The Finance function in the digital era

The Finance function in the digital era

The purpose of this research is to investigate the relationship between transformational leadership and how employees working in the finance function perceive the outcome of

digitalisation. Various individuals working in this area at different companies were interviewed. Their answers were used to measure transformational leadership, anxiety and affinity for digitalisation and the expected outcome of digitalisation. I found no strong support

for a significant relationship between transformational leadership and expectations of a positive outcome. Also, I did not find strong evidence that transformational leadership amplified expectations in cases where individuals had some anxiety about digitalisation. I found support that transformational leadership had no effect on individuals who already had

an affinity for digitalisation. However, the generalizability of this research is limited by the small number of respondents (80). Moreover, I found contractionary results in testing for my

relations between transformational leadership and outcome expectations. This research benefits the academic field since it provides a starting point for more research of transformational leadership during digitalisation of the finance function. Thanks to my research, organisations and managers in the finance sector are now aware that they should

1/34

1. INTRODUCTION

In recent years, digitalisation of financial and administrative processes has become

increasingly more important. Universities have been changing their business programmes to include IT and prepare students, while data driven audits and big data have become more important (Cao, Chychyla and Stewart, 2015). Currently, more than 98% of information is electronic (J. Donald Warren, Jr., Kevin C. Moffitt, and Paul Byrnes, 2015). In short, more and more is being done with and by IT. The finance function is no different. Möller, Schäffer and Verbeeten (2020) already found that digitalisation is changing the finance department. To keep up with these trends, CFOs and their finance department needs to change.

Digitalisation has accelerated in the past few years (Cao et all, Vial, 2019, Möller et all, 2020). As stated by Möller et all, (2020), the digital transformation (or digitalisation) is changing the finance function. Employees need to develop new competencies to cope, and it is having a major impact on them. Employees’ expectations of the future may have been altered by changes to the finance sector and their own jobs. Meuter, Ostrom, Bitner and Roundtree (2003) already found that this digital change is not easy for everyone, it could lead to anxiety for some people. However, in other research, such as Geissler and Edison (2005), it is discussed that people with affinity for digitalisation do enjoy such change. These changes have an impact on employees. To coop with these impacts, leaders play an important role. Dimitrios Belias and Athanasios Koustelios (2014) stated that leaders have an important role in change management, more specifically in creating an atmosphere where employees dare to engage in new behaviour. This means that their leadership is important to guide these

employees. Venkatesh, Morris, Davis and Davis (2003) studied user acceptances of

technologies. They found that social influences are important in accepting technologies. These social influences change the expectations of employees. Ronning (2004) found that this is a key responsibility of managers. In this research, I will discuss the leadership style of CFOs, and examine how it influences the perception of digitalisation of finance employees. The research question that this document will answer is, “How does the leadership style contribute to employee perception of digitalisation of the finance function?” This research question will be answered with the help of three hypothesises. My first hypothesis entails Transformational

leadership tends to positively influence the expectations of employees in the finance function

and my second hypothesis is Transformational leadership has a stronger influence on

2/34

hypothesis: Transformational leadership has no positive influence on employees that have an

affinity with digitalisation.

The research is important for CFOs to understand how their leadership could influence their employees’ perception of digitalisation. As discussed in other research (Belias and Koustelios, 2014, Faupel and Süß, 2019), transformational leadership is important during change.

Transformational leadership is a style which is more focussed on a process than the outcome. Leaders adopting such a style help their employees to cope with embracing change, so it has a positive effect on employees. In addition, employees who feel anxiety about digitalisation, IT or any other related topic will benefit more from transformational leadership. This particular style helps people feel more confident about digitalisation, so it has a stronger influence on expectations. In their 2020 meta analyses, Henner Gimpel, Vanessa Graf and Valerie Graf-Drasch looked at how attitude is one of the strongest determinants of adopting new

technologies. I therefore expect that having a positive attitude will diminish the effect of transformational leadership, while having a negative attitude will increase the effect of transformational leadership.

Several companies were contacted and employees in their financial departments were interviewed for this research. These interviews were conducted using a survey about topic digitalisation. The surveys were used to measure the respondents’ affinity and anxiety, and their expectations of the results of digitalisation. Furthermore, the respondents gave some insight into their CFOs, which allowed me to distinguish transformational leaders and non-transformational leaders. In total 124 surveys were done, 80 of them were useful for my research. In my research I found contradictory evidence that a CFO with a transformational leadership style has a significant effect on the expectations of employees. This contradiction evidence is also found in for employee that feel anxious when dealing with digitalisation. I find support in literature for my hypothesises, and my results do confirm my hypothesis. However, the contradiction originated in my robustness tests. These tests do not provide evidence for my hypothesises. This occurred most likely to the small number of respondents. However, I found that transformational leadership had no significant effect on employees with an affinity for digitalisation.

My research fills a gap in the academic field, as there are few studies that focus on the effect of the CFO’s current leadership style on employees during digital transformation. Most research focusses on how digitalisation influences CFOs (or boards), or what CFOs have to deal with during digitalisation. I find supportive literature for my proposition that

3/34

transformational leadership is important for digitalisation, especially for anxious employee. But my research cannot confirm it. My research should be used as a starting point for future research on this topic. Moreover, this research contributes to the practical field. Companies can use my research to improve the process of digitalisation. Although I did not find support in my results, in describing theory (chapter 2) for my research I found supportive evidence of the proposition. Companies and CFO could use my literature research as evidence that transformational leadership plays an important role, when dealing with digitalisation.

The remainder of this report is structured as follows: In the second part, the theory used in this report is discussed, followed by the methodology in the third part. The results are then

examined in the fourth part, and the report ends with a discussion and conclusion of the results.

2. THEORY 2.1 Digitalisation

This research focusses on the change of digitalisation in the finance function. When we talk about change, literature refer it often to as “planned alterations of organizational

components” (Cawsey, Deszca and Ingols, 2016, p. 20). It is done to improve the

effectiveness of the business. According to Ronning (2004) change can and will influence organizational performance. Cawsey et all (2016) further discuss that it is important that employees understand the vision, direction and perspective of the organization and the changes. In a change, employees must alter their behaviour to make a change work. This requires convincing employees of the need of change (Ronning, 2004, Cawsey et all, 2016). In a worst case, if employees do not alter their behaviour, the change will not happen.

Although it seems important, this research does not focus on defining digitalisation. There are several definitions of digitalisation in the academic world. Leviäkangas (2016, p. 2) defined it as “the application of modern information and communication technologies”. Muktiarni, Widiaty, AG Abdullah, Ana and Yulia used the term Industry 4.0 in their research in 2019, which entails cloud computing, data analysis and the Internet of Things. Gartner, a research company, defined digitalisation as using digital technologies in business (2020). Vial (2019) looked more into expected outcomes. Digitalisation will transform business processes.

Sometimes these processes are optimised, or digitalisation leads to efficiency gains. Möller et all (2020) used their definition to include jobs, such as the finance department. The research

4/34

of Möller et al also confirms that digitalisation affects financial controllers (and thus the finance function).

In terms of my research, digitalisation can be seen as the process of changing the finance function through the use of new technologies. This is in line with the previous mentioned research, as well as with the concept of digital transformation by Vial (2019). He defined digitalisation as “a process that aims to improve an entity by triggering significant changes to

its properties through combi-nations of information, computing, communication, and connectivity technologies” (p. 118). Westerman, Bonnet and McAfee (2014) specifically

mention that the best companies use strong leadership with digital activity (p. 2). The definitions above have one thing in common, digitalisation is about change.

2.2 Expectations

As stated in the previous paragraph, change is effective when employees are convinced of the necessity of change. By convincing employees, the expectations of those employees are being managed to start using the new technology (Ronning 2004). Gimpel et all (2020) found that expectations significantly influence the process of adopting new technologies on an individual level. Which means that positive expectations are quite important to have a successful

digitalisation.

Viswanath Venkatesh, Michael G. Morris, Gordon B. Davis and Fred D. Davis (2003) defined the expectations as, “An individual believes that using the system will help him or her to attain

gains in job performance” (p. 447). This is in line with the statement that change lead to

expected improved organisational result (Ronning 2004). Venkatesh et all (2003) discusses in their research important items for digitalisation. In their discussion they mention the most important expectations. These expectations entail increased work efficiency, increased work quality and increased personal gains (such as raise or promotion). These expectations arise by being informed by others (Venkatesh et all, 2003), for example the leaders. In the research of Venkatesh et all (2003) a survey was used to measure the expectations (based on the research of Compeau and Higgings, 1995 and Compeau, Higgings and Huff, 1999).

2.3 Leadership

In the previous paragraph it is mentioned that expectations are formed by information of others. Venkatesh et all (2003) talks about social influence and digitalisation. This refers to the individual behaviour (of the employees) that is influenced by important others to start

5/34

using the new system. This is mainly important in the early stages of the new technology. The individual behaviour is influenced by three mechanism (Venkatesh et all, 2003), which compliance is mainly relevant for my research: an individual intends to use the technology, due to social pressure. Using the paraphrase of Lo, Ramayah, Wei Min & Songan (2010) leadership is deemed as a good source of influence. Thus, to achieve this social influence, the leader of the individual is the appropriate person to address the expectations of employees. Also, previous literature on digitalisation pointed out that leadership is an important aspect during such changes (Vial, 2019). Ronning (2004) mentioned that change is a key

management responsibility. Managers must recognize it and convince others of it. There needs to be a positive attitude and people needs to be convinced. In the view of Jung, Wu and Chow (2008) you have transformational leadership and transactional leadership. Two opposite styles of leadership. As Jung et all (2008) shortly describes these leadership styles.

Transformational leadership is about creating personal and professional commitment from employees. This will increase intrinsic motivation and facilitates innovation. Transactional leadership focus on the status quo and motivates employees using extrinsic rewards to increase motivation (extrinsic motivation). Using the definitions of Jung et all (2008) of the different styles of leadership, transformational leadership by a manager displays the proper characteristics to convince employees of the necessary change. This is in line with other literature such as Belias & Koustelios (2014) and Ritu Agarwal, Steven Johnson and Henry Lucas (2011). Those researches discussed the importance of transformational leadership during change. Moreover, a study by Faupel and Suß (2019) found that transformational leadership results in a more positive view of the outcome of change. Moreover, Barker (2001) states that a successful transformational leader convinces their followers (employees) to get behind the organisational goals.

In summary, theory states that expectations are influenceable by important others, and leaders can be (and must be) these important others. Moreover, transformational leadership is deemed most appropriate for organisations in change. Based on this, I propose my first hypothesis in order to answer the research question.

H1: Transformational leadership tends to positively influence the expectations of

employees in the finance function

Sally Carless and Alexander Mann discussed in their 2000 paper titled “a short measure of transformational leadership” how to measure transformational leadership. They looked at the

6/34

Global Transformational Leadership scale to measure seven concepts that transformational leaders have, or should have. These concepts are vision, staff development, supportive leadership, empowerment, innovative or lateral thinking, lead by example, and charismatic leadership. Carless and Mann translated these seven concepts into seven questions, which describe a leader’s behaviour. High scores on the questions indicates that a leader shows extensive transformational behaviour, while low scores shows the lack of transformational behaviour on those topics.

2.4 Anxiety

As stated by Ronning (2014) for a change process, employees need to be convinced to make it successful. Dealing with new technologies is not easy for everyone, and it could lead to anxiety for some people (Meuter, Ostrom, Bitner, Roundtree, 2003). Katja Gelbrich and Britta Sattler (2014) defined technology anxiety in short as “an apprehensive feeling towards a new

technology”. Gelbrich and Sattler (2014) discussed that anxiety can lead to stress for the

employee or it slows down convincing employees of the need for change.

However, in the model developed by Venkatesh et all (2003), called the UTAUT, the authors proposed and found evidence for the hypothesis that anxiety does not have a significant influence on behaviour in the context of new technologies. Gimpel et all (2020) found that there is no significant relationship between anxiety and behavioural intention, but there is between attitude and behavioural intention. These different authors hypothesised that anxiety has limited influence on the usage or the outcome expectations of digitalisation.

In short, anxiety slows the digitalisation process. Therefore, for a smooth digitalisation, reducing anxiety is important. The research of Gimpel et all and Venkatesh et all did not consider transformational leadership. According to Jung (2008) transformational leadership creates an atmosphere in which employees must feel safe in trying out new innovations. Moreover, transformational leadership creates a vision (Belias & Koustelios, 2014) that enables employees. It seems that these characteristics of transformational leadership can reduce the impact of the anxiety and improves expectations. This leads to my second hypothesis:

H2: Transformational leadership has a stronger influence on employees who are more

7/34

2.5 Affinity

Gary L. Geissler and Steve W. Edison (2005) researched affinity for technology and market mavens. They found that those that have affinity for technology have a positive influence on others. In their paper they stated, “Affinity for technology can be defined as the degree to

which an individual likes or looks forward to learning about and being involved with new technology” (p. 77).

Transformational leadership focus on enabling employees and increasing intrinsic motivation (Belias and Koustelios, 2014, Jung, 2008, Faupel and Suß, 2019 and Barker, 2001). An employee with affinity is already motivated and eager to be involved with the digitalisation. This leads to my third and last hypothesis:

H3: Transformational leadership has no positive influence on employees that have an

affinity with digitalisation.

I propose this hypothesis on the basis that an employee with affinity is already convinced about the benefits of digitalisation. A transformational leader will not convince the employee more (the employee already supports the digitalisation goal). The positive characteristics of transformational leadership do not have an added value for the employee.

Edison and Geissler (2005) used their previous research from 2003, in which they created an affinity scale based on questions to measure affinity. In their 2005-article, this affinity is measured by focussing on using technologies.

3. RESEARCH METHOD 3.1 Study design

The main data was gathered by surveying financial professionals in companies. Various students at the Faculty of Economics and Business surveyed professionals for their own research. The outcomes of these surveys were collected in one large data pool. Every student used the same survey which entailed the same closed-ended questions. Using the survey, data was gathered about the companies (industry, number of employees, growth expectations), digitalisation (strategy, personal anxieties and expectations) and the leadership style.

The survey was created using previous research but adapted to the finance & control function. Questions that focussed on affinity, anxiety, outcome expectations and leadership style were

8/34

measured with a seven-point Likert scale, with 1 representing completely disagree, 4 neutral and 7 completely agree. Other aspects, such as age, title and education, were categorised into fixed options. The survey questions and answer options are included in Appendix A. For all the used surveys, I performed factor analysis to test whether questions of a single subject (the variables that I want to test) form a coherent construct. I consider a factor loading beneath 0.5 low. Furthermore, I checked the eigenvalue of those factor analyses. I expect an eigenvalue higher than 1 for the first factor loading and an eigenvalue lower than 1 for the other factor loadings. I communicate the eigenvalues of the first factor and the second factors in the tables (eigenvalues decrease with new every factor).

Most of the surveys were done by telephone and videocalls, and a few during face-to-face meetings. These interviews involved closed-ended questions. Although there was room for explaining or elaborating during the interviews, the respondents had to choose a fixed option that best described their situation. This made it possible to use the surveys of different

students, although they had all different research goals. Moreover, the closed-ended questions method made it possible to compare the different answers of the respondents.

3.2 Sample

In total, 124 observations were taken. Of these 124, 80 observations were usable. The 44 missing observations were dropped due to incompleteness of the other answers. The results are illustrated in Table 1.

TABLE 1

Descriptive statistics of the sample

Variable Obs Mean Std.Dev. Min Max

Age 100 2.93 1.10 1 5 Gender 103 1.17 .37 1 2 Title 91 2.38 1.61 1 5 Education 100 3.83 .87 2 5 FTE_FC 85 674.26 2,853.70 1 20,000 FTE_Tot 85 10,136.56 45,176.61 1 375,000

Table 2 shows the main characteristics of the sample. Most respondents were between 40 and 60 years old and male. The majority held a managerial role, the commonest being CFO (or equivalent). Most of the respondents had a bachelor or higher degree. The respondents worked in various industries. Most respondents worked in the finance, insurance and real estate industry, followed by the service industry.

9/34

TABLE 2

Sample characteristics of the sample

Freq. Percent Cum.

Age Less than 30 16 16.00 16.00 30-40 13 13.00 29.00 40-50 36 36.00 65.00 50-60 32 32.00 97.00 Over 60 3 3.00 100.00 Gender Male 86 83.50 83.50 Female 17 16.50 100.00 Title CFO 42 46.15 46.15

First level below the CFO (e.g., Head of Controlling or Accounting)

17 18.68 64.84

Second level below CFO 6 6.59 71.43

Below the second level, but management responsibility

7 7.69 79.12

Employee in the finance and control function

19 20.88 100.00

Education

Vocational education or equivalent 5 5.00 5.00

Bachelor’s degree or equivalent 32 32.00 37.00

Master’s degree or equivalent 38 38.00 75.00

Post academic 25 25.00 100.00

Industry

Agriculture, Forestry, and Fishing 4 4.40 4.40

Construction 1 1.10 5.49

Manufacturing 7 7.69 13.19

Transportation, Communications,

Electric, Gas, and Sanitary Services 7 7.69 20.88

Wholesale Trade 6 6.59 27.47

Retail Trade 7 7.69 35.16

Finance, Insurance, and Real Estate 19 20.88 56.04

Services 13 14.29 70.33

Public Administration 3 3.30 73.63

10/34

3.3 Variable measurement: dependent variable

As described previously, the research investigates individual outcomes. To measure these outcomes, I used the same survey questions that Venkatesh et all (2003) used in their research for user acceptance of technologies. They based their survey on the social cognitive theory of Compeau and Higgings (1995) and Compeau et all (1999). These questions are shown in Table 3 (and Appendix A).

In the questionnaire the respondents were given six possible benefits for their own working activities. Each of these benefits was rated on a seven-point Likert scale. Using these results, a mean was calculated, this can be seen in the top of Table 3 (descriptive statistics of outcome expectations). Furthermore, a factor analysis was performed. These factor loadings are shown in Table 3. As can be seen, all factor loadings are above 0.637 and an eigenvalue larger than 1, indicating that the variables are answered in correspondence with each other. The

eigenvalue of the other factor loadings are below 0.42 (indicating there is no other construct). In short, the measured individual expectations (Ind_expectations) is formed by a coherent construct, which indicates how positive the respondents think about the possible outcomes.

TABLE 3

Descriptive statistics of individual outcome expectations

Variable Obs Mean Std.Dev. Min Max

Ind_expectations 82 4.7 1.53 1 6.86

Factor analysis of individual outcome expectations

To what extent do you agree or disagree that making greater use of digital technologies would enable me to

Factor …execute my working procedures more efficiently. 0.643 …coordinate work across processes, units and levels more easily. 0.637 …recognise changes that are relevant for our business much earlier. 0.743 …understand influences on and interdependencies of our business more deeply. 0.778 …contribute to our firm's business development more effectively. 0.765

…focus more on value-creating activities. 0.646

Number of observations = 87, Eigenvalue = 2.981, Eigenvalue of the second factor = 0.419 I also looked at questions about company expectations and combined these with the individual outcomes. Table 4 shows these questions and the factor loadings. The questions of the

company outcomes are based on the expected gains for the company. As change theory dictates, change is expected that performance improves (Ronning, 2014) and should keep the long-term viability of an organisation (Möller et all, 2020). The questions of company outcomes are based on these assumptions.

11/34

The factor loading of the last question, about new business, is below 0.5. All other loadings are higher than 0.6. The eigenvalue of the first factor is also quite high, indicating that the construct explains more than a single question. However, as the last question has a low factor, I will not use it for my general testing. I will use this combination for the robustness testing.

TABLE 4

Factor loadings of all outcome expectations To what extent do you agree or disagree that making greater use of digital technologies would enable me to

Factor …execute my working procedures more efficiently. 0.619 …coordinate work across processes, units and levels more easily. 0.632 …recognise changes that are relevant for our business much earlier. 0.735 …understand influences on and interdependencies of our business

more deeply.

0.776 …contribute to our firm's business development more effectively. 0.759 …focus more on value-creating activities. 0.638 To what extent do you agree or disagree with the following

statements? Regarding our organisation (the organisational unit I provide services to), I believe that making greater use of digital technologies within the finance & control function

…is important for its survival 0.625

…would improve its performance 0.602

… would bring new business 0.442

Number of observations = 87, Eigenvalue = 3.857, Eigenvalue of the second factor = 0.626 3.4 Independent variable: Transformational leadership

The questions from the Carless and Mann paper (2000) were used to measure leadership. Carless and Mann used several theories to develop a quick test to see how transformational leaders are.

My research focus on the CFO as the leader of change. However, not every company has a CFO, or is the CFO responsible for digitalisation of the finance function. In those situations, the equivalent of the CFO must be used. Thus, when I mention the CFO, it might also be the equivalent of CFO. If the respondent was the CFO, the CEO (or someone equivalent) had to be rated.

The respondents had to rate their CFO how they behaved on several topics. Seven of these topics came from the Carless and Mann paper to measure transformational leadership. To check if these seven topics form a coherent construct, factor analysis was performed. The factor analysis showed that these items were related, the factor loadings are extremely high (minimum of 0.868) and the eigenvalue was above 5. The eigenvalue of the second factor

12/34

(and others) are below 1, which indicates that there is no other construct. See also Table 5. Using the results, I calculated the average mark for transformational leadership.

In the view of Jung et all (2008) transformational leadership and transactional leadership are two opposites. For this research I interpret that at a certain point, a leader is transformational. A new dummy variable (Transf_dummy) was created of the variable from transformational leadership. The mean of transformational leadership (Transf_ldr in Table 5, not being

dummy) is 4.7 with a standard deviation of 1.53. In my hypothesis testing, the score of a CFO who is a transformational leader on the seven question should be at least 5 (first positive score above neutral). The CFO leader is therefore only considered transformational if the score is above 5, the dummy variable will be 1 in those cases.

TABLE 5

Descriptive statistics of transformational leadership

Variable Obs Mean Std.Dev. Min Max

Transf_ldr 82 4.7 1.53 1 6.86

Descriptive statistics of the dummy variable of transformational leadership

Transf_dummy

Freq. Percent Cum.

0 37 45.12 45.12

1 45 54.88 100.00

Factor analysis of questions about transformational leadership

The CFO of our organisation [if you are the CFO refer to your superior]… Factor …communicates a clear and positive vision of the future. 0.894 …treats staff as individuals, supports and encourages their development. 0.923

…gives encouragement and recognition to staff. 0.868

…fosters trust, involvement and cooperation among team members. 0.902 …encourages thinking about problems in new ways and questions assumptions. 0.886 …is clear about his/her values and practices what he/she preaches. 0.886 …instils pride and respect in others and inspires me by being highly competent. 0.945 Number of observations = 82, Eigenvalue = 5.682, Eigenvalue of the second factor = 0.133

However, other literature states that a leader can exhibit transformational leadership as well as transactional leadership. The research of Carless and Mann (2000) does imply such opposing view of leadership styles. The developed survey by Carless and Mann is about to show how often a leader shows transformational leadership behaviour. This could mean that a dummy variable for transformational leadership is not applicable and I should use the mean variable (Transf_ldr) to measure transformational leadership.

13/34

To correct for these different views, I will perform robustness tests with the real average variable (Transf_ldr). This way, I can use the dummy variable in line with the view of Jung et all (2008) and see if my results still holds when using the real variable (in line with Carless and Mann, 2000).

3.5 Moderating variable: Anxiety

Six questions were asked to measure the anxiety of the respondents. The questions are based on the research of Venkatesh et all (2003) and are adapted to digitalisation.

Exploratory factor analysis of the questions showed two of those questions were insufficiently related to the others, and were dropped when calculating anxiety. This was due to the

questions asked. The four questions I keep relate to the personal feelings and competences about digitalisation. The first question dropped was about the personal expectation of the company possibly failing because of digitalisation. The second question dropped was about data protection issues. The questions used are shown in Table 6 with the corresponding factor loadings. The eigenvalue was higher than 1 (while the eigenvalues of the other factors were below 1) indicating that these questions form a coherent construct.

TABLE 6

Descriptive statistics of anxiety

Variable Obs Mean Std.Dev. Min Max

Anxiety 88 2.11 1.05 1 6

Descriptive statistics of the dummy variable of anxiety Anxiety_dummy Freq. Percent Cum.

0 72 81.82 81.82

1 16 18.18 100.00

Factor analysis of questions about anxiety

To what extent do you agree or disagree with the following statements? Factor I get an uncomfortable feeling when I think of using digital technologies 0.768 I am concerned that I am not fully qualified to handle digital technologies 0.775 Due to digital technologies, I am worried that my personal competencies will be less

needed.

0.733 Due to digital technologies, I am worried that my actual job will be less relevant. 0.726 Number of observations = 88, Eigenvalue = 2.256, Eigenvalue of the second factor = 0.156

To test the influence of transformational leadership on anxiety, the results needed to be divided into respondents that were anxious and non-anxious about digitalisation. Another dummy variable was created for this. However, almost all respondents scored low on this

14/34

rating (mean 2.11 with a 95% confidence interval of max. 2.34). For this research, somebody is considered to be anxious if the score is higher than 3. A score of 3 is the first score below neutral. This led to 72 observations of no anxiety and 16 observation of anxiety. See the summary in Table 6.

3.6 Moderating variable: Affinity

The affinity was measured by four questions. These questions are based on the research of Geissler and Edison (2005). They developed a survey for their research to find consumers with an affinity for technologies. Their survey is adapted to digitalization for this research. Exploratory factor analysis showed that one question scored low. This question was about how informed the respondent was about digital training courses at the respondent’s company, while the other three questions were related to the respondent’s actions. When calculating the average affinity, this low-rating question was dropped. The eigenvalue is 1.767, while the second eigenvalue is below 0. The used construct is thus valid for calculating affinity. Just as in the case of anxiety, to test for the influence of transformational leadership, the answer had to be divided. Thus, a dummy variable was also created for affinity. The mean of affinity is 5.2. Following the same logic as in the calculation of transformational leadership, someone with a score of at least 5 (first positive score above neutral on the Likert scale) is considered as having affinity with digitalisation. Table 7 shows a summary of the real variable (top), of the dummy (in the middle) and the factor loadings (bottom of Table 7).

TABLE 7

Descriptive statistics of affinity

Variable Obs Mean Std.Dev. Min Max

Affinity 88 5.24 1.14 2 7

Descriptive statistics of the dummy variable of affinity Affinity_dummy Freq. Percent Cum.

0 28 31.82 31.82

1 60 68.18 100.00

Factor analysis of questions about affinity

To what extent do you agree or disagree with the following statements? Factor I frequently inform myself about trends and developments related to digital

technologies.

0.785 I regularly look for opportunities to develop knowledge and skills in the field of

digital technologies.

0.826 I enjoy dealing with new technologies and innovations. 0.685 Number of observations = 88, Eigenvalue = 1.767, Eigenvalue of the second factor = -0.087

15/34

3.7 Control variables

I checked several variables during my calculations. These control variables were included to see if they do not interfere with the expectations of the respondents. Answers that were neglected were not considered for the calculations.

The gender of the respondent was the first variable. During the investigation, respondents could choose between male or female. The answers were stored as a binary variable (male = 1 female = 2). The position (hierarchical level) in the company was the second variable that was tested. The answers in the survey were already categorised, with a low score for a high

position and a high mark for a low position (1 = CFO and 5 = non-managerial employee). I also used level of education as a control variable. I added the variable education to control any chances that different educated people would be less (or more) able to foresee what changes would bring, leading to more (or less) anxiety or lower (higher) expectations. Education was categorised from 1 as only high school to 5 for post-academic education.

The survey with the possible answers and how they were labelled in the datafile is shown in Appendix A. The descriptive statics of the control variables are shown in Table 1 and Table 2. For my second hypothesis, I also added the control dummy variable Low_affinity. Low

affinity is calculated using the same method as Affinity_dummy but applies on scores below 4.333 on Affinity.

3.8 The model

Using statistical analysis, the leadership style of the CFO and the individual outcomes were tested. STATA was used for the data calculations. In STATA, all 124 observations were used for the calculations. If STATA found that an observation was not useable, it would drop them automatically. Linear regression was used for determining any relationships.

My model to test my first hypothesis:

Ind_expectations = α + β1 ∗ Transf_dummy + γij ∗ (control variables)ij + ℇij

To test for hypothesis one, a simple regression analysis was performed using all variables, followed by an analysis using only the variables shown in the formula.

To investigate the second and third hypothesis, I include two interaction terms. For hypothesis two the dummy variable ‘transformational leadership’ was combined with the dummy

variable ‘anxiety’. The combined and single variables were both taken into account in the calculations. For the second hypothesis, this led to the interaction Transf_dummy

16/34

*Anxiety_dummy. I also added a second control variable ‘low_Affinity’. The tested model

became:

Ind_expectations = α + β1 ∗ Transf_dummy + β2 (Transf_dummy

*Anxiety_dummy) + β3 ∗ Anxiety_dummy + β4 ∗ Low_Affinity + γij ∗ (control variables)ij + ℇij

The same method also applied to hypothesis three, which combined the dummy variable ‘transformational leadership’ with the dummy variable ‘affinity’. The interaction term used was therefore Transf_dummy *Affinity_dummy. This led to the following model:

Ind_expectations = α + β1 ∗ Transf_dummy + β2 (Transf_dummy *

Affinity_dummy) + β3 ∗ Affinity_dummy + γij ∗ (control variables)ij + ℇij

In the models above, γij ∗ (control variables)ij relates to all used control variables (age, gender,

title and education) as a separate part of the model. For readability, these variables are not separately shown in the models.

When a regression was positive, the regression was tested again for robustness, using the robust option in STATA. The results were only taken into account if the p-value was lower than 0.05 (my minimum threshold for significance).

4. RESULTS 4.1 Descriptive statistics

Table 8 shows the statistics of the variables before any changes were made. Only the variables used in the results are shown in the table. For the hypothesis tests, the mean individual

expectation is 5.54 with a small standard deviation (0.87), meaning the spread is not large. However, the information about the FTEs is more limited (due to less observations) and widely spread. The standard deviation is larger than the mean, indicating a large spread.

TABLE 8 Descriptive Statistics

Variable Obs Mean Std.Dev. Min Max

Ind_expectations 80 5.54 .87 3 7

Transf_ldr 80 4.65 1.52 1 6.86

Age 80 2.98 1.10 1 5

FTE_FC 75 578.41 2,817.97 1 20,000

17/34

TABLE 9

Correlation Matrix of regression variables

Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (1) Ind_ expectations 1.00 (2) Transf_ ldr 0.14 1.00 (3) Affinity 0.50*** -0.04 1.00 (4) Anxiety -0.15 -0.01 -0.34*** 1.00 (5) Transf_ dummy 0.38*** 0.80 *** 0.15 -0.05 1.00 (6)Affinity_ dummy 0.50*** -0.02 0.77*** -0.42 ** 0.19* 1.00 (7)Anxiety_ dummy -0.26** -0.03 -0.28** 0.86 *** -0.10 -0.43 *** 1.00 (8) Age -0.26** 0.11 -0.19* 0.08 0.14 -0.19* 0.07 1.00 (9) Gender -0.08 0.05 -0.13 0.05 0.04 -0.12 0.08 -0.05 1.00 (10) Title -0.03 -0.10 -0.12 0.16 -0.14 -0.20* 0.16 -0.29*** 0.30*** 1.00 (11) Education 0.04 0.02 0.18 -0.14 0.08 0.19* -0.13 0.20* -0.20* -0.19* 1.00 (12) FTE_FC 0.06 0.03 -0.13 0.05 0.02 -0.07 -0.04 -0.04 -0.05 0.09 0.01 1.00 (13) FTE_Tot -0.16 0.04 -0.08 0.16 -0.10 -0.20* 0.28** 0.00 0.27** 0.17 0.18 0.05 1.00 *** p<0.01, ** p<0.05, * p<0.1

Table 9 shows the pairwise correlation matrix. As can be seen, the dummy transformational leadership (Transf_dummy) is positive significant in respect of the individual expectations (Ind_expectations), but the real value (Transf_ldr) is not significant. The same applies to Affinity and Affinity_dummy, while Anxiety (not significant) and Anxiety_dummy have a negative relationship with the individual outcomes. Furthermore, we can see a negative but significant relationship between Age and the individual outcomes.

4.2 Hypothesis testing

Three hypotheses were developed for testing. The table below shows the regression analysis of hypothesis one. This hypothesis means that transformational leadership positively

influences the outcome expectations.

As can be seen in Table 10, being a transformational leader (Transf_dummy) has a positive coefficient, indicating that a transformational leader positively influences the individual outcomes (strongly significant). Moreover, age has a negative coefficient (significant), which indicates that younger audiences expect more benefits from digitalisation.

18/34

TABLE 10

Robust Linear regression of all variables (hypothesis 1) Ind_expectations Coef. St.Err.

t-value

p-value [95% Conf Interval] Sig

Transf_dummy 0.727 0.188 3.87 0.000 0.352 1.101 *** Age -0.268 0.089 -3.01 0.004 -0.445 -0.091 *** Gender -0.242 0.292 -0.83 0.410 -0.825 0.340 Title -0.017 0.054 -0.31 0.756 -0.124 0.090 Education 0.050 0.094 0.53 0.598 -0.137 0.237 Constant 6.073 0.551 11.02 0.000 4.975 7.170 ***

Mean dependent var 5.535 SD dependent var 0.869

R-squared 0.257 Number of obs 80.000

F-test 4.334 Prob > F 0.002

Akaike crit. (AIC) 191.694 Bayesian crit. (BIC) 205.986

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

TABLE 11

Robust linear regression of anxiety (hypothesis 2) Ind_expectations Coef. St.Err.

t-value

p-value [95% Conf Interval] Sig

Transf_dummy 0.384 0.168 2.29 0.025 0.049 0.719 ** Anxiety_dummy -0.909 0.362 -2.51 0.014 -1.630 -0.188 ** Transf_dummy# Anxiety_dummy 1.178 0.439 2.69 0.009 0.303 2.054 *** Low_Affinity -0.894 0.288 -3.10 0.003 -1.470 -0.319 *** Age -0.166 0.078 -2.12 0.038 -0.322 -0.010 ** Gender -0.044 0.225 -0.20 0.844 -0.493 0.404 Title 0.018 0.054 0.33 0.741 -0.090 0.126 Education -0.116 0.091 -1.27 0.209 -0.298 0.066 Constant 6.469 0.470 13.78 0.000 5.533 7.406 ***

Mean dependent var 5.535 SD dependent var 0.869

R-squared 0.460 Number of obs 80.000

F-test 6.238 Prob > F 0.000

Akaike crit. (AIC) 172.208 Bayesian crit. (BIC) 193.647

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

Table 11 contains the hypothesis two test results. This hypothesis tested whether anxious respondents benefit from transformational leadership, and thereby have a higher score on the individual outcomes. The Anxiety_dummy is significant (p < 0.05) and the product of anxiety and transformational leadership (Transf_dummy #Anxiety_dummy) is significant and

positive. This means that I find that anxiety significantly influences the individual expectations, but more important I did find that transformational leadership has a greater impact on individual expectations of anxious employees. The interaction is quite significant

19/34

and strong (coefficient over 1). Low affinity (Low_Affinity) was added as a control variable. This made it clear whether low affinity, instead of anxiety, might affect individual outcomes. Because I test the interaction of transformational leadership and anxiety, my result still holds.

TABLE 12

Robust linear regression of affinity (hypothesis 3) Ind_expectations Coef. St.Err.

t-value

p-value [95% Conf Interval] Sig

Transf_dummy 0.965 0.369 2.61 0.011 0.228 1.701 ** Affinity_dummy 0.997 0.351 2.84 0.006 0.298 1.696 *** Transf_dummy# Affinity_dummy -0.511 0.412 -1.24 0.219 -1.332 0.310 Age -0.177 0.099 -1.79 0.077 -0.373 0.020 * Gender -0.196 0.215 -0.91 0.367 -0.625 0.234 Title 0.023 0.062 0.37 0.714 -0.100 0.145 Education -0.053 0.087 -0.61 0.546 -0.227 0.121 Constant 5.421 0.659 8.22 0.000 4.107 6.735 ***

Mean dependent var 5.535 SD dependent var 0.869

R-squared 0.406 Number of obs 80.000

F-test 6.589 Prob > F 0.000

Akaike crit. (AIC) 177.818 Bayesian crit. (BIC) 196.874

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

For the third hypothesis, a test was performed to see whether employees with affinity do not benefit from transformational leadership. Table 12 show the robust regression results. The table shows that the affinity dummy has a strong coefficient (significant), which increases the individual outcome expectations. However, the variable Transf_dummy#Affinity_dummy, has no significance. This variable is for those with affinity and who are under a

transformational leader. This means that a transformational leader has no effect on the outcome expectations of an employee with affinity.

4.3 Robustness checks

To check the results of my analysis, I considered several different options. First, I checked different outcome expectations (I change the depended variable). Then I changed several dummy variables for the original data.

Change the dependent variable

For each of the hypotheses, I changed the outcome variable to general outcomes (individual and company expectations combined, model 1, 3 and 5) and to company expectations (model 2, 4 and 6). See also Table 4 for the used questions (Table 4 completely is the combined

20/34

results, while the bottom three questions of Table 4 entail the company expectations). Table 13 shows the six outcomes of my Robustness testing. Model 1 and 2 test the alternatives for hypothesis one. This alternative continues to support the results found. Transformational leadership still improves outcome expectations. For hypothesis two (model 3 and 4), the results are less conclusive. When dealing with anxiety, transformational leadership has no significance influence on the company expectations, but they still hold for the combined expectations. The outcome expectations proposed by my third hypothesis remain the same. Transformational leadership does not influence outcome expectations.

In the robustness test, the variable Age loses it significant (except in model 1). In the original calculations, Age was significant with hypothesis 1 and 2.

TABLE 13

Robust linear regressions with alternative depended variables (hypothesis 1, 2 and 3)

(1) (2) (3) (4) (5) (6)

outcomes com_out outcomes com_out outcomes com_out Transf_dummy 0.653*** 0.514** 0.343** 0.266 0.744** 0.299 (0.179) (0.231) (0.161) (0.247) (0.353) (0.409) Age -0.211** -0.093 -0.111 0.004 -0.114 0.015 (0.080) (0.101) (0.074) (0.106) (0.082) (0.099) Gender -0.229 -0.145 -0.055 -0.038 -0.183 -0.126 (0.287) (0.353) (0.201) (0.278) (0.203) (0.280) Title -0.024 -0.040 0.012 -0.000 0.020 0.013 (0.052) (0.073) (0.052) (0.076) (0.055) (0.072) Education 0.051 0.059 -0.104 -0.080 -0.043 -0.020 (0.092) (0.131) (0.091) (0.137) (0.091) (0.136) 1.Anxiety_dummy -0.743** -0.420 (0.328) (0.381) Transf_dummy#Anxiety_dummy 0.997** 0.639 (0.409) (0.489) Low_Affinity -0.937*** -1.030*** (0.257) (0.308) 1.Affinity_dummy 0.923*** 0.777** (0.299) (0.340) Transf_dummy#Affinity_dummy -0.311 0.100 (0.394) (0.477) _cons 5.872*** 5.383*** 6.236*** 5.713*** 5.209*** 4.731*** (0.497) (0.709) (0.451) (0.742) (0.542) (0.703) Obs. 80 81 80 81 80 81 R-squared 0.223 0.083 0.436 0.215 0.387 0.199

Standard errors are in parenthesis

21/34

Change the independent variable

TABLE 14

Robust linear regressions with alternative independent variables (hypothesis 1, 2 and 3)

(1) (2) (3) (4) (5) Ind_expectations H1 H2 H3 H2 H3 Transf_ldr 0.099 0.066 0.318 -0.082 0.727** (0.065) (0.058) (0.271) (0.125) (0.354) Age -0.248*** -0.135 -0.134 -0.136 -0.117 (0.090) (0.083) (0.092) (0.084) (0.090) Gender -0.182 -0.170 -0.211 -0.180 -0.121 (0.327) (0.279) (0.236) (0.322) (0.294) Title -0.038 0.031 0.025 0.015 0.004 (0.053) (0.058) (0.059) (0.057) (0.055) Education 0.071 -0.100 -0.064 -0.088 -0.069 (0.106) (0.102) (0.096) (0.107) (0.089) Anxiety_dummy -1.702 (1.283) c. Transf_ldr #Anxiety_dummy 0.283 (0.250) Low Affinity -1.077*** -1.105*** (0.312) (0.323) Affinity_dummy 2.079 (1.415) c. Transf_ldr #Affinity_dummy -0.247 (0.277) Anxiety -0.448 (0.295) c.Transf_ldr#Anxiety 0.093 (0.056) Affinity 0.897*** (0.297) c.Transf_ldr#Affinity -0.109* (0.056) _cons 5.843*** 6.362*** 4.211*** 7.020*** 0.827 (0.662) (0.655) (1.321) (0.993) (1.842) Obs. 80 80 80 80 80 R-squared 0.117 0.345 0.331 0.316 0.342

Standard errors are in parenthesis

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

In Table 14, I show the results for the robustness test for alternative independent variables. First, I changed the dummy variable of transformational leadership into the real value (Transf_ldr), those results are visible in model 1, 2 and 3. Furthermore I used the average affinity and anxiety variables (model 5 and 6).

The table shows that the real values of hypothesis 1 (model 1) are not significant. This indicates that on a sliding scale (using the real value) transformational leadership has no influence on individual outcomes. For the interaction terms for hypothesis two and three

22/34

(model 2 and 3), there is also no relationship. Using the source variables for the dummy variables appears to invalidate hypothesis one (model 1) and hypothesis two (model 2) and confirm hypothesis three (model 3).

For using the real value of anxiety (model 4), the coefficient of transformational leadership becomes even negative (and remains non-significant). Support for hypothesis two with only real values (model 4) is not found. In model 5 I find two significant relations, between

transformation leadership and the individual outcomes, and affinity and individual outcomes. An explanation for the difference of results (compared to our testing in the previous

paragraphs) might be found in Graph 1. This graph shows the real values of transformational leadership on the y-axis and individual outcomes on the x-axis. Using STATA, I added a fitting value line. With the help of this line, it becomes visible that many observations are outliers, they are not even close to the fitting line. If there are many outliers, a relation cannot be found, it will invalid any regression analysis.

GRAPH 1

Graphical overview of observations of individual expectations and transformational leadership

To explain the found remarkable, significant results in model 5, which did not exist when I tested in the previous paragraph, I present Graph 2. This graph shows the observations of transformational leadership (Transf_ldr) and affinity. In this graph, it is clearly visible that

23/34

between x=0, x=4 and y=0, y=4, there are almost no results, while in the remainder of the graph there are observations.

GRAPH 2

Graphical overview of observations of affinity and transformational leadership

5. CONCLUSION AND DISCUSSION 5.1 Discussion of the results

The research question of this paper is, “How does the leadership style contribute to employee perception of digitalisation of the finance function?” This question was answered using the three hypotheses.

The first hypothesis (Transformational leadership tends to positively influence the

expectations of employees in the finance function) and the second hypothesis

(Transformational leadership has a stronger influence on employees who are more anxious about digitalisation) are not proven to be completely true.

I expected a relation between expectations and transformational leadership. In other literature, such as Faupel and Suß (2019), Belias and Koustelios (2014) similar relations were found. For the first hypothesis, my regression analysis found a significant positive relationship. This

24/34

confirms my expectations that if a CFO has a transformational leadership style, he or she positively influences the digitalisation expectations of his or her employees. However, when I did my robustness test and used a sliding scale for transformational leadership, I did not find any support for a relationship. The coefficient dropped massively and no significance at all was found for this relation. This could mean that my found results might falsely support my first hypothesis. One of the reasons I performed a robustness test, was due to the Carless and Mann paper (2000). But, in that paper, the authors mention the notion that transformational leadership helps amongst other with group goals (page 390). An explanation of the different result of hypothesis one might be found in my observations. For the result only 80

observations were useable. The plot of those variables (see Graph 1) shows the values are quite widely spread, there are many outliers. In short, with my found results, I cannot confirm hypothesis one.

For the second hypothesis, I expected that transformational leadership would help anxious employees with better outcome expectations. I find support in literature for the relation

between transformational leadership and anxiety (Jung, 2008, Belias and Koustelios, 2014). In my results it is shown that the interaction term of anxiety and leadership has a significant positive coefficient. This indicates that transformational leadership is important for those who are anxious when it comes to digitalisation. This is as expected and is supported by Jung (2008) and Belias and Koustelios (2014). However, with checks on the robustness of independent variables, just like hypothesis one, I could not find more support for this hypothesis. The coefficient of the interaction dropped and the significance went from remarkably high (significant on p<0.01 when tested) to non-existence (robustness test). The same explanation can be used for this altered result as for hypothesis one: when using the real value, there are many outliers which prevents any regression to be possible. Like hypothesis one, I think with my found results, I cannot positively be sure that my hypothesis is true. Important side note for my second hypothesis: in some literature (Venkatesh et all, 2003, Gimpel et all, 2020) they also found no relation between anxiety and outcome expectations. By formulating my third hypotheses, I did not expect that transformational leadership had an affect on affine employees. I formulated these expectations using Jung (2008), Faupel and Suß (2019), Belias and Koustelios (2014) and Barker (2001). In the testing I performed, I did not find significant evidence that the outcome expectations of employees with an affinity for digitalisation are higher if the CFO has a transformational leadership style. Hypothesis three (Transformational leadership has no positive influence on employees that have an affinity

25/34

with digitalisation) is true. The testing of hypothesis three confirmed my expectations, and

this result is supported by literature. Hypothesis three was tested both for the dummy variables (for affinity and transformational leadership) and real average score during robustness testing. The robustness testing confirms hypothesis three. However, as I use the real values for both transformational leadership and affinity, the interaction between those two variables become significant with a negative coefficient. This indicates that transformational leaders have a negative impact on employees with affinity. This result is remarkable, since I did not expect this, nor is it supported in literature. An explanation for this is provided in Graph 2. This shows that in my observations a lot of employees showed affinity, with leaders that both show and do not show transformational behaviour. While there are no employees with a low score on affinity and a leader that does not show transformational leadership. Since I cannot find any explanation in literature for this, most likely this occurred just for my

observations. I expect that when the survey is applied again to other respondents, this

situation will not occur again. Moreover, the found relation in the robustness testing does not alter the conclusion of hypothesis three. I did not find any positive influence of

transformational leadership on outcome expectations of employees with an affinity for digitalisation.

In all hypothesis testing, other variables such as age, gender, title (hierarchical level) and education are taken into account. In my testing, only the factor age seemed to be significant negatively related to the outcome expectations. This is in line with the research of Venkatesh et all (2003). They also found that gender and age did have an impact on the outcome

expectations. I did not find such connection for gender, but that is most likely due to the 84% of my observations are male.

However, in the robustness tests, age did lose some significance in the alternative regressions (except for model 1 of Table 13 and Table 14). Also important finding from the robustness test (which were not mentioned before) is that by using the alternative variable combined outcomes, my results confirms my three hypothesis., When I only use the variable expected company outcomes by employee, I find supportive evidence for hypothesis one and three, but hypothesis two is not supported anymore.

5.2 Limitations

This research was performed during the Covid-19 pandemic. Although this pandemic did not immediately interfere with the outcome of this research, I noticed that several respondents

26/34

said they were slightly more negatively influenced because of the uncertainty of the virus. Moreover, the data was gathered by multiple students, all with different research subjects. This could mean every student had a slightly different understanding of the surveyed questions. However, I do not expect vastly different outcomes if there had not been a

pandemic or all the interviews had been performed by the same students. The questions of the survey were set up in a manner that external influences or understandings would not

significantly affect the answers provided, if at all. Moreover, using factor analysis I tried also to compensate for questions that might be interpreted differently.

During the surveys, I noticed that the respondents were not very keen on providing ‘negative’ feedback on their CFOs. This is quite understandable, as social expectations might prevent people criticising their leaders. However, in my research, a CFO had to score highly (5+) to be considered a transformational leader. This mitigated the ‘social expectations’ problem. This research found supportive evidence for all hypotheses. However, the tested sample was quite small (80 respondents). Moreover, when I performed robustness testing, hypothesis one and two were not supported anymore.

As discussed previously, I used a dummy variable for transformational leadership. Although this method might find some support in literature (Jung et all, 2008), other research (Carless and Mann, 2000) found that a leader shows sometimes more or less forms of transformational leadership (it is a sliding scale). I tried to compensate for these different views using

robustness testing. However, my robustness testing showed different results compared to my normal testing. I think to compensate this in future research, transformational leadership should be measured more properly. One way to do so, is to ask multiple respondents with the same CFO. This might also mitigate the social expectations problem of the previous

paragraph. But more important, multiple respondents provide a better insight in the

transformational behaviour of these leaders in the finance function. This will lead to a more trustworthy measurement of the CFO (or their equivalent). And ultimately, I expect that this mitigates the differences between a dummy variable and the sliding scale.

Another limitation of this research has to do with the surveys. The surveys used for this report were quite long and extensive. Sometimes, the interviews tired the respondents. For future research, surveys should focus only on leadership, expectations and personal feelings (affinity and anxiety) and be conducted among more respondents. This also makes it easier for

27/34

any researcher present during the survey, respondents might be more ‘honest’ about their CFOs.

5.3 Implications and future research

My research contributes to the literature in various ways. Mostly, I provided new insights into the influence of CFOs during digitalisation. Other research has already made it clear that transformational leadership is important during change. I researched the concept that transformational leadership becomes even more important for anxious employees.

Furthermore, I tested the hypothesis about transformational leadership for anxious employees and employees with affinity. As discussed previously, my findings are contradictory and makes it hard to answer my research question. Previous research pointed towards support for my hypothesis that transformational leadership is important during digitalisation, especially for anxious employees. This was supported by some of my results. However, when I

performed my robustness testing, this conclusion is not supported. This means that more research on this topic is required to contribute to the academic field.

In that future research, more attention needs to be given to measuring transformational leadership. The survey of Carless and Mann (2000) is quite useful for this topic, but more respondents are required. Also, it might be worthwhile to ask several employees with the same CFO to fill in a questionnaire. This creates a more reliable description of the transformational behaviour of these CFOs.

After more research have been done, my research (with the follow up research) can be used to create a framework about the outcome expectations of digitalisation and transformational leadership. This framework should show how the CFO scores on their transformational leadership skills (using the paper of Carless and Mann, 2000) and how that reflects on the expected outcome results of their anxious and non-anxious employees.

Once the framework is ready, based on my research and the follow up research, CFOs can use this in times of digitalisation. The framework will tell them how their leadership will

influence the outcomes of the digitalisation. Also, without such framework, my research emphasizes the importance of transformational leadership during digitalisation. When companies are looking how to improve their digitalisation, they should look for transformational leaders.

28/34

5.4 Conclusion

My research shows that transformational leadership might influences employees’ expectations of digitalisation. Evidence suggest also this effect is stronger for anxious employees, while transformational leadership has no effect on employees with affinity. However, my research also shows evidence that contradicts these relations.

6. REFERENCES

Agarwal, R., Johnson, S. L. & Lucas, H. C. (2011). Leadership in the Face of Technological Discontinuities: The Transformation of EarthColor. Communications of the Association for

Information Systems, Vol. 29, pp. 627-644. DOI: 10.17705/1CAIS.02933

Barker, R.A. (2001). The nature of leadership. Human Relations, Vol. 54, No. 4, pp. 469-494. Belias, D. & Koustelios, A. (2014). The impact of leadership and change management

strategy on organization culture. European Scientific Journal: March 2014, Vol. 10, No. 7, pp. 451-470.

Cao, M., Chychyla, R. and Stewart, T. (2015). Big Data Analytics in Financial Statement Audits. Accounting Horizons: June 2015, Vol. 29, No. 2, pp. 423-429.

Carless, S.A., Wearing, A.J., & Mann, L. (2000). A Short Measure of Transformational Leadership. Journal of Business and Psychology, Vol. 14, No. 3, pp. 389-405.

Cawsey, T. F., Deszca, G., & Ingols, C. (2016). Organizational change: an action oriented toolkit (3e ed.). Thousand Oaks, United States of America: SAGE Publications, Inc. Compeau, D.R. and Higgins, C.A. (1995). Computer Self-Efficacy: Development of a Measure and Initial Test. MIS Quarterly, Vol. 19, No. 2, pp. 189-211.

Compeau, D.R., Higgins, C.A. and Huff, S. (1999). Social Cognitive Theory and Individual Reactions to Computing Technology: A Longitudinal Study. MIS Quarterly, Vol. 23, No. 2, pp. 145-158.

Edison, S. & Geissler, G. (2003). Measuring attitudes towards general technology: Antecedents, hypotheses and scale development. Journal of Targeting, Measurement and

Analysis for Marketing, Vol 12, pp. 137–156. https://doi-org.proxy-ub.rug.nl/10.1057/palgrave.jt.5740104

29/34

Faupel, S. & Süß, S. (2019). The Effect of Transformational Leadership on Employees During Organizational Change–An Empirical Analysis. Journal of Change Management, Vol 19, No. 3, pp. 145-166. https://doi.org/10.1080/14697017.2018.1447006.

Gartner glossary. (2020). https://www.gartner.com/en/information-technology/glossary/digitalization.

Geissler, G.L. & Edison, S.W. (2005). Market Mavens' Attitudes Towards General Technology: Implications for Marketing Communications. Journal of Marketing

Communications, Vol. 11, No. 2, pp. 73-94. https://doi-org.proxy-ub.rug.nl/10.1080/1352726042000286499.

Gelbrich, K. & Sattler, B. (2014). Anxiety, crowding, and time pressure in public self-service technology acceptance. Journal of Services Mark, Vol. 28, No. 1, pp. 82-94.

Gimpel, H., Graf, V. & Graf-Drasch, V. (2020). A model for individuals’ acceptance of smart energy technology – A meta-analysis. Energy Policy, Vol. 138.

https://doi.org/10.1016/j.enpol.2019.111196.

Jung, D., Wu, A. and Chow, C.W. (2008). Towards understanding the direct and indirect effects of CEOs' transformational leadership on firm innovation. The Leadership Quarterly, Vol. 19, No. 5, pp. 582-594.

Lawson, R. (2019) Management Accounting Competencies: Fit for Purpose in a Digital Age? Institute of Management Accountants. Montville, NJ.

Leviäkangas, P. (2016) Digitalisation of Finland’s transport sector. Technology in Society:

November 2016, Vol. 47, pp. 1-15. https://doi-org.proxy-ub.rug.nl/10.1016/j.techsoc.2016.07.001.

Lo, M.-C., Ramayah, T., Min, H.W. and Songan, P. (2010). The relationship between leadership styles and organizational commitment in Malaysia: role of leader–member exchange. Asia Pacific Business Review, Vol. 16, No. 1-2, pp. 79-103. https://doi-org.proxy-ub.rug.nl/10.1080/13602380903355676.

Meuter, M. L., Ostrom, A. L., Bitner, M. J., & Roundtree, R. (2003). The influence of technology anxiety on consumer use and experiences with self-service technologies. Journal

of Business Research, Vol 56, No. 11, pp. 899–906. https://doi.org/10.1016/S0148-2963(01)00276-4.

30/34

Möller, K., Schäffer, U. & Verbeeten, F. (2020). Digitalization in management accounting and control: an editorial. Journal of Management Control¸ Vol 31, pp. 1–8.

https://doi.org/10.1007/s00187-020-00300-5.

Muktiarni, M., Widiaty, I., Abdullah, A. G., Ana, A. & Yulia C., (2019) Digitalisation trend in education during industry 4.0. Journal of Physics: Conference Series, Vol. 1402, No. 7. https://doi.org/10.1088/1742-6596/1402/7/077070.

Quattrone, P. (2016) Management accounting goes digital: Will the move make it wiser?

Management Accounting Research, Vol 31, pp 118–22.

Ronning, P. L. (2004). Change and the Learning Organization. Journal of Cardiovascular

Management, Vol. 15, No. 1, pp. 7–8.

Venkatesh, V., Morris, M., Davis, G., & Davis, F. (2003). User Acceptance of Information Technology: Toward a Unified View. MIS Quarterly, Vol 27, No. 3, pp. 425-478.

doi:10.2307/30036540

Vial, G. (2019). Understanding digital transformation: A review and a research agenda.

Journal of Strategic Information Systems, Vol 28, pp. 118-144.

https://doi.org/10.1016/j.jsis.2019.01.003.

Warren, J. D., Moffitt, K. C. & Byrnes, P. (2015). How big data will change accounting.

Accounting Horizons, Vol. 29, No 2, pp. 397–407.

Westerman, G., Bonnet, D., McAfee, A. (2014). The Nine Elements of Digital Transformation. MIT Sloan Management Review, pp. 1-8 Available on

http://sloanreview.mit.edu/article/the-nine-elements-of-digital-transformation

Wilson, J. P. & Campbell, L. (2016) Financial functional analysis: a conceptual framework for understanding the changing financial system. Journal of Economic Methodology, Vol. 23, No. 4, pp. 413–431. http://dx.doi.org/10.1080/1350178X.2016.1157200.

Yoo, Y., Boland, R.J., Lyytinen, K. & Majchrzak, A. (2012) Organizing for innovation in the digitized world. Organization Science, Vol. 23, No 5, pp. 1398–1408.