THE ROLE OF BRANDS IN JOB CHOICE DECISIONS: HOW A STRONG BRAND PERSONALITY MAKES RECRUITMENT EASY

THE ROLE OF BRANDS IN JOB CHOICE DECISIONS:

HOW A STRONG BRAND PERSONALITY MAKES

RECRUITMENT EASY

NIELS OLIVIER

University of Groningen

Faculty of economics and business

MSc Marketing

1

THE ROLE OF BRANDS IN JOB CHOICE DECISIONS: HOW A STRONG BRAND PERSONALITY

MAKES RECRUITMENT EASY

By Niels Olivier

Faculty of economics and business

Master thesis June 2018 Toellan 35 8447 DZ Heerenveen (06) 53 88 47 70 E-mail: niels.olivier92@gmail.com Student number: 25 64 815

2

ABSTRACT

The purpose of this study is to research the role of brands in job choice scenarios. This study uses employee based brand equity (EBBE) to measure the extent to which the brand has an impact in job choices. It then decomposes the brand effect into employer brand awareness and employer brand image, to measure which aspects of the brand cause this brand effect. In the last part of this paper, the moderating role of individual personality on employer brand image is tested. Data was collected via an online choice based conjoint analysis, which resulted in 235 respondents. The brand image effects were measured using brand personality (Aaker 1997). Individual personality was measured using the Big Five model (Goldberg 1990). The main findings of this paper indicate that brands indeed have a significant positive effect in job choice decisions, this effect is mainly caused by high levels on the brand personality dimensions: Excitement and Sincerity. This indicates that companies who are perceived as exciting and sincere have a higher EBBE and would therefore theoretically have to pay less salary and still be able to attract future employees. Practical methods and calculations to asses and maximize the EBBE effect for companies are provided. No moderating role of individual personality was found.

Keywords: Employer branding, Employee based brand equity, employer attractiveness, Employer

3

Table of Content

2.1 Decomposition employer branding……….….………..7

2.2 Employer attractiveness………..………8

2.3 Employee based brand equity………..………..9

2.4 Employer brand awareness……….…10

2.5 Employer brand image………...11

2.6 Brand personality………...11

2.7 Individual personality……….…….13

2.8 Conceptual framework………..15

3. METHODOLOGY……….………...16

3.1 Research method……….…….…16

3.2 Choice based conjoint analysis……….……...16

3.3 Experimental design………..…..…19 4. RESULTS………..……….……….24 4.1 Sample……….……….…...24 4.2 Benchmark model………..…….26 4.3 Extended model 1……….26 4.4 Extended model 2……….29

4.5 Comparison benchmark model and extended models………..……..29

4.6 Hypotheses testing………..29

4.7 Additional analyses………..32

5. DISCUSSION………..………...34

5.1 Employee based brand equity………..34

5.2 Brand personality………..35

5.3 Employer brand awareness………35

5.4 Individual personality……….36

6. IMPLICATIONS……….………...36

6.1 Practical implications……….……….36

6.2 Creating and altering brand personality……….………..38

7. LIMITATIONS……….………..………..….……….……….39

8. FUTURE RESEARCH………..………...…...40

9. LITERATURE………41

Appendix A: R code………..………46

Appendix B: Example questionnaire……….……..56

Appendix C: Additional tables and results……….……….…….68

4

1. INTRODUCTION

After the economic recession in 2008, the world economy is now growing again at the same pace as it did before the recession (International Monetary Fund 2018). With an expected GDP(Gross domestic product) growth of 3.7% for 2018, the world is looking forward to a prosperous time. The same trend is shown in unemployment. The unemployment rate steadily declined to 5.8% (The

World Bank 2018). For The Netherlands, similar patterns occurred. Economic growth, measured in

GDP, is at the pre-crisis level again (3.1%) and the unemployment rate is now at 4.9%, while it was at 7.4% at its highest point in 2014.

These trends are highly attractive from the perspective of society, however, it also comes with a number of challenges, especially from the business perspective. Due to the fact that the pool of potential new employees is shrinking, companies face a severe struggle in attracting new talent (ManpowerGroup 2014; Lievens & Highhouse 2003). To illustrate this: In 2014, there were 7 unemployed persons per vacancy in The Netherlands, in 2017, this number declined to only 2.2 (Groenewegen, Van Schoot & Vriesselaar 2017). This problem will be even more relevant in the coming years when macro-economic trends, such as declining employment rates and population ageing, continue (Michaels, Handfield-Jones & Axelrod 2001).

Finding qualified staff is essential, as the role of human capital has a strong positive effect on business success (Benson 1998), measured as an increase in profitability. The contribution of

employee knowledge, skills, and abilities to organizational performance has also become more important in recent years, because knowledge has become a critical ingredient in achieving competitive advantage in the new economic landscape (Breaugh & Starke 2000; Hitt et al. 2001). Therefore one can conclude that in the coming years businesses will face an increasingly difficult problem, to attract high skilled people and that the consequences of not succeeding in this process can have a severe impact on business success in the long run.

The effects of this problem are already visible. In a global study, over one-third of the 37.000 employers in 42 countries reported talent shortages (ManpowerGroup 2014). This is the highest percentage in seven years.

1.1 Employer branding

5 Barrow (1996) were among the first authors to study the effect of brand equity in the recruitment phase. This lead to the concept of employee based brand equity (EBBE), which is a derivative of customer based brand equity (Kashive & Khanna 2017; Keller 1993). Applying the structure of the definition of customer based brand equity to EBBE, one can be define it as: the differential effect that brand knowledge has on potential employee response towards working for that brand. This will be further elaborated on in the theory section. EBBE has shown to positively affect organizational attractiveness (Jiang & Iles 2011). However, no research has yet been done on how impactful EBBE is as a predictor of realistic job choice decisions and which characteristics of the brand are responsible for this effect. This research paper makes three key contributes to the field of employer branding and recruitment research by studying the following items:

(1): The employer branding literature is strengthened by quantifying the brand effect (EBBE) in realistic job choice decisions. Multiple studies concluded that different aspects of a brand (brand awareness and brand image) have positive effects on organizational attractiveness and intentions to apply (Chapman et al. 2005; Jiang & Iles 2011; Del Vecchio 2007), however the effect of a brand in actual job choice decisions is still unknown. In fact, only a handful of papers is responsible for what we know about which job choice predictors affect actual job choice (Chapman et al. 2005). Multiple studies highlight the need for more research to assess whether actual job choice can be reliably predicted (Barber 1998; Ryan & Ployhart 2000). However, real job choice decisions are hard to research, because the recruitment process is a relative long, individual and most often

unstandardized process. Therefore it is hard for researchers to gather enough data from multiple applicants in order to generalize conclusions. This study uses a conjoint experiment to bridge this gap. In this experiment, actual job choices are simulated in order to quantify the brand effect, using EBBE as the measured construct. This approach brings the field of research one step closer to predicting actual job choice, by researching the brand effect in job choice situations.

6 and more holistic understanding on which aspects of the brand cause the effect that a brand can have in job choice decisions.

(3): The final contribution of this paper is to test whether the brand (image) effect is moderated by the individual personality of the respondent. The aim of the last part of this paper is to generate a deeper understanding about the effect of individual personality on the brand effect in job choice scenarios. Lievens and Highhouse (2003) stated that symbolic attributes allow employees to maintain or enhance their self-image. Brand image consists of symbolic attributes about the brand (Keller 1993). So therefore, it is expected that different types of people, value brands differently on the basis of their own personality. To test this, this study uses the Big Five model in order to test the moderating effect of individual personality dimensions on the brand personality dimensions

(Goldberg 1990). This has been tested in customer marketing research by Mulyanegara, Tsarenko and Anderson (2009). They found empirical evidence about the impact of the individual dimensions on brand personality. such that individuals with high levels on certain dimensions valued

companies/brands higher with these matching brand personality dimensions. This study builds upon the foundations from customer marketing research, to test whether this effect is also present in employer branding.

1.2 Practical relevance

The results of this study are highly relevant for brand managers and recruiters as well. This paper will generate insights into how important brands are in job choice decisions. It then explains what the brand effect consists of and how this differs for people with different personalities. Doing so, managers and recruiters can learn how to fully leverage and optimize their own brand in their attempts to attract new employees. Practical strategies and methods on how to work with EBBE and how to maximize the effect are provided in the implications chapter. As a useful side effect, this paper also generates knowledge about the attractiveness of different industries and locations. These results can be used as well by managers, to either emphasize or avoid the type of industry and location in their marketing efforts for a specific job offer. Lastly, the effect of the salary level on job choice is included as well. Managers can use the results of this model to make predictions about the height of the salary level which should be offered to successfully attract new employees. How these predictions can be made and how you can calculate the financial impact is on the company, will be presented in the implication chapter as well.

1.3 Thesis structure.

7 used in this study. Next, based on the literature, hypotheses are proposed. After that, the choice based conjoint expirements, that are used to research the hypotheses will be explained. The results of this paper will be displayed and further explained and discussed in the discussion section. The implication chapter focuses on the practical consequences of this study. Limitations and suggestions for future research will be presented last.

2. THEORY

Employer branding is generally seen as the overarching theme of research on employer attractiveness, recruitment marketing, employee based brand equity and employee satisfaction. There is little unified understanding about the constructs that are being used in the research field of employer branding (Theurer et al. 2016). Employer branding is defined as an approach to

recruitment and retention that ‘involves internally and externally promoting a clear view of what makes a firm different and desirable as an employer’(Lievens 2007 p. 51). Related constructs and terms such as employment image, employee branding, employer attractiveness, employer image, employer knowledge and employer brand equity are inconsistently defined and applied (Theurer et al. 2016). This paper uses EBBE as the main construct to measure the brand effects in job choices. However, to achieve a thorough understanding about the brand effect in the recruitment phase, based on past research, it is important to analyze the findings from the most important constructs that are being used in employer branding. This will provide the clarity to develop a systematic approach to research the brand effect in job choice decisions in a realistic context.

2.1 Decomposition employer branding

To decompose employer branding into the different constructs that are being used, it is useful to first describe the process of employer branding. Employer branding consists of three stages (Lievens 2007): The first step is to develop a concept of value that is offered to current and potential employees. This means for example the salary level, job benefits and promotion opportunities. Next, that concept forms the central message in the external marketing strategy, in order to attract applicants from the target group. The last step is to incorporate this value concept in the organizational culture. This is also called internal marketing (Lievens 2007).

8 2.2 Employer attractiveness

Employer attractiveness is defined as: ‘the envisioned benefits that a potential employee sees in working for a specific organization’ (Berthon et al. 2005, p. 151). This concept has mostly been studied by organizational and personnel psychologists (Jiang & Iles 2011). Research on Employer attractiveness has identified many factors that influence the attractiveness to work for a certain organization. These range from job and organizational characteristics, such as type of work and working hours, to perceived fit and recruiter characteristics (Chapman et al. 2005).

Certain brand effects are also incorporated in research on employer attractiveness, for example: organization image. But these effects did not incorporate all the aspects of the total brand effect (Jiang & Iles 2011; Chapman et al. 2005). To study the total brand effect, marketing

researchers introduced the concept of EBBE (Jiang & Iles 2011). This concept is a derivative of

customer based brand equity (Keller 1993). Cable & Turban (2001) argued that the same principles of brand effects on customer behavior are applicable to (potential) employee behavior. Therefore, EBBE is used in this study to measure the total brand effect in job choice scenarios. This will be further elaborated upon in the next section.

To be able to determine how important brand effects are in realistic job choice decisions, other factors that influence job choice decision-making are incorporated as well. Therefore, it is worthwhile to analyze the most important factors that influence employer attractiveness, that are identified in past research. Predictors of employer attractiveness can be decomposed into

instrumental(functional) and symbolic attributes (Lievens 2007; Keller 2013).

Instrumental attributes. Instrumental attributes are those attributes of a job and/or organization that can be classified as functional or utilitarian (Theurer et al. 2016; Keller 2013). A lot of research has been done on utilitarian job and organizational attributes (Chapman et al. 2005; Lievens 2007; Theurer et al. 2016). Some of the examples of utilitarian attributes in research papers are: Payment level, advancement opportunities, job security, working with customers, flexibility, responsibility, sociality and team climate (Berthon et al. 2005; Chapman et al. 2005; Lievens & Highhouse 2003; Theurer et al. 2016). These attributes have an effect on employer attractiveness. Therefore it is expected that these attributes also play an important role in actual job choice decisions. However, not all attributes are applicable to every job choice situation, for example: Advancement

9 predictors in employer attractiveness research (Chapman et al. 2005). Although multiple studies differ in their conclusions on the importance of payment levels, it is widely accepted that payment plays an important role in job choice decisions (Theurer et al. 2016). In their meta-analysis on employer attractiveness, Chapman et al. (2005) concluded that the payment level is of medium importance in predicting organizational attractiveness in comparison to other factors. This was against the expectation of the researchers, since it is a widely believed fact that payment is one of the most important attributes in job choice decisions. They suggested that the lower importance of payment was due to the fact that the majority of research on payment importance were field studies in which pay varied naturally. So there might have been a limited range of payment offers to job seekers. Attributes that have a low variation, receive less weight than attributes with greater variability (Rynes, Schwab and Heneman 1983) Therefore, it is interesting to test whether the importance of payment differs in a lab experiment in which there are no natural payment

restrictions. However, the effect of payment on job choice is not the focus of this paper, but since it is expected to explain a significant amount of the variance in job choice, it is included as a control variable in the conceptual framework.

Symbolic attributes. Symbolic attributes are defined as self-expressive organizational attributes that are part of the employer image (Theurer et al. 2016). This concept closely relates to EBBE (Keller 1993). The difference between the two concepts is that EBBE also includes the effect of brand awareness, while Symbolic attributes only account for employer image effects. To make the distinction clearer: Employer image is a concept that is derived from brand image, a concept that originated in customer marketing research. Brand image in turn is part of Keller’s framework for building customer based brand equity (CBBE), which he defined as: ‘the differential effect of brand knowledge on consumer response to the marketing of the brand’. Within Keller’s (1993) framework he states that brand knowledge consists of brand awareness and brand image. CBBE can been transformed into EBBE, such that EBBE consists of employer brand image and employer brand awareness. In this research, the focus is on the effect that a brand has in job choice decisions for potential employees. Therefore, EBBE is proposed as the construct to measure the differential effect that brand knowledge has on potential employee response towards working for that brand. The construct effectively measures all symbolic attributes and includes the effect of brand awareness.

2.3 Employee based brand equity

10 focused on the conceptualization (Cable & Turban 2001; Lievens & Highhouse 2003; King & Grace 2010). There are some articles that indicate that (parts of) EBBE play(s) a significant role in job choice decisions. Chapman et al. (2005) for example showed that employer image strongly and positively affects the intention to apply for a job. Collins (2006) found that employer knowledge strongly predicted interest in applying for a job and actual application behavior. Berthon et al. (2005) found a link between employer attractiveness and EBBE. He stated that the more attractive an employer is perceived, the stronger that particular organization’s EBBE. Jiang & Iles (2011) concluded that marketers should focus on EBBE in order to enhance organizational attractiveness and intentions to apply. So there is evidence that EBBE plays a significant role in the recruitment process. This research aims to strengthen the evidence, by researching and decomposing the effects of EBBE in job choice decisions, in comparison to other job choice predictors.

To research the concept of EBBE, it is important to separate the brand effects from other effects that are directly linked to the brand. Therefore, the effects of the location and the effect of the industry in which the brand is active are included as control variables. Location preference has shown to have a positive effect on job acceptance by applicants ( Chapman et al. 2005). Industry preference has not directly been researched, but closely related concepts such as type of work and work environment have both shown to have a positive effect on job acceptance intentions (Chapman et al. 2005).Therefore, both variables are expected to play a significant in job choice decisions. The following hypothesis will be tested in this research paper, with regard to EBBE:

H1: Employee based brand equity has a significant positive effect on job choice by potential applicants.

2.4 Employer brand awareness

EBBE is a relative large concept. If it turns out that EBBE plays an extensive role in job choice decisions, the impact on practical implications for managers would be rather low, since it is not clear which aspects of EBBE result in a more extensive role of EBBE in job choice decisions. It can be that brand awareness is high/low for a certain brand, or that product related associations cause a

11 different purchase and consumption situations in which the brand comes to mind and depth refers to how easily a brand comes to mind. The depth and breadth of brand awareness is referred to as brand familiarity in the field of employer branding(Cable and Turban 2001). Multiple studies

concluded that a high brand familiarity positively affected employer attractiveness (Cable & Graham 2000; Gatewood, Gowan, &Lautenschlager 1993; Turban 2001;Turban & Greening 1996; Brooks, Highhouse, Russell, & Mohr 2003). A possible underlying reason for this is the mere exposure effect (Zajonc 1968). This theory states that people that are repeatedly exposed to a stimuli, tend to develop a preference for this stimuli. Therefore it is expected that people who are familiar with a brand, tend to feel more positive about the brand and would be more likely to be willing to work for this brand. In this way, brand awareness positively contributes to a higher brand effect in job choice decisions. This results in the following hypothesis:

H2a: Brand awareness has a positive effect on Employee based brand equity in job choice decisions

2.5 Employer Brand Image

Brand image is defined by Keller (1993) as perceptions about a brand, as reflected by the brand associations held in consumer memory. Brand image has shown to have a positive effect on recruitment, since applicants remembered more from advertisements from companies with a strong brand image, and were willing to earn less in order to work for that company (Cable & Turban 2003). The issue, however, in studying brand image effects in the recruitment phase, is that brand image is concerned with associations, and these associations can be either related to the quality of the product as well as symbolic associations about the organization. This makes it hard to generalize findings between companies in different industries, since they do not serve the same function (Keller 1993). Another aspect that makes comparison hard is the fact that companies strive for a unique and favorable brand image (Keller 2013). In order to research which characteristics of the brand’s image result in a higher brand effect in job choice scenarios, a brand personality approach is taken. 2.6 Brand personality

12 The first dimension in Aaker’s (1997) model, excitement, describes brands that can be seen as daring, spirited, imaginative and up-to-date. The second dimension, sincerity, concerns brands that are down to earth, honest wholesome and cheerful. Competent brands are associated with reliableness, intellegence and succesfullness. Sophisticated brands are those that can be described by their charming and upper class appearance. And the last dimension, ruggedness, concerns brands that have a masculine and tough appearance.

The field of consumer research already uses personality dimensions to describe intangible aspects of brands for a longer while (Keller and Lehmann 2006). This is very relevant for marketers, since it has shown that a strong brand personality positively affects brand equity (Valette-Florence, Guizani & Merunka 2011; Su & Tong 2015). Surprisingly however, there has been little evidence on which dimensions of a brands personality result in a high brand equity. Su & Tong (2015) have recently researched this in the sportswear industry. They found that competence & sincerity significantly contributed to a higher brand equity.

The concept of brand personality has also been studied in the field of employer branding in recent years. Brand personality conceptualizations are often used to derive symbolic attributes of the employer image (Slaughter et al. 2004). Dimensions of brand personality have shown to have an effect on employer brand attractiveness (Rampl & Kenning 2014). They found that the sincerity dimension most positively influenced employer brand trust and affect. Employer brand trust and affect in turn explained 71% of the variance in employer brand attractiveness. Furthermore, they found that the dimensions excitement and sophistication positively affected employer brand affect as well. The ruggedness dimension showed a slight negative effect on employer brand affect. The effect of the competence dimension on brand affect and brand trust were not significant. Other authors came to somewhat different conclusions. Lievens & Highhouse (2003) for example found that the competence dimension had simplified differentiation between different banks employer brands and could therefore be an important indicator of organizational

attractiveness. Lievens (2007) also concluded that sincere and competent brands are of greater importance for actual applicants than other dimensions.

The effect of brand personality dimensions in job choice decisions and more specifically, on EBBE has not been researched before. This is important for marketers since a high EBBE enhances recruitment success (Theurer et al. 2016). Which in turn reduces costs associated with recruitment. It can also be useful in the sense that a higher EBBE means that people are willing to receive less salary in order to work for that brand, which is a very useful competitive advantage.

13 to form an opinion about the attractiveness to work for a certain brand (Theurer et al. 2016).

Therefore, it is expected that certain dimensions of the brand personality create a desirable image towards working for that brand. When we combine this theory with the results of previous papers, one can expect a positive relationship for the following dimensions: Excitement, Sincerity,

Competence and Sophistication. A negative relation is expected for Ruggedness. This results in the following hypotheses:

H2b: The degree of excitement has a positive effect on Employee based brand equity in job choice decisions.

H2c: The degree of sincerity has a positive effect on Employee based brand equity in job choice decisions.

H2d: The degree of competence has a positive effect on Employee based brand equity in job choice decisions.

H2e: The degree of sophistication has a positive effect on Employee based brand equity in job choice decisions.

H2f: The degree of ruggedness has a positive effect on Employee based brand equity in job choice decisions.

2.7 Individual personality.

14 the perceived attractiveness of the company. Following the reference group theory and the brand personality approach, it is expected that individuals find companies more attractive when they possess personality dimensions that maintain or enhance their self-image. Therefore, this paper researches the moderating role of individual personality dimensions on the relationship between brand personality dimensions and EBBE.

To study this, the Big five model is applied (Goldberg 1990). This model is widely accepted as a tool to categorize and describe individual personality. The model defines the following dimensions: extraversion, agreeableness, conscientiousness, emotional stability and openness in order to

describe individual personality. To research this moderating effect, this paper builds upon the foundations of Mulyanegara, Tsarenko and Anderson’s (2009) study on the relationship between the big five dimensions and the brand personality dimensions in Aaker’s (1997) model in customer marketing. They found empirical evidence about the impact of the individual dimensions on brand personality and found a link between: excitement with extraversion, sincerity with agreeableness and competence with conscientiousness, such that individuals with high levels on these dimensions valued companies/brands higher with these matching brand personality dimensions. The

expectation in this paper is that this effect will also be present in employer branding.

The big five model has also been applied to study employer attractiveness for existing employees (Anitha & Madhavkumar 2012). They found that the conscientiousness dimension of individuals was most positively related to employer attractiveness. A study by Davies et al. (2004) found that agreeableness was the strongest of the five dimensions in influencing initial attraction, employee satisfaction and affinity for the brand.

15

H3a: The degree of extraversion moderates the relationship between excitement and Employee based brand equity in job choice decisions, such that extraverted employees react more positively to

employers categorized as exciting.

H3b: The degree of agreeableness moderates the relationship between sincerity dimension and Employee based brand equity in job choice decisions, such that agreeable employees react more positively to employers categorized as sincere.

H3c: The degree of conscientiousness moderates the relationship between competence and employee based brand equity in job choice decisions, such that conscientious employees react more positively to employers categorized as competent.

2.8 Conceptual framework.

Based on the theory section and the proposed hypotheses, the following conceptual model forms the basis of this research paper:

16

3. METHODOLOGY

The basis of this study relies on two separate conjoint analysis and an additional survey, in order to measure the effects of brand (EBBE), location, industry, payment level and the

decomposition of the brand effect. The reason why this study applies a dual choice based conjoint experiment is to separate the employer brand effect from the other job choice predictors that are interrelated with the brand, which are industry and location. These attributes cannot be varied independently. For example, you cannot assign three different locations to the same brand if that brand is only located in one particular city. The consequence is that one can only calculate the total effect of location, industry and the brand itself. Therefore, two separate conjoint experiments are being performed. The first experiment only calculates the utility values for industry, location and payment . The second Choice based conjoint experiment calculates the utility value for the combined effect of the employer (brand, industry and location) and payment level. By subtracting the industry and location values (experiment 1) from the total employer brand effect (experiment 2), one can effectivity measure the true brand effect, which is expressed in this study as EBBE. after the two conjoint experiments, respondents will be asked to fill in an additional questionnaire. The questionnaire consists of questions regarding brand awareness, brand personality, individual personality and descriptive measures. The results are used to decompose the effect of EBBE in job choices.

3.2 Choice based conjoint analysis.

A choice based conjoint analysis builds upon random utility theory as the theoretical framework (Mc Fadden 1974). The theory states that the overall utility U of an individual i, for an object j is a latent construct that includes a systematic component V and an error component ℇ. The error effect accounts for all effects that are not systemic (Manski 1977). This results in the following formula:

𝑈

= 𝑉

+ 𝜀

17 systematic utility (𝑉𝑖𝑗) for a specific job offering can be estimated with the MNL model (McFadden 1974). The MNL model presents the probability (prob) that a job offering (a) is chosen from a choice set of J alternatives:

𝑝𝑟𝑜𝑏(𝑎|𝐽) = exp⁡(𝑉𝑖𝑎) ∑ exp⁡(𝑉𝐽_{𝑗 𝑖𝑗})

Eq. 2 This model allows for an estimation of the utility functions for each job offering in a choice set. Each job offering consists of different attributes, in this study: brand, location, industry and payment. And each attribute can have multiple levels, such as presented in table 1 and 2. This study is interested in the part worth utility values of these different levels. Therefore, preferences for attribute levels have to be decomposed statistically from the overall evaluation of the job offering. In order to be able to estimate the part worth utility values (β) for each attribute and their

corresponding levels, this study uses the Log likelihood function in order to multiply the MNL models that are calculated for each choice set (Eq. 3). The formula multiplies the probabilities of chosen alternatives a across consumers n and choice sets c. This step is followed by a maximum likelihood estimation procedure. In this procedure, the set of part worth utility values (β),that maximizes the log likelihood function is identified.

𝐿𝐿 = ∑ ∑ 𝐿𝑛(𝑝𝑟𝑜𝑏(𝑎𝑛𝑐|𝐽𝑛𝑐) 𝑐

𝑛

Eq. 3

Benchmark model. The model that forms the basis for this study, called the benchmark model, calculates the systematic utility (𝑉𝑖𝑗) of job offering j for respondent i, based on the summation of the part worth utility values (β) of the four attributes and their corresponding levels: brand, location, industry and payment level. This is visualized in equation 4:

𝑉

= ⁡

18 The βEBBE⁡ effect in equation 4 cannot be calculated directly, as explained in the research approach paragraph. The effect is derived from two separate experiments. The second experiment calculates the total employer brand effect, which is a combined effect of brand, location and industry (𝛽𝑒𝑚𝑝) The first experiment calculates the beta values for location (βloc), industry (βind) and the payment level β(pay). By subtracting the beta values from experiment 1 from the employer brand effect in experiment 2, one can derive the beta value for EBBE:

𝛽

= ⁡ 𝛽

− ⁡ 𝛽

− 𝛽

Eq. 5 Extended model 1. As explained in the theory section, it is expected that employer brand awareness and employer brand image (via brand personality) explain a significant part of the variance of the brand effect (EBBE) in job choices (figure 1). Therefore, the corresponding dimensions and constructs are added in this study. The employer brand image effects are measured through brand personality dimensions, derived from Aaker (1997): Excitement (exc), Sincerity (sin), Competence (com), Sophistication (sop) and Ruggedness (rug), Brand awareness (awa) effects are tested through brand recognition. The corresponding model forms extended model 1 in this study and is based on the following formula, in which 𝛽𝐸𝐵𝐵𝐸𝑟 𝐸𝐵𝐵𝐸𝑗⁡represents the residual brand effect that is not accounted for by these constructs:

𝑉𝑖𝑗= ⁡ 𝛽𝑙𝑜𝑐𝐿𝑜𝑐𝑗+ 𝛽𝑖𝑛𝑑𝐼𝑛𝑑𝑗+ 𝛽𝑝𝑎𝑦𝑃𝑎𝑦𝑗+ ⁡ 𝛽𝐸𝐵𝐵𝐸𝑟 𝐸𝐵𝐵𝐸𝑗+ 𝛽𝑒𝑥𝑐𝐸𝑥𝑐𝑖𝐸𝐵𝐵𝐸+ 𝛽𝑠𝑖𝑛𝑆𝑖𝑛𝑖𝐸𝐵𝐵𝐸 + 𝛽𝑐𝑜𝑚𝐶𝑜𝑚𝑖𝐸𝐵𝐵𝐸+ 𝛽𝑠𝑜𝑝𝑆𝑜𝑝𝑖𝐸𝐵𝐵𝐸+ 𝛽𝑟𝑢𝑔𝑅𝑢𝑔𝑖𝐸𝐵𝐵𝐸+ 𝛽𝑎𝑤𝑎𝐴𝑤𝑎𝑖𝐸𝐵𝐵𝐸

Eq. 6 Extended model 2. Furthermore, this study also includes individual personality characteristics, in order to measure the moderating effect (γ) on the brand personality dimensions. As presented in the theory section, extraversion (ext), agreeableness (agr) and conscientiousness (con) are included to test the moderating role on excitement, sincerity and competence. This results in extended model 2. This model captures the systemic utility component V for individual i, and job offering j:

19 3.3 Experimental design.

The conjoint analysis part of this study measures the effect of the brands, locations, industries and payment levels. The brands are chosen into four categories of industries, namely energy, electronics, consumer goods and finance. These industries were chosen, because of the fact that they were among the biggest industries in The Netherlands and were therefore a realistic representation of future work opportunities for respondents (Graydon 2013). Another reason why the decision has been made to incorporate these industries is because the results speak to wider audience (e.g. more companies). When selecting the brands, two main aspects were taken into account. The first one was the minimization of brand locations, in order to avoid the number of levels effect and to achieve a parsimonious research design. This means that there are only companies in the sample with a headquarter in the following cities: The Hague, Amsterdam, Rotterdam or Utrecht. The second aspect that was taken into account was the estimation of brand knowledge of

respondents for various companies. Therefore, companies with different expected brand awareness ratios were included, in order to get variation in the results for the brand effect for various

companies. One fictional brand is included in the research design, in order to validate the results and to get the estimates for a brand with zero EBBE, because it is assumed that a fictional brand has no brand equity. The included brands, locations, industries and payment levels are presented in table 1 and 2.

Conjoint experiment 1. Before the conjoint experiments, respondents were asked to fill in their preferred job and their expected salary. Respondents were then presented with 7 choice sets in experiment 1. Each choice set consisted of 3 suggested job offerings for their preferred job and a no choice option. Figure 2 presents an example of such a choice set. The first experiment only includes the attributes and corresponding levels from table 1. These levels were varied independently. When designing experiment 1, as well as experiment 2, efficiency criteria were taken into account. The design is balanced, orthogonal and has minimum overlap (Kuhfeld et al. 1994). Experiment 1 leads to separate estimates for industry, location and payment level.

attribute Level 1 Level 2 Level 3 Level 4 Level 5

Industry Finance Consumer goods Electronics Energy

Location The Hague Amsterdam Utrecht Rotterdam

Payment -15% -7,5% Expected salary +7,5% + 15%

20 Conjoint experiment 2. The second experiment included 7 choice sets as well. Again, it consisted of 3 suggested job offerings and a no choice option. This time, with two attributes, such as presented in table 2. The effects for brand, industry and location could not be varied independently. Therefore the second experiment resulted in a combined estimate for the total employer brand effect, consisting of the brand, industry and location. An example of a choice set is presented in figure 2. The entire survey can be found in Appendix 2.

attribute Level 1 Level 2 Level 3 Level 4 Level 5

Employer* Brands Rabobank ING ASN bank Unilever Douwe Egberts Heineken KPN TomTom Elektrolight** Eneco Nuon Royal Dutch Shell

Industry Finance Consumer goods Electronics Energy

Location Rab: Utrecht ING: Amsterdam

ASN: The Hague

Uni: Rotterdam DE: Utrecht Hei: Amsterdam KPN: The Hague Tom: Amsterdam Ele: Utrecht Ene: Rotterdam Nuo: Amsterdam Shell: The Hague

Payment -15% -7,5% Expected salary +7,5% + 15%

* Brand, industry and location cannot be varied independently. ** Fictional brand

21 FIGURE 2: Example of choice sets in experiment 1 and 2

Construction benchmark model The results of both experiments are used to calculate the benchmark model as presented in equation 4. This is done by subtracting the beta values from industry and location in experiment 1, from the combined beta value of brand, location and industry (𝛽𝑡𝑜𝑡) in experiment 2. This is mathematically presented in equation 5.

Construction extended models. The constructs of the brand personality dimensions are

22 they are aware of and if they associate a brand with a certain personality dimension (Sonnier 2011). In order to measure individual personality, this study uses a simplified version of the question markers, presented by Goldberg (1992). This choice has been made in order to balance the length of the questionnaire, but without losing the essence of each dimension. The results of the different constructs are included in the extended model, such as presented in equation 7. The following methods are used to operationalize and measure the different constructs (table 3, 4 and 5):

Construct Operationalization Scale

Excitement Which of the following brands would you describe as exciting, daring and spirited?

0-1 binary scale:

0= no 1=yes

Sincerity Which of the following brands would you describe as sincere, down to earth and honest?

0-1 binary scale:

0= no 1=yes

Competence Which of the following brands would you describe as competent, intelligent and successful?

0-1 binary scale:

0= no 1=yes

Sophistication Which of the following brands would you describe as sophisticated, upper class and charming?

0-1 binary scale:

0= no 1=yes

Ruggedness Which of the following brands would you describe as rugged, tough and masculine?

0-1 binary scale: 0= no

1=yes

TABLE 3: Operationalization brand personality

Construct Operationalization Scale

Employer brand awareness

Please select all brands that you are familiar with, for example, you have previously seen or heard of them.

0-1 binary scale

0= not familiar 1= familiar

23

Construct Operationalization Scale

Extraversion How would you describe yourself in a group of people? 0-5 Likert scale: 0= very introvert 5= very extrovert 0-5 Likert scale: 0= very silent 5= very talkative 0-5 Likert scale: 0= very unassertive 5= very assertive

Agreeableness How would you describe yourself in response to others? 0-5 Likert scale: 0= very unkind 5= very kind 0-5 Likert scale: 0= very disagreeable 5= very agreeable 0-5 Likert scale: 0= very uncooperative 5= very cooperative

Conscientiousness How would you describe yourself when performing tasks? 0-5 Likert scale 0= very careless 5= very thorough 0-5 Likert scale 0= very disorganized 5= very organized 0-5 Likert scale 0= very irresponsible 5= very responsible

24 4. RESULTS

4.1 Sample

To gather the required data, a questionnaire was sent out, via multiple channels, towards a diverse target group. This resulted, after cleaning the dataset, in a number of 235 respondents. The sample consists of 50.6% male respondents and 49.4% female respondents. The majority of the respondents has a Dutch background (54.9%), followed by respondents with a Chinese background (30.6%). The rest of the sample consisted of people from countries as Norway, Germany, Spain, USA and 17 more. Most respondents were students (64.3%), who were on average 23 years old. The other part of the sample, the working class (35.7%), has an average age of 36 years old. The sample consists of both students looking for a first job, as well as people that are working for a longer while, whom might be looking for a next step in their career.

4.2 Benchmark model

The aim of the first part of this study is to separate the brand effects from the combined employer effect of location, industry and brand. In order to do so, two separate conjoint experiments were performed, which resulted in dataset 1 and 2. Using dataset 2, one is able to calculate the employer brand effect. Using dataset 1, one can find the separate effects of location and industry. By subtracting the data of experiment 1 from the data of experiment 2, one creates a clean measure of the true brand effect. This is presented in table 6. The separated effects are presented in the Dataset 2-1 column under ‘Combined'. This model was then compared with the full dataset (1 and 2

25

Dataset 2 - 1 Full dataset(1 and 2) (Benchmark model)

separate combined Combined

Brand Brand Brand

Rabobank 0.21 (0.09)* -0.04 0.01 (0.13) Ing 0.16 (0.10) ® -0.04 -0.02 (0.13) ASN Bank -0.18(0.11) ® -0.08 -0.07 (0.14) Unilever 0.52 (0.09)* 0.25 0.33(0.13)* Douwe Egberts 0.40 (0.09)* -0.15 -0.08 (0.12) Heineken 0.58 (0.09)* 0.08 0.13 (0.12) KPN -0.44 (0.12)* 0.12 0.08 (0.15) Tom Tom -0.28 (0.11)* -0.02 -0.05 (0.14) ElektroLight -0.33 (0.11)* -0.12 -0.13 (0.14) Eneco -0.33 (0.11)* -0.14 -0.12 (0.15) Nuon -0.37 (0.11)* -0.41 -0.43 (0.14)* Shell 0.07 (0.10) 0.33 0.34 (0.13)* Industry Industry Finance 0.08 (0.06) 0.07 (0.06) Consumer goods 0.38 (0.06)* 0.36 (0.05)* Electronics -0.38 (0.07)* -0.35 (0.06)* Energy -0.08 (0.06) -0.08 (0.06) Location Location Rotterdam -0.11 (0.06) -0.13 (0.06)* Amsterdam 0.12 (0.06)* 0.14 (0.06)* The Hague -0.18 (0.06)* -0.17 (0.06)* Utrecht 0.17 (0.06)* 0.16 (0.06)* Payment level 5.49 (0.36)* 4.88 (0.23)* No Choice 5.83 (0.38)* 5.01 (0.24)* *= significant at p<0.05 ®= marginally significant at p<0.10

26 4.3 Extended model 1.

In order to decompose the EBBE effect, this study uses 6 variables (figure 1). Table 7 shows the mean score for each variable on each brand. Table 8 shows that there is moderate significant correlation between these variables. However, the VIF scores, as presented in Table 9, show that adding the variables to the model will not lead to any critical variance inflation factors.

Rab ING ASN Uni DE Hei KPN Tom Ele Ene Nuon Shell

Awareness 0.74 0.81 0.54 0.87 0.66 0.83 0.72 0.67 0.06 0.59 0.57 0.89 Excitement 0.17 0.26 0.13 0.43 0.21 0.51 0.13 0.17 0.06 0.13 0.09 0.30 Sincerity 0.19 0.21 0.21 0.28 0.31 0.26 0.12 0.14 0.06 0.12 0.1 0.15 Competence 0.38 0.51 0.16 0.62 0.31 0.56 0.31 0.26 0.06 0.19 0.2 0.52 Sophistication 0.25 0.43 0.2 0.34 0.18 0.29 0.18 0.09 0.03 0.09 0.06 0.33 Ruggedness 0.17 0.25 0.09 0.16 0.07 0.41 0.13 0.13 0.09 0.11 0.17 0.51

TABLE 7: Mean scores brand decomposition effects

Awareness Excitement Sincerity Competence Sophistication Ruggedness

Awareness - 0.43** 0.37** 0.58** 0.41** 0.37** Excitement 0.43** - 0.39** 0.48** 0.42** 0.31** Sincerity 0.37** 0.39** - 0.36** 0.35** 0.22** Competence 0.58** 0.48** 0.36** - 0.50** 0.36** Sophistication 0.41** 0.42** 0.35** 0.50** - 0.30** Ruggedness 0.37** 0.31** 0.22** 0.36** 0.30** -

** = significant at the 0.01 level

TABLE 8: Correlation matrix brand decomposition effects

Awareness Excitement Sincerity Competence Sophistication Ruggedness

VIF score 1.7 1.5 1.3 1.9 1.5 1.2

TABLE 9: VIF scores brand decomposition effects

The brand awareness and brand personality variables were added to the benchmark model (equation 4). This resulted in the extended model 1 (equation 5), as presented in table 11.

4.4 Extended model 2.

27 variance and by including three factors, 65% of the total variance was explained, as opposed to 55% with two factors. After rotating the component matrix, the result showed high loadings (> 0.5) for the three questions about extraversion on factor 1, the three questions about agreeableness on factor 2 and the three questions on conscientiousness on factor 3 (table 10). Afterwards, the Chronbach’s Alphas were tested for each factor. The test indicated that there were no problems with the internal consistency, since all factors scored well above 0.6 (table 10). As a next step, the rotated (mean centered) factor scores were saved as new variables and included as moderators (Extraversion, Agreeableness and Conscientiousness) in extended model 2 (table 11).

Component Chronbach's alpha

1 2 3 ext01introvert 0.868 0.09 -0.064 0.769 ext02silent 0.852 0.088 0.17 ext03unassertive 0.722 -0.03 0.247 agr01unkind 0.067 0.739 0.326 0.732 agr02disagreeable -0.03 0.813 0.058 agr03uncooperative 0.135 0.766 0.261 con01careless 0.032 0.213 0.732 0.664 con02disorganized 0.114 0.111 0.784 con03irresponsible 0.223 0.289 0.666

TABLE 10: Factor analysis moderation effects

28

Benchmark model Extended model 1 Extended model 2

Brand Brand residual Brand residual

Rabobank 0.01 (0.13) 0.03 (0.13) 0.03 (0.13) ING -0.02 (0.13) -0.08 (0.13) -0.08 (0.13) ASN Bank -0.07 (0.14) -0.02 (0.14) -0.02 (0.14) Unilever 0.33(0.13)* 0.06 (0.13) 0.06 (0.13) Douwe Egberts -0.08 (0.12) -0.16 (0.13) -0.16 (0.13) Heineken 0.13 (0.12) -0.17 (0.13) -0.16 (0.13) KPN 0.08 (0.15) 0.20 (0.15) 0.20 (0.15) Tom Tom -0.05 (0.14) 0.04 (0.14) 0.04 (0.14) ElektroLight -0.13 (0.14) 0.10 (0.15) 0.10 (0.15) Eneco -0.12 (0.15) -0.01 (0.15) -0.01 (0.15) Nuon -0.43 (0.14)* - 0.24 (0.14) ® - 0.24 (0.14) ® Shell 0.34 (0.13)* 0.25 (0.14) ® 0.25 (0.14) ®

Industry Industry Industry

Finance 0.07 (0.06) 0.07 (0.06) 0.07 (0.06)

Consumer goods 0.36 (0.05)* 0.37 (0.07)* 0.37 (0.06)*

Electronics -0.35 (0.06)* -0.37 (0.07)* -0.37 (0.07)*

Energy -0.08 (0.06) -0.07 (0.06) -0.07 (0.06)

Location Location Location

Rotterdam -0.13 (0.06)* -0.11 (0.06) ® -0.11 (0.06) ® Amsterdam 0.14 (0.06)* 0.12 (0.06)* 0.12 (0.06)* The Hague -0.17 (0.06)* -0.17 (0.06)* -0.17 (0.06)* Utrecht 0.16 (0.06)* 0.16 (0.06)* 0.16 (0.06)* Payment level 4.88 (0.23)* 4.95 (0.23)* 4.96 (0.23)* No Choice 5.01 (0.24)* 5.25 (0.24)* 5.25 (0.24)*

Brand effects Brand effects

Awareness 0.001 (0.08) 0.001 (0.08) Excitement 0.74 (0.09)* 0.82(0.21)* Sincerity 0.53 (0.09)* 0.27(0.35) Competence 0.04 (0.08) 0.31(0.31) Sophistication 0.10 (0.09) 0.10 (0.09) Ruggedness 0.01 (0.09) 0.01 (0.09) Individual personality Extraversion -0.03(0.08) Agreeableness 0.09(0.12) Conscientiousness -0.09(0.10) *= significant at p<0.05 ®= marginally significant at p<0.10

29 4.5 Comparison benchmark model and extended models.

The three different models in table 11 reflect equation 4 (benchmark model), 6 (extended model 1) and 7 (extended model 2). In order to analyze the quality and added value of each model, a number of tests were performed. First, the log-likelihood values were calculated for the three models, including a NULL model. The NULL model assumes that all beta’s are 0 and that the probabilities in a choice set are equal for all alternatives. So in the NULL model, all choices are random. By comparing the log likelihood values of the three models with the NULL model, one can calculate whether the models perform better in explaining choice behavior. This comparison is done via a Mc Fadden R² calculation. In table 12 one can see that all three models are better in predicting choices than the NULL model, since the R² values are different from zero. Significance of the results is tested using a likelihood ratio test. The test shows that the benchmark model is performing

significantly better than the NULL model. It also shows that extended model 1 performs better than the benchmark model. However, extended model 2 is not a significant improvement relative to extended model 1. Model 2 does, however, includes more parameters, so the model is less parsimonious than model 1.

Null model benchmark model

Extended model 1 Extended model 2

log-likelihood -4560.9 -4209.6 -4115.9 -4115.2

Mc Fadden R² 0.077 0.098 0.098

Likelihood ratio test p<0.001 p<0.001 p=0.60

MAE 2.20% 4.00% 3.5%

TABLE 12: Analysis of the different models

The MAE value calculates the Mean Absolute Error. This is an indicator for the predictive validity of the models. It compares the predicted values with the actual values. The average difference reflects the Mean absolute error. As can be seen in table 12, the predictive validity of all 3 models is rather high, which indicates that all three models are capable of predicting choices rather well. The benchmark model shows the highest predictive validity. The calculation of the MAE values can be found in appendix 3.

4.6 Hypotheses testing

30 including brands provides a highly significant better model (p=0.001). To further analyze the effect of brands, this study calculates the importance of all attributes in equation 4 (brands, location, industry and payment level), as can be seen in figure 3. The results show that the importance of brands is 23% in job choice decisions. It is the second most important aspect in job choice decisions. These analysis provide clear evidence that hypothesis 1 is supported.

FIGURE 3: Attribute importance benchmark model

Brand awareness and brand personality. To analyze whether brand awareness and brand personality dimensions significantly affect EBBE in job choice decisions, extended model 1 is used (table 11). One can see that the dimensions Excitement and Sincerity show a significant positive effect. Brand awareness, Competence, Sophistication and Ruggedness show no significant effect. However, when brand personality dimensions are excluded, brand awareness does show a significant positive effect. Due to the correlation between the variables and the light multicollinearity, the effect of brand awareness spreads over multiple brand personality dimensions (table 8 and 9).

Furthermore, when extended model 1 is compared with the benchmark model (table 11), one can conclude that the values for location, industry and payment did hardly change, while the brand effects did. Which indicates that the brand personality dimensions and brand awareness

decomposed the brand effect rather well. To further illustrate this, see Figure 4. The figure shows that the importance of payment, location and industry did not change very much, relative to Figure 3. However, the brand effect is reduced from 23% to 13%, which means that the included effects for brand personality and awareness were able to explain a large deal of the brand (EBBE) effect. Furthermore, one can state that exciting and sincere brands together have a 15% importance in job

Brand 23% Industry 22% Location 10% Payment level 45%

31 choices, which is rather extensive. Therefore one can conclude that there is strong support for hypothesis 2b and 2c and that these dimensions explain an extensive part of the total brand (EBBE) effect.

FIGURE 4: Attribute importance extended model 1

Individual personality. In order to test the moderating effects of extraversion, agreeableness and Conscientiousness on excitement, sincerity and competence, the mean centered factor scores were added as interaction effects in extended model 2, such as presented in equation 7. The results (table 11) show that there is no significant effect for any of the moderating effects: Extraversion on

excitement (p=0.65), agreeableness on sincerity(p=0.45) and conscientiousness on competence (p=0.36). Extended model 2 also did not significantly improved extended model 1. Therefore, no support has been found for hypotheses H3a, H3b and H3c.

32 4.7 Additional analyses

Willingness to receive less salary. Based on the brand and payment results in table 11, one can calculate the willingness for potential employees to receive less salary, in order to be able to work for a specific brand. This can be done by dividing the brand effects by the payment effect. This is an important measure for companies in order to estimate the financial effect of their employer

branding efforts. A higher EBBE means that a company, in theory would have to pay less salary to still attract new employees. Since extended model 1 shows the highest goodness of fit and includes the brand decomposition effects, which can be affected by marketers, this model is used to calculate willingness to receive less salary. In order to do so, the total brand effects need to be calculated. These are calculated as the sum of the beta values of each construct multiplied by the mean rating (table 7). For example, Rabobank has the following brand effect:

𝑏𝑟𝑎𝑛𝑑𝑒𝑓𝑓𝑒𝑐𝑡(𝑅𝑎𝑏𝑜𝑏𝑎𝑛𝑘)

= 0.03(𝑟𝑎𝑏) + 0.001 ∗ 0.74(𝑎𝑤𝑎) + 0.74 ∗ 0.17(exc) + 0.53 ∗ 0.19(𝑠𝑖𝑛) + 0.04 ∗ 0.38(𝑐𝑜𝑚) + 0.10 ∗ 0.25(𝑠𝑜𝑝) + 0.01 ∗ 0.17(𝑟𝑢𝑔) = ⁡0.30

Eq. 8

Zero centering the data. The calculated effects are absolute values, however, all brand effects in the benchmark model are mean centered. Therefore, the mean value of the absolute brand effects is subtracted from each brand. This results in table 13. The table shows that these effects are almost similar.

Total brand effect extended model

Total brand effect extended model 1 (mean centered)

Total brand effect benchmark model Rabobank 0.30 0.01 0.01 ING 0.29 0.00 -0.02 ASN bank 0.22 -0.07 -0.07 Unilever 0.59 0.30 0.33 Douwe Egberts 0.19 -0.10 -0.08 Heineken 0.40 0.11 0.13 KPN 0.39 0.10 0.08 Tom Tom 0.26 -0.03 -0.05 Elektrolight 0.18 -0.11 -0.13 Eneco 0.17 -0.12 -0.12 Nuon -0.10 -0.39 -0.43 Shell 0.61 0.32 0.34

33 To calculate willingness to receive less salary, this study uses the mean centered total brand effects and divides them by the payment level beta value in table 11. This results in the following results.

Willingness to receive less salary Absolute WTRLS

Rabobank 0.19% 2.29% ING 0.00% 2.10% ASN bank -1.33% 0.77% Unilever 5.71% 7.81% DE -1.90% 0.20% Heineken 2.10% 4.20% KPN 1.90% 4.00% Tom Tom -0.57% 1.53% Elektrolight -2.10% - Eneco -2.29% -0.19% Nuon -7.43% -5.33% Shell 6.10% 8.20%

TABLE 14: Willingness to receive less salary (WTRLS)

These results show the willingness to receive less salary, based on the average EBBE of the included brands. This is due to the fact that all brand effects are mean centered. However, brands are interested in the absolute effect of their brand equity. Therefore this study included the fictional brand: Elektrolight. As stated in the methodology, it is assumed that this brand has no brand equity. Therefore, the difference between the willingness to receive less salary value of Elektrolight and the values of other brands can be considered as the absolute willingness to receive less salary due to the brand. The results show relative large differences between brands. It also shows brands, such as NUON and Eneco, who have an EBBE level that is worse than that of a company with zero brand equity.

34 5. DISCUSSION

5.1 Employee based brand equity

The analysis provided a unique quantitative view on job choice decisions. As Chapman et al. (2005) argued, there are only a handful of papers responsible for what we know about job choice predictors in actual job choice scenario’s. In this paper, actual job choice scenarios were simulated in a conjoint experiment. This allowed for an analysis on the role of different job choice predictors, in particular the true brand effect (EBBE). The analysis showed a significant role for brands in job choices, in which the differences between brands (EBBE) became more clear. Furthermore, it showed that the importance of the brand in job choice decisions is rather high (23%). A similar effect is present for the industry type (22%) and location amounted up to 10%. However, the biggest effect is the influence of the payment level (45%). So, the analysis showed that job seekers are mostly motivated by money, but the role of brands is also significantly large. As discussed in the

introduction, many studies contributed to a list of varying factors that play a role in the recruitment process, but we do not know much about how these factors truly contribute and how they

interrelate in actual job choices (Theurer et al 2016; Chapman et al. 2005; Barber 1998; Ryan & Ployhart 2000). This study provides a clear insight in the role of brands, locations, industries and payment level and the relative importance of each factor in job choice decision. This study brings the field of employer branding and recruitment research one step closer towards accurately predicting job choice behavior, by providing a more holistic view on the decomposition of job choices.

35 5.2 Brand personality

After including the brand personality effects to the conjoint model, there was clear evidence that the brand personality dimensions excitement and sincerity have a significant positive effect on the brand effects. This means that brands with an exciting, daring, sincere and honest brand image enjoy a higher brand effect in job choice decisions. The analysis did not show significant results for competence, sophistication and ruggedness. This is partly in line with previous research. Rampl & Kenning (2014), for example found a positive effect for excitement and sincerity on brand affect and trust, which is a predictor of brand attractiveness. However, Lievens & Highhouse (2003) and Lievens (2007) indicated that competent brands might also have a significant effect in job choices.

Furthermore, Rampl & Kennig (2014) found a positive effect for sophistication and a negative significant effect for ruggedness. So, the results partly support previous research, but also provided new insights in how important these dimensions are in job choice decisions. As presented in figure 4, the combined total effect of sincerity (5%) and excitement (10%) in job choice decisions is 15%, while the total brand effect (EBBE) in the benchmark model was 23%. This clearly indicates that these two dimensions explained a large part of the total brand effect (EBBE). Based on the theory that brand associations create expectations of the workplace, one can conclude that employees value an exciting, daring, sincere and honest work environment (Keller 2013; Theurer et al. 2016). Based on this study, marketers should therefore focus on increasing the perception of excitement and sincerity in order to increase EBBE. A small increase in these perceptions can have a large positive effect on EBBE, which will be shown in the next chapter. A higher EBBE means that potential employees are more willing to work for that specific brand and are also willing to receive less salary in order to get the job (table 14). Therefore, it is important for marketers to focus their marketing efforts on twitching the employer brand image towards an exciting and sincere brand to work for. How this works and what marketers can do to realize a higher EBBE effect is also presented in the next chapter.

5.3 Employer brand awareness

36 be included in the brand image (brand personality) effect. This is also supported by the relative high correlations with the brand personality dimensions and the VIF score (table 8 and 9). However, this does not mean that increasing brand awareness is not helpful in maximizing the brand effect. It seems that increasing brand awareness is supportive to the effect of the brand personality

dimensions. Therefore, the recommendation is to increase brand awareness, while simultaneously focusing on signaling excitement and sincerity.

5.4 Individual personality

The general expectation, in the theory section, was that the included personality

characteristics would enhance the positive effect of the brand personality dimension: Excitement, sincerity and competence (Anitha & Madhavkumar 2012). However, no significant results were found. This shows that the effects of excitement, sincerity and competence were not moderated by extraversion, agreeableness and conscientiousness. This implicates that brands that score high on excitement, sincerity and competence do not necessary attract job seekers with matching individual characters. One can conclude that companies with an employer branding strategy focused on excitement and sincerity do not attract a uniform pool of potential employees. Which can be an important consideration, since diversity on the work floor is an important factor in firm performance (Richard et al. 2004). This result can be caused by the fact that strong brand signal value for the environment (Del Vecchio 2007). So, working for an exciting, sincere and/or competent brand might be desirable from a social perspective. This would attract all sorts of individual characters, since different characters might all aspire to be associated with excitement, sincerity and competence (Hyman 1942).

6. IMPLICATIONS

6.1 Practical implications

In the final part of the results chapter, the respondents willingness to receive less salary is calculated (table 14). This method is usually applied in customer based brand equity (CBBE)

37 yearly salary of € 36.000,- , which is based on research by CPB (2018). The number of employees were based on research by Elsevier (2014) and the final assumption that has been made, is that companies hire on the same rate as the yearly turnover rate, in order to remain the same size. SHRM (2017) found that the average turnover rate is 18%.

EBBE effect on salary Impact on average

Dutch yearly salary Employees

Impact on hiring (18% hiring rate) Rabobank 2.29% € 823 57.000 € 8.442.514 ING 2.10% € 754 84.000 € 11.404.800 ASN bank 0.77% € 274 161 € 7.949 Unilever 7.81% € 2.811 173.000 € 87.547.886 DE 0.20% € 69 7.600 € 93.806 Heineken 4.20% € 1.509 81.000 € 21.994.971 KPN 4.00% € 1.440 23.000 € 5.961.600 Tom Tom 1.53% € 549 3.500 € 345.600 Eneco -0.19% € -69 7.100 € -87.634 Nuon -5.33% € -1.920 4.800 € -1.658.880 Shell 8.20% € 2.949 92.000 € 48.828.343

TABLE 15: Impact EBBE on companies

These results show how much money these companies would have to pay extra per year if they would have lost their current brand equity. Interestingly, the results also show that NUON and ENECO have a lower EBBE than the fictional brand. So the perception of these brands is so low that they would actually benefit from not including their brand names in a job offer, although this seems impossible. The results clearly indicate the financial consequences of each company’s EBBE. It also implicates the high relevance for companies to maximize the brand effect in job choice decisions, which means that a marketing budget for maximizing the brand effect is justified. This paper found that marketers should focus on increasing brand excitement and sincerity perceptions in order to enhance the brand effect. To illustrate what an increase in these dimensions would mean for a specific company and how much money could be saved on salary expenses, see the following example:

38 𝑏𝑟𝑎𝑛𝑑𝑒𝑓𝑓𝑒𝑐𝑡(𝐾𝑃𝑁)

= 0.20(𝐾𝑃𝑁) + 0.001 ∗ 0.72(𝑎𝑤𝑎) + 0.74 ∗ (0.13 + 0.20)(exc) + 0.53 ∗ (0.12 + 0.20)(𝑠𝑖𝑛) + 0.04 ∗ 0.31(𝑐𝑜𝑚) + 0.10 ∗ 0.18(𝑠𝑜𝑝) + 0.01 ∗ 0.13(𝑟𝑢𝑔) = ⁡0.64

Eq. 9 This result is then zero centered by subtracting the mean brand effect of all brands, which is 0.31. One can derive this value by calculating the average of column 1 in table 13 and subtracting this from all brand effects, after filling in the new brand effect for KPN in column 1. This results in a mean centered value of: 0,334. This value is then divided by the beta for payment level (5.25). As a last step, the difference with the fictional brand (-2.48%) is calculated, which results in an EBBE level of 8.83%. The effect of this enhanced EBBE effect on the willingness to receive less salary is shown in table 16.

EBBE effect on salary

Impact on average

Dutch yearly salary Employees

Impact on hiring (18% hiring rate) Current effect EBBE KPN 4,00% € 1.440 23.000 € 5.961.600 Future effect EBBE KPN 8,45% € 3.042 23.000 € 12.593.856

Difference +4.45% € - 1.602 - € 6.632.256

TABLE 16: Effect employer branding on KPN

The results show that an increase of 20% in excitement and sincerity perceptions can lead to cost savings of over 6 million euros per year in salary spending for KPN. It does have to be mentioned that one cannot vary the excitement and sincerity levels, without affecting the beta values in the model. So, one cannot use the model to calculate and predict effects of changing attribute levels in the competitive environment. However, slightly changing two values for a single brand will not have an high impact on the model. Therefore, one can use this method to create an idea of the impact of certain changes on the EBBE level of a company. These results are rather theoretical and would be influenced by many factors in practice, buy it does give an unique insight in the effect of EBBE and why it is useful to put time and effort into twitching the employer brand personality towards

excitement and sincerity. Doing this, companies can either increase their recruitment success and/or lower salary levels.

6.2 Creating and altering brand personality