 Female 0.42 0.493 Age 39.28 11.127 Firm-Size 288.77 299.859