The role of supplier-buyer relationships on the organisation’s ability to exploit emerging technologies : a mixed methods study in the context of artificial intelligence based robotic process automation

(1)

Master’s Thesis Digital Business

The role of supplier-buyer relationships on the Organisation’s

Ability to Exploit Emerging Technologies. A mixed methods study

in the context of Artificial Intelligence based Robotic Process

Automation.

Name: Michiel van Ommen Student number: 10197559

Study Program: Msc Business Administration University: University of Amsterdam

Supervisor: Andreas Alexiou Date: 22 June 2018

(2)

2

Statement of originality

This document is written by Michiel van Ommen who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

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

(3)

3

Abstract

This research is aimed at finding out what determinants in the buyer-supplier relationship influences the organisation’s ability to exploit new technology gained from the supplier, in the context of Artificial Intelligence based Robotic Process Automation. The research question is: How does trust in supplier affects the exploitation of new technology within an organisation and to what extend is this moderated by knowledge asymmetry and informal relationships, in the context of the implementation of Artificial Intelligence based Robotic Process

Automation? This research makes use of an explanatory mixed-method research design. The first phase consists of a quantitative analysis of a survey that was held at 150 companies that recently implemented Artificial Intelligence based Robotic Process Automation, in which C-level managers were asked to fill out this survey. The second phase, qualitative, consists of multiple case studies and is cross-sectional, with nine different (medium to large) companies, and is used for an in-depth analysis of the outcomes from the first phase. The data is collected with the use of semi-structured interviews. From the results it follows that, among other findings, trusting the supplier positively influences the organisation’s ability to exploit new technology. Moreover, informal relationships and knowledge asymmetry are found to moderate the exploitation of technology. Furthermore, from the post-hoc analysis it follows that management should pay more attention on making their employees ready for change (e.g. implementation of RPA) with the use of change management. Also, trail-and-error practices and inter-organisation knowledge sharing are found to positively influence the organisation’s ability to exploit technology.

Keywords: Artificial Intelligence, Automation, Emerging Technology, Exploitation of technology,

(4)

4

1. – Introduction

Automation has been used for over hundred years to reduce labour costs and speed time to market. In 1913, Ford Motor Company introduced the assembly lines for their car production. Benefits of automation include the reduction of direct labour costs, set-up time and time to market, and improving the quality of the produced product or service (Co, Patuwo & Hu, 1998). Nowadays, a new and more advanced way of automation is being used, namely Robotic Process Automation (hereafter RPA). RPA performs tasks in the background and many decisions are made on the foundation of the software output of these ‘‘robots’’. With the execution of business processes employees spend a substantial time dealing with Customer Relationship Management, Enterprise Resource Planning, spread sheets, and legacy systems in manual repetitive tasks that are highly time consuming (Lacity & Willcocks, 2015). Nowadays, these manual tasks that are highly structured and routine based can be handled by a robot, in which the employees that execute these tasks has more time for value added tasks or can be replaced with the use of RPA. However, with any type of automation there are certain degrees of risk. For example, if within the RPA solution an error within its algorithms is present while working continuously (24/7), and without a human checking the work that is done, dramatic results could occur (KPMG, 2017). With this risk, good training and practice from the supplier are needed for a perfect execution of the RPA.

The concept of Artificial Intelligence (hereafter AI) has been present for more than 60 years, however only in the last decade it started to gain a lot of attention. AI is the idea to empower machines and computer systems to exhibit human-like intelligence, judgements and thoughts (Minsky, 1961; Kalogirou, 2003; Stone & Veloso, 2000). Burgess (2017) indicates that RPA and AI are very different types of technology, however they are able to complement each other. The combination of the two technologies can become extremely powerful. When the input data within RPA is unstructured (e.g. customer mail) or semi-structured (e.g.

(7)

7 invoices) AI can be introduced to turn this data into a structured format, which makes it possible to automate this unstructured and semi-structured data.

For this research the term ‘’AI-based RPA’’ is used, which indicates the combination between Artificial Intelligence and Robotic Process Automation. With the use of AI-based RPA, organisations have the opportunity to reduce costs, speed time to market, and eventually gain a competitive advantage. To gain a competitive advantage, organisations have to improve their speed of learning. According to Nooshinfard & Nemati-Anaraki (2014) organisations need to understand how this knowledge is created, shared and used, in order to capitalize this knowledge. The organisation’s ability to learn and understand a new technology and to diffuse it across other aspects within the organisation is the Exploitation of Technology. The more education and training an employee receives, the higher his or her ability to assimilate and exploit this technology (Cohen & Levinthal, 1990). Research has found that the ability to exploit new technology within an organisation creates a competitive advantage (Cohen & Levinthal, 1990). To develop the exploitation of technology, the organisation needs to collaborate with suppliers in order to promote continuous learning and accumulation of knowledge (Ritala & Hurmelinna-Laukkanen, 2013).

Knowledge about emerging technologies is mostly gained via suppliers (Argote & Ingram, 2000; Kamaşak & Bulutlar, 2010). This entails the importance to gain this knowledge from the supplier in order to exploit the solution within the organisation, in which the organisation is able to reduce costs and gain a competitive advantage. Moreover, it highlights the importance of the factors surrounding the supplier-buyer relationship on the ability to exploit the new technology. The transferred knowledge can be affected by the inter-organisational trust in the supplier. When the supplier is less open and honest the buyer will gain less knowledge, since it affects the perceived ability, benevolence, integrity, and predictability of the buyer and supplier in which they meet the business norms (Panteli &

(8)

8 Sockalingam, 2005). However, when the organisation has little prior knowledge about the new technology (e.g. knowledge asymmetry exists) it is more difficult to determine the trustworthiness of the supplier (Schoorman, Mayer, and Davis, 2007), which can affect the transferred knowledge. Furthermore, an informal relationship between the buyer and supplier has the possibility to increase the transferred knowledge from supplier to buyer (Dillon & McQueen, 2016). So far literature on the topic of informal relationships and trust on the ability to exploit technology primarily focuses on intra-relationships (Andersén, & Kask, 2012; Camisón & Forés, 2011; Costa & Monteiro, 2016). This research aims to provide deeper insights in the supplier-buyer relationship on the Exploitation of Technology within the organisation. For this reason the research question is:

How does trust in supplier affects the exploitation of new technology within an organisation and to what extend is this moderated by knowledge asymmetry and informal relationships, in the context of the implementation of Artificial Intelligence based Robotic Process Automation?

To answer this question this research will make use of an explanatory mixed-method research design. The first part is quantitative with the use of a dataset that is conducted at 150 companies that recently adopted AI-based RPA. From this dataset evidence is found for the hypotheses. Second, the post-hoc analysis, with the use of multiple case studies the factors and hypotheses of interest will be illuminated at eight organisations that recently adopted AI-based RPA and one supplier.

The next part of this research will provide a review of literature and formulation of the hypotheses (Chapter 2). Then the research method and data will be discussed (Chapter 3) followed by an overview of the results (Chapter 4). Thereafter, the results will be discussed (Chapter 5), and this is followed by the conclusion (Chapter 6).

(9)

9

2. – Literature Review

When an organisation hires a supplier to transfer knowledge in order to exploit the AI-based RPA, research identified several factors influencing this process (Van Wijk, Jansen & Lyles, 2008). The social factors within this research consist of the trustworthiness of the supplier (Vignola et al., 2016) and the communication environment between the two that can be formal or informal (Nooshinfard & Nemati-Anaraki, 2014). Also, identifying the trustworthiness of the supplier is affected by the amount of prior knowledge of the organisation (Hardin, 2002). This chapter will go into more detail about RPA, AI and the factors that affect the ability to exploit.

2.1 – Robotic Process Automation

According to Du Plessis and Mwalemba (2016) emerging technologies are those technologies whose knowledge base is expanding, as well as their application to existing markets is still undergoing innovation that has the potential of creating new markets. An example of an emerging technology is RPA.

‘‘Robotic Process Automation (RPA) takes the robot out of the human. The average knowledge worker employed on a back-office process has a lot of repetitive, routine tasks that are dreary and uninteresting. RPA is a type of software that mimics the activity of a human being in carrying out a task within a process. It can do repetitive stuff more quickly, accurately, and tirelessly than humans, freeing them to do other tasks requiring human strengths such as emotional intelligence, reasoning, judgement, and interaction with the customer.’’ - Leslie Willcocks (2015).

RPA is being used to tackle typical challenges organizations face. Namely, cost reduction, increase of quality, and faster processes. RPA is considered to have the highest potential for automation in finance, supply chain, and in human resource departments (Anagnoste, 2017). Furthermore, The Institute for Robotic Process Automation (2015)

(10)

10 indicates that there will be a significant reduction of full-time employees (FTEs). It can be assumed that one robot will replace at least two humans and this can go up to five humans. However, these jobs are more affected in countries such as India and China, which will provide an opportunity for business process outsourcing providers. They will be able to reach new and different customers as well as helping their client to find a balance between FTEs and robots. Instead of replacing humans, RPA will elevate them, due to the fact that RPA provides a faster approach and will enhance speed and cost effectiveness.

2.2 – Artificial Intelligence

AI is a term that, in its broadest sense, would indicate the ability of a machine to perform the same kind of functions and behaviour that characterizes human thought. Furthermore, AI is part of computer science that is concerned with designing intelligent computer systems, which can learn, reason, solve problems, and understand behaviour and language. With AI the physical part humans perform, can be substituted by robots (Kalogirou, 2003).

2.3 – Differences between RPA and AI

Below the differences between RPA and AI are summarized (Table 1).

RPA AI

Mimics users’ activity Mimics human thought

Can process structured and semi-structured data Can process structured, semi-structured, and unstructured data

Automation is rule-based Can learn or change behaviour

Table 1: Differences between RPA and AI (Van Ommen, 2018)

2.4 – AI-based RPA

When RPA is combined with the adaptability and awareness of artificial intelligence, it is called: ‘’Automation of automation’’. This technology is able to learn and respond to

(11)

11 problems that normally were not possible. These software robots are able to perform the tasks of knowledge workers whose jobs have not been affected by the impact of automation, which will elevate the employees (The Institute for Robotic Process Automation, 2015).

2.5 – Exploitation of Technology

2.5.1 – Definition exploitation of technology

In this research, the ability to exploit new acquired technology from the supplier is based on the concept of Absorptive Capacity (hereafter ACAP) by Cohen and Levinthal (1990). ACAP is the ability of an organisation to recognize the new value, assimilate this, and apply it to commercial ends. The ability to exploit external knowledge (e.g. RPA technology) is a critical component of innovative capabilities. The ability to evaluate and utilize outside knowledge is largely based on prior knowledge, which can include basic skills or even shared language. Thus, this prior knowledge confers the ability to recognize the value of new information. Furthermore, ACAP is affected by the cumulative nature of knowledge. The more education and training an employee receive, the higher his or her ability to assimilate and exploit this new acquired technology (Cohen & Levinthal, 1990).

Also, Volberda, Foss, and Lyles (2010) explain ACAP as the ability to acquire knowledge and the motivation to learn from the individual to facilitate the identification, selection, and implementation of new and innovative practices. In addition, Lane and Lubatkin (1998) proposed that ACAP is a dyad-construct (e.g. relative absorptive capacity) and not a firm-level construct. This entails that information is recognized and assimilated on individual level and applied within the organisation. Moreover, Zahra and George (2002) call the ability to transform and exploit knowledge the realized absorptive capacity, which consists of the constructs of assimilation and application. Assimilation entails the understanding of the automation solution provided by the supplier. Application has its focus

(12)

12 on the usage and further implementation of the automation solution within the organisation (Cohen & Levinthal, 1990).

In this research organisations do have a prior knowledge base about RPA, which entails that they recognize the benefits, importance and potential RPA has to offer, and in turn enable a supplier to acquire external knowledge to understand and learn how to implement and adjust the new RPA solution. For this reason the focus lies on the organisation’s ability to exploit new acquired technology within their own organisation, which entails the assimilation and application of external knowledge and is called: The Exploitation of Technology.

2.6 – Trust

2.6.1 – Definition trust

Trust is defined as the perceived credibility and benevolence of a target of trust. The perceived credibility of an exchange partner focuses on the expectancy that the partner’s word or written statement is reliable. The benevolence of the exchange partner focuses on the extent to which the partner is genuinely interested in the other partner’s welfare and is

motivated to seek joint gain (Doney & Cannon, 1997). Furthermore, the concept of trust is not the behaviour or a choice of the buyer, but trust is an underlying psychological condition that can result from these actions (Qureshi & Evans, 2013). Moreover, Kee and Knox (1970) define trust as the willingness of a party (e.g. organisation) to be vulnerable to the actions of the other party (e.g. supplier) and are based on the expectation that the supplier will perform a particular action important to the organisation, irrespective of the ability to monitor or control this supplier.

2.6.2 – Trust on Exploitation of Technology (C-path)

When both parties trust each other, this stimulates favourable attitudes and actions and will increase openness and tolerance (Zhang, Zhao & Lyles, 2018). When the organisation

(13)

13 trusts the supplier, it will be more willing to work together and share their personal know-how (Mayer, Davis & Schoorman, 1995). Nahapiet and Ghoshal (2000) found that trust provides a social system and can motivate employees to engage in the exploitation of technology and will be more willing to take risks that are associated with exploring novel and creative ideas. Also, the organisation who trusts its supplier will tend to perceive the new technology as more credible and is more likely to exploit this technology, compared to technology obtained from less trustworthy suppliers (Montazemi, Pittaway, Saremi & Wei, 2012). Thus, trust increases the reliability in the supplier and improves the organisation’s ability to exploit the technology within the organisation. Within this research it is expected that trust in the supplier has a positive influence on the exploitation of technology within an organisation. From this the following hypothesis follows:

H1 = Trusting the supplier has a positive impact on the ability to exploit the new technology within the organisation.

2.6.3 – Trust on Knowledge Transfer

Inter-organizational trust is an important factor for business partnerships, because it nurtures the intentions of knowledge acquisition and sharing (Cai, Goh de Souza & Li, 2013; Chen, Lin & Yen, 2014). Based on trust, business partners determine the extent and nature of knowledge transfer. Furthermore, trust is an important predictor of knowledge transfer, since it affects the perceived ability, benevolence, integrity, and predictability of the buyer and supplier in which they meet the business norms (Panteli & Sockalingam, 2005). Also, trust reflects the scope and depth of the relationship between buyer and supplier, which can lead to stronger partnerships and better knowledge sharing. Additionally, Qureshi and Evans (2015) indicate that trust is an antecedent to transfer knowledge and a social capital resource that is embedded in relationships among people. A lack of trust between the buyer and supplier

(14)

14 decreases the transfer of knowledge. A reason for this decrease is that people who have limited openness are less honest. From these findings the hypothesis follows:

H2= Trusting the supplier positively influences the transferred knowledge from supplier to buyer.

2.7 – Knowledge Transfer 2.7.1 – Definition knowledge

Nooshinfard and Nemati-Anaraki (2014) explain the concept of knowledge as a critical resource that should be well managed for effective performance, which applies for both profit- and non-profit-, and either product or service organisations. Also, knowledge is a justified belief that increases the organisation’s capacity for effective action. Whereas Xiong and Deng (2008) explain knowledge as a combination of experts’ experiences, values and insights that may help evaluate and incorporate new information and experience. From this it follows that knowledge is the skill, intuition, and experience that influences decision-making. According to Polanyi (1966) and Saint-Onge (1996) there are two types of knowledge:

(1) Explicit knowledge is defined as ‘’hard’’ knowledge that can be expressed in numbers and words, and is shared formally and systematically in the form of data, specifications and manuals. This type of knowledge is part of everyday professional life, which can easily be captured and shared with others through courses or books for self-learning.

(2) Tacit knowledge is defined as ‘’soft’’ knowledge, which includes insights, intuitions, and hunches. This type of knowledge is difficult to express and formalize, which makes it difficult to share. Tacit knowledge includes skills and know-how that are part of a person and of the practices of the employee within the organisation, which are acquired over several years. This complicates the sharing of this type of knowledge.

Within this research the knowledge transfer of tacit knowledge is measured, because the knowledge is acquired and learned from the external automation supplier.

(15)

15 2.7.2 – Definition Knowledge Transfer

Performance in various parts of the organisation is enhanced when people share or gain knowledge. Knowledge sharing implies that individuals or organisations mutually adjust their beliefs and actions through interactions (Liao, Fei, & Chen, 2007). To gain new knowledge, interaction and communication about its tacit and explicit knowledge is required (Kamaşak & Bulutlar, 2010).

Knowledge Transfer is a process through which a unit is (e.g. organisation) is affected by the experience of the other (e.g. supplier), which may result in the joint creation by those units by sharing tacit knowledge (Argote & Ingram, 2000). Moreover, Ali, Musawir, and Ali (2018) explain knowledge transfer as the process of exchanging tacit knowledge through social and collaborative processes. This implies the transfer of knowledge that is framed within a specific context and is subject to the interpretation of the buyer (e.g. organisation). The process of knowledge transfer can take place through direct interpersonal means (e.g. face-to-face meeting, telephone conversations, e-mail, or online communication). Knowledge transfer has been found to be a positive indicator of the exploitation of technology within an organisation (Sharratt & Usoro, 2003).

According to several studies (Argote & Ingram, 2000; Van den Hooff & De Ridder, 2004) consists knowledge sharing of two processes: knowledge donation and knowledge collection. Knowledge donation is the communication of knowledge that is based on the desire of an individual to transfer their knowledge. Knowledge collection is the attempt of an organisation to pursue others to share their knowledge. Knowledge collection is the process of interest within this research. Furthermore, the transfer of knowledge between organisations is called inter-organisational knowledge transfer, which increases the learning between individuals from different companies and entails the conversion of individual learning into

(16)

16 organisational learning through internal mechanisms. The transferred knowledge from supplier to buyer is the point of interest within this research.

2.7.3 – Knowledge Transfer on Exploitation of Technology

The knowledge transfer process is a driving force for developing the organisation’s ability to exploit new technology within the organisation. This transfer process entails the exchange of tacit knowledge possessed by the project team in the form of past experiences and lessons learned, which forms the basis for a mutual understanding of the projects goals and challenges. In this way the process builds the exploration of the organisation by enabling team members to identify and acquire relevant external knowledge (Ali, et al., 2018; Biedenbach & Müller, 2012; Rafique, Hameed, & Agha, 2018).

Moreover, knowledge transfer has a positive effect on the exploitation ability of the organisation by creating an enabling environment for knowledge transfer. This environment facilitates the assimilation and transformation of relevant knowledge to address the needs of the project. When there is a difference in the knowledge transfer culture between the organisation and supplier, the employees of the organisation are more likely to be affected (Reich, Gemino & Sauer, 2012). Through interactions between the two, employees can acquire new knowledge that increases their learning. Knowledge transfer can develop and increase an organisation’s ability to exploit the new technology, which is achieved by integrating external knowledge about the RPA solution and transforming this into the organisation’s competence (Liao et al., 2007). Therefore, the following hypothesis is given:

H3 = Knowledge Transfer from supplier to buyer positively influences the organisation’s ability to exploit new technology within the organisation.

(17)

17 2.8 – Knowledge asymmetry

2.8.1 – Definition knowledge asymmetry

A definition of knowledge and asymmetry can be found in the Oxford dictionary, and is as follows:

Knowledge: ‘’Facts, information, and skills acquired through experience or education; the

theoretical or practical understanding of a subject.’’

Asymmetry: ‘’Lack of equality or equivalence between parts or aspects of something; lack of

symmetry.’’(Oxford dictionary, 2018).

Lin, Geng and Whinston (2005) indicate two types of knowledge asymmetry within their sender-receiver framework of knowledge transfer, which is based on information structures that typically exist within the transfer of knowledge. First, the sender-advantage asymmetric information structure, where the sender (e.g. supplier) of information has complete information whilst the receiver (e.g. buyer) of information has incomplete information and is unable to evaluate the received information and knowledge. Second, the receiver-advantage asymmetric information, where the receiver has more knowledge compared to the sender. These two structures are in line with the concept of knowledge asymmetry presented in the paper by Cimon (2004).

Many studies have been conducted on information asymmetry (Muthusamy & White, 2005; Roberts, 2000), however knowledge asymmetry is a concept that is not studied intensively within business research. For this research an overview is made (shown in Table 2) to indicate the differences between knowledge asymmetry and information asymmetry.

Grant (1996) explains that skills are learned when a deeper understanding of the solution is gained, which entails teaching and studying in order to get the know-how of the solution. From this it follows that gaining knowledge is more time consuming and has higher costs compared to gaining information, due to the fact that knowledge requires skills,

(18)

18 experience and attitude. Furthermore, knowledge is bound to an individual, because it is a capability of a person that allows that person to execute a certain action. Last, knowledge is not transferable in contrast to information. Knowledge is bound to an individual, which limits the possibility to transfer this knowledge. The supplier provides information to understand and learn the technology, consequently the organisation gains knowledge from skills and experience by doing, learned form the supplier. When these skills and experiences are low this information is not absorbed and/or transformed into new knowledge.

Information Knowledge

Boundary Unbound Bound to individual

Cost Lower Higher

Learning Not needed Needed

Transferable Transferable Not transferable

Table 2: Differences between information and knowledge (Van Ommen, 2018)

2.8.2 – Moderating effect of knowledge asymmetry

The organisation assesses the trustworthiness of a supplier to determine whether the organisation can trust this supplier. Knowledge is an important factor to determine the trustworthiness of a supplier. To determine the trustworthiness of a supplier, the organisation needs to process information about the technology (Schoorman et al., 2007). Additionally, Hardin (2002) indicates that knowledge is developed with everyday experiences of the technology and more knowledge will allow the organisation to assess the trustworthiness of the supplier more accurate. Consequently, it can be seen that when knowledge asymmetry is not present (e.g. the organisation possesses knowledge similar to the supplier) the organisation is more able to indicate whether the supplier is trustworthy or not, which indicates the moderating effect on the relationship between inter-organisational trust and knowledge transfer. From these findings it follows that the hypothesis will be:

(19)

19

H4 = Knowledge Asymmetry between the buyer and supplier negatively moderates the relation between Trust and Knowledge Transfer.

2.9 – Informal relationship between buyer and supplier 2.9.1 – Definition Informal Relationship

Zimmermann, Oshri, Lioliou and Gerbasi (2017) call the informal relationship between two parties the captive/external sourcing relationship. This social capital aspect has been found to influence the amount of shared knowledge. It is found that in practice there seems to exist a trend towards less confrontation and more cooperation among organisations (Zimmermann, et al., 2017). Moreover, innovation relies strongly on interaction and the ability to interact. During the process of knowledge transfer, obstacles within the organisation can form a barrier for this transfer. The ties of organisations are key for improving knowledge transfer. Firms with multiple sources of information are less likely to miss out on information. These multiple channels provide the opportunity to discover new information that can be combined in novel ways to generate innovation (Zhang & Xu, 2008). Moreover, Cheng and Fu (2013) explain inter-organisational relationships as two different types of relationships, which are based on the type of collaboration that can be either a close or a simple relationship. When organisations develop inter-organisational relationships, they are confronted with risk. Both parties strive to gain value and control risk, which requires trust, commitment and flexibility within this informal relationship.

2.9.2 – Moderating effect of Informal Relationships

Chen et al. (2014) indicate that knowledge transfer is an interactive process in which the organisation accumulates and develops new knowledge. This entails that managers are able to detect and understand associated business problems in which they are able to develop appropriate solutions, in which collaboration is required to speed up the transfer of knowledge. Moreover, a long-term collaboration relationship is an important condition for

(20)

20 effective knowledge transfer between supplier and buyer. Collaboration forces the involved parties to develop common goals and formulate joint strategies, which eliminates concerns (Möller & Svahn, 2004). Also, organisations need to collaborate with the supplier and share knowledge in order to gain more knowledge. This collaboration influences how much knowledge there will be transferred. A strong informal relationship between buyer and supplier improves the transferred knowledge (Saifi, Dillon & McQueen, 2016; Todorova & Durisin, 2007). Furthermore, Zhang and Xu (2008) find that an informal relationship between both parties has a positive moderating effect on the relationship between the transferred knowledge and the exploitation of technology within the organisation. From these findings the hypothesis will be as follows:

H5 = The relationship Knowledge Transfer on Exploitation of Technology is positively moderated by the Informal inter-organisational Relationship between buyer and supplier.

2.10 – Control variables

This research makes use of two control variables, namely firm size and industry of the organisation. First, the size of an organisation can affect the innovation performance. Larger organisations are more likely to have more resources, which enhance their innovative behaviour and performance (e.g. implementing and understanding the RPA solution). However, large organisations are less adaptable compared to small organisations, but small organisations are usually less powerful than large organisations (Tsai, 2001). Second, the industry of the organisation is categorized into five industries, namely financial services, manufacturing, retail, public sector, IT/computer services. It will be researched whether the industry of an organisation has an influence on the exploitation of technology within the organisation.

(21)

21 2.11 – Conceptual Model

This research investigates to what extent Trust in supplier has an effect on the Exploitation of Technology within the organisation, which is mediated by Knowledge Transfer from supplier to the organisation. Furthermore, from the literature it follows that Knowledge Asymmetry (e.g. the organisation has less knowledge compared to the supplier) negatively moderates the relation between Trust and Knowledge Transfer. Also, the Informal Relationship positively moderates the effect of Knowledge Transfer on the Exploitation of Technology within the organisation. The relationships are visualized in the conceptual model below (Figure 1).

(22)

22

3. – Data and Method

In this chapter the research methodology within this research is explained. First, the overall research design and procedure are described. Second, the quantitative part of this research is explained. Last, the qualitative part of this research is described.

3.1 – Overall research design and procedures

This research is a combination of a deductive and inductive approach and will make use of an explanatory mixed-method research design. First a deductive approach, which is about testing composed theory through research (Saunders et al., 2012), and will test the assumptions that are based on existing theory. However, the existing literature on the exploitation of technology is scarce, especially regarding the inter-organisational relationship between buyer and supplier that are examined in this research. For this reason an inductive approach is used, which provides an open mind-set that will result in interesting relationships. The literature review provides a foundation for the conceptual building and analysis of the results in this research and to provide a better understanding of the subject.

This research will be a sequential explanatory research method, where the first part will be quantitative, with the use of a dataset. The second part is qualitative, with the use of multiple case studies. Furthermore, this research will be cross-sectional where different organisations are researched who experienced a similar implementation process. The method of research indicates how the data is approached and gathered. Quantitative research is mostly applied to find causal relationships between different variables (Saunders et al., 2012) and within this research it provides a general picture of the research problem and is executed to identify the significance of the predictive variables that affect the exploitation of technology within an organisation. The qualitative method is used to determine the motivations, beliefs, and perceptions of phenomena. Within this research a combination of both methods is used, in

(23)

23 order to comprehend the logic and theory underlying the relationships in this research (Eisenhardt, 1989). The aim of this research is to identify the factors surrounding the relationship between buyer and supplier that affect the organisation’s ability to exploit the new technology gained from the supplier, in the context of Artificial Intelligence based Robotic Process Automation.

3.1.1 – Conceptual model sequential explanatory multiple case study

3.2 – Quantitative procedure 3.2.1 – Quantitative sample

The quantitative part consists of a secondary dataset, which originally was collected by Dr. A. Alexiou, prof I. Oshri and Dr. S. Khanagha. The scales used are all adapted versions of previously validated scales. This dataset consists of a survey that is conducted at 150 companies that recently adopted AI-based RPA. C-level managers within the organisation, that were part during the implementation of the RPA solution, filled out this survey. This dataset is based on a five-point Liker scale. This dataset will be used to find evidence for the hypotheses and possibly statistics that will be illuminated in the multiple case studies. These factors are related to the buyer-supplier relationship of new technology with a focus on the

Phase 2 Literature review Analysis of dataset Phase 1 Semi-structured

(24)

24 exploitation of technology. Respondents were selected on the basis of four requirements, namely (1) if the organisation has more than 10 employees, (2) the organisation did use a third party provider to automate a task that was previously performed by a human, (3) the automation solution include AI, and (4) the respondent was familiar with how the automated solution has been implemented.

3.2.2 – Measurement

This research has its focus on the relationship between the organization (e.g. buyer) and supplier. The objective of this research is to identify and examine determinants of the exploitation of technology within organisations that recently implemented AI-based RPA. The exploitation of technology is based on trust between buyer and supplier, knowledge transfer, knowledge asymmetry, and the informal relationship between the two. The variables that will be used follows from the literature review. The dependent variable (DV) consists of the exploitation of technology within the organisation, the independent variable (IV) consists of trust in the supplier, the mediating variable is the knowledge transfer from the supplier to the organisation, and the relations are moderated through knowledge asymmetry and informal relationship. To measure the internal consistency the researcher makes use of the Cronbach’s alpha, which indicates how closely related a set of construct are as a group and shows the reliability. Furthermore, the dataset is checked whether some constructs needed recoding because the intent was opposite. The recoded constructs are indicated in Appendix IV with brackets indicating (recoded).

3.2.3 – Cronbach’s Alpha

To measure the reliability of the variables within this research, the Cronbach’s Alpha is calculated with the use of IBM SPSS (version 25). The outcomes are indicated in Table 3 below. The threshold for acceptance is when the alpha is .700 or higher. If the Cronbach’s alpha is below .700 it indicates a low reliability. The Cronbach’s Alpha was checked for the variables and 3 variables (Exploitation of Technology, Trust, and Knowledge Transfer) have

(25)

25 an alpha > .700, which indicate a reliable variable. However, 2 variables (Knowledge Asymmetry and Informal Relationship) have an alpha < .700, which would imply that these items should not form a construct. Appendix III summarizes the details of every variable.

Variables Cronbach's Alpha

Exploitation of Technology .843

Trust .779

Knowledge Transfer .864

Knowledge Asymmetry .677 Informal Relationship .690

Table 3: Cronbach’s Aplha for each variable

3.2.4 – Dependent variable

Exploitation of Technology (7 items, α = .843). The measurement for the dependent variable

(DV) of the exploitation of technology is based on two constructs from Cohen and Levinthal (1990). These two constructs consist of assimilation, and application. The participants rated their ability to assimilate and apply the new technology within their organisation. The scale ranges from strongly disagree (1) to strongly agree (5). This item consists of whether the organisation understands and is able to adapt the RPA solution. A high score represents a high ability to exploit the new technology within the organisation, which entails that the organisation is able to further assimilate and exploit the automation solution within their organisation. The explanation of the items are indicated in Appendix IV

3.2.5 – Independent variables

Trust (8 items, α = .779). The independent variable trust is based on the definition by

Doney and Cannon (1997). They define trust as the perceived credibility and benevolence of a target of trust. The participants were asked to rate eight statements about trust (see Appendix IV) with the scale ranging from strongly disagree (1) to strongly agree (5). Two statements are recoded because they indicate the opposite outcome of trust. A high score represents that the organisation trusts its automation supplier.

(26)

26

Knowledge Transfer (7 items, α = .864). Knowledge Transfer is used as a mediating

variable for the relation between trust and the exploitation of technology of the organisation. The measure knowledge transfer is based on the data collected by by Dr. A. Alexiou, prof I. Oshri and Dr. S. Khanagha. Knowledge transfer has been found to positively influence the organisation’s ability to exploit technology (Ali, et al., 2018; Biedenbach and Müller, 2012; Rafique, Hameed, & Agha, 2018; Reich, Gemino and Sauer, 2012). The participants were asked to rate seven statements (see Appendix IV) with the scale ranging from strongly disagree (1) to strongly agree (5). A high score represents that the supplier was able to transfer their knowledge to their buyer (e.g. the organisation).

3.2.6 – Moderating variables

Knowledge Asymmetry (KA) (3 items, α = .677). Knowledge asymmetry is expected to

moderate the relationship between trust and knowledge transfer. When knowledge asymmetry does not exists - the organisation is more able to assess the trustworthiness of the supplier - it positively moderates this relationship (Hardin, 2002; Schoorman, Mayer & Davis, 2007). The measure knowledge asymmetry is based on the data collected by by Dr. A. Alexiou, prof I. Oshri and Dr. S. Khanagha. The participants were asked to rate the statements (see Appendix IV) with the scale ranging from strongly disagree (1) to strongly agree (5). One statement is recoded, because it indicates the opposite of knowledge asymmetry. A high score indicates that knowledge asymmetry exists.

Informal Relationship (IR) (4 items, α = .690). Informal relationship is the moderating

variable on the relationship between knowledge transfer and the organisation’s ability to exploit technology. The measurement of Informal Relationship is based on the paper by Zimmermann et al. (2017), which they call the captive/external sourcing relationship. The participants were asked to rate four statements (see Appendix IV) with the scale ranging from

(27)

27 strongly disagree (1) to strongly agree (5). A high score indicates that an informal relationship is present.

3.2.7 – Control variables

For this research the following control variables are selected: firm size and the industry of the organisation. These control variables are included, because they are expected to affect the dependent, independent, mediating, and moderating variables.

The first control variable is firm size and is measured by the number of employees from the organisation. Within the dataset three dummy variables are created to indicate whether the organisation is a small, medium or large organisation and is used to identify the differences between the three types of organisations. The criteria for small (11-50 employees), medium (51-249 employees) and large (>250 employees) organisations are in line with the European Commission (2014). The second control variable is industry, which consists of five options. Within the dataset five dummy variables are created to identify in which industry the organisation is active. The different items for the industry and size are indicated in Appendix IV.

3.2.8 – Quantitative data analysis

Within IBM SPSS (version 25) the regression models by Hayes are used to find effects for the moderating and mediating variables. This section will be twofold, where the first part will investigate the basic model (e.g. trust influences the exploitation of technology, mediated by knowledge transfer) and the second part will investigate the moderating variables (e.g. knowledge asymmetry and informal relationship).

3.2.8.1 – Data file

In order to check the data on any missing values a response analysis has been executed. Furthermore, the dataset is checked on scores that were out of range. The dataset that is used has no missing values or outliers, which means that the total amount within the dataset contains of 153 observations.

(28)

28

3.2.8.2 – Unengaged responses

Unengaged responses or straight-liners are respondents who give the same answer to all questions, which indicates that the questions are not answered seriously. In order to check for unengaged responses the standard deviation over all the main variables are executed by the SPSS SD function. This standard deviation is referred to the within-subjects deviation. Thereafter, the standard deviations are sorted. When the standard deviation is zero (0) it indicates that there is an unengaged response detected and has to be excluded from the dataset. Within this dataset no standard deviation of zero is detected, which indicates that there are no unengaged responses.

3.2.8.3 – Distribution dependent variable

To check whether the distribution of the dependent variable is normal, Graph 1 indicates that the dependent variable is almost normal distributed. It is slightly (positive) skewed to the right.

Graph 1: Distribution dependent variable 3.2.8.4 – PROCESS macro Hayes

In order to analyse the theorized hypotheses in SPSS, this research makes use of the PROCESS macro (version 3.0) developed by Hayes (2018). With this software this research makes use of model 4, which results in a simple mediation, moderation and conditional

(29)

29 indirect effects. This model is used to analyse hypothesis 1, hypothesis 2 and hypothesis 3. Furthermore, model 21 will be used to measure the moderating effect of knowledge asymmetry on the relationship between trust and knowledge transfer (e.g. hypothesis 4) and the moderating effect of informal relationship on the relationship between knowledge transfer and exploitation of technology (hypothesis 5).

3.3 – Qualitative procedure 3.3.1 – Research strategy

This research makes use of a multiple case study strategy. Case studies enable the researcher to get a detailed understanding of the context within the research (Saunders et al., 2012). For this research nine organisations are analysed. The unit of analysis for this research are organisations that recently adopted AI-based RPA.

3.3.2 – Selection criteria

Interviewees are selected with the use of multiple-stage sampling techniques (Saunders et al., 2012), where the criteria entails that the case (e.g. the organisation) has more than ten employees, used a third party for the implementation of the RPA, was familiar with the implementation process, and used AI within their RPA solution.

3.3.3 – Data collection and sample

Together with two other students from the University of Amsterdam the data is collected. This will provide other and hopefully deeper insights, because all three students have a different angle of approach on the process of the implementation of RPA.

The data collection is realized through semi-structured interviews. These types of interviews are proven to be a valid method for multiple case studies (Eisenhardt, 1989). These

interviews are used to determine the interviewee’s thoughts, feelings, and intentions about the effect of the buyer-supplier relationship on the Exploitation of Technology within the organisation. Moreover, with a semi-structured interview it is possible to ask additional

(30)

30 questions during the interview when new or surprising themes are discussed (Saunders et al., 2012). The length of the interviews varied between forty-five to sixty minutes and all are conducted in Dutch, which is their native language. Using the interviewee’s native language is a powerful route to acceptance and can be seen as an indicator of one’s willingness to enter into the world of the interviewee (Rubin & Rubin, 1995). The interviewees were all present and involved during the implementation of the RPA solution.

Before reaching out we first held a pilot interview in order to check the interview questions. Also, personal contact (via telephone) with suppliers in the Netherlands are held to gain deeper insights about the current development of RPA. For this research nine organisations were analysed. The unit of analysis for this research are organisations that recently adopted AI-based RPA. From these nine cases, eight cases are organisations that are active in the Netherlands and implemented RPA via a third party. Within these eight cases there are differences concerning the type of automation, which varies between process automation to fraud detection. However, these differences do not influence the learning during the implementation, because the RPA solution is a software package that is implemented as a basic package and thereafter is extended. Also, from these eight cases there is one case that is a small organisation (Case H), the other seven cases consists of large organisation. Furthermore, one case (Case I) is a RPA consultant in the Netherlands, which can provide insights on the other side of the story. Table 4 indicates the summarized details of all cases.

(31)

31

Table 4: Summarized details cases A to I (Van Ommen, 2018)

3.3.4 – Interview questions

The general subject of the interview was the ability to exploit new technology within the organisation. The questions are divided into different categories. The first set of questions was aimed at the RPA solution and the respondent’s role during the implementation. The respondents were asked to specify one RPA solution and use this as a remainder during the rest of the interview. The second set of questions was focused at the relationship between the buyer and supplier and to what extend this could have an effect on the ability to exploit. The third set of questions was about knowledge asymmetries between buyer and supplier (e.g. in which the supplier has more knowledge about the RPA solution compared to the buyer). The fourth set of questions has its focus on trust in the supplier, which entails that the organisation trusts what the supplier says and promises. The fifth set of questions was about the ability to exploit within the organisation and how this could be improved. The last set of questions has its focus on the exchanged knowledge and was aimed at providing more insights in the implementation process. The complete interview protocol is attached in appendix V. The interviews were recorded, which creates a transparent research (Shenton, 2004).

Case Firm Size Industry Number of

interviewees Language Job Interviewee 1

Job Interviewee 2

A Large Public Sector 1 Dutch Implementation

Manager n/a

B Large Telecommunication 2 Dutch Manager RPA

implementer

C Large Financial Services 1 Dutch Manager RPA n/a

D Large Financial Services 2 Dutch Manager RPA IT manager

E Large Financial Services 1 Dutch Manager RPA n/a

F Large Beverage 1 Dutch IT business Partner n/a

G Large Financial Services 1 Dutch Senior RPA-expert n/a

H Small Retail/Distribution 1 Dutch CEO / Manager n/a

(32)

32 3.3.5 – Data Analysis

The next step in the process is transcribing the interviews. For the transcription of the interviews the transcribing tool Trint, an online transcribing website, was used. From this the audio files were converted into Word documents. The names of the cases are made anonymous to protect their identity and privacy. The transcripts have the same language as the language of the interview, namely Dutch. The quotations that are used are translated into English. After the transcripts, the next step in analysing the data is coding. The goal is to identify what the different categories from the data are. Thereafter, the codes that are in line with each other are grouped in corresponding thematic codes. The following themes, indicated in Table 5, are identified as relevant for analysing the organisation’s ability to exploit. On one hand these themes follow from the literature review and on the other hand from the interviews. For this part a computerised data system called QRS NVivo 12 was used for the analysis of the data. This software makes it possible to identify and analyse patterns between constructs within the data (Braun & Clarke, 2006). In order to create themes, first the researcher make use of open coding, which is a method of defining particular factors that are of importance within this research. Thereafter, axial coding is used, which creates a connection between categories and subcategories. An overview of the codes is provided in Appendix VI.

Table 5: General themes after coding transcripts (van Ommen, 2018) General Information about the organisation and the RPA solution Trust

Knowledge

Supplier-buyer relationship Exploitation of Technology Other factors influencing EoT

(33)

33

4. – Results

In this chapter the results of both the quantitative and qualitative analyses are discussed. The first paragraph focuses on the quantitative results. In the second paragraph the qualitative results are discussed.

4.1 – Results quantitative phase

In order to get a better perception of the data within this research, an insight has been gained in the number of respondents, means, standard deviations (SD), minimum and maximum scores and the correlation among all the predictors and control variables within this research. The descriptive statistics are shown in table 6, furthermore a Pearson Correlation analysis is shown in table 7. To test the hypotheses within this research, model 4 and model 21 of the PROCESS macro (version 3.0) by Hayes (2018) are used. Model 4 is used to perform a simple mediation analysis. Model 21 is used to test a moderation analysis and to perform an analysis on the moderated mediation.

4.1.1 – Descriptive statistics

The descriptive statistics in Table 6 indicate a skewed distribution to the right side. The respondents had a choice between one and five; therefore it is expected that most means should be around two and a half. In this dataset the means are all above 3, which indicates a slightly positive skewed distribution.

(34)

34 Below (see Table 7) the Pearson Correlation analysis is given. When looking at the main antecedents Trust and Knowledge Transfer, it can be noticed that between Trust and Exploitation of Technology a strong positive significant relationship occurs (r = .401, p < .01). Also, the relationship between Knowledge Transfer and Exploitation of Technology is strongly significant (r = .533, p < .01). When looking at the other variables in this research, Knowledge Asymmetry correlates significantly with Exploitation of Technology (r = .163, p < .05) and strong with Trust (r = .504, p< .01) and strong with Knowledge Transfer (r = .438, p < .01). Also Informal Relationship positively correlates strong with Exploitation of Technology (r = .436, p < .01), Trust (r = 497, p < .01), Knowledge Transfer (r = .488, p < .01), and Knowledge Asymmetry (r = .287, p < .01).

Table 7: Pearson correlation

4.1.2 – Simple mediation analysis

Simple mediation is used to estimate and test the hypotheses 1, 2 and 3 which indicate the paths of causal influence from Trust in supplier to the Exploitation of Technology of the organisation, one through the proposed mediator Knowledge Transfer and a second independent of Xà M à Y mechanism. To calculate the direct and indirect effects of this mediation, model 4 of the PROCESS macro by Hayes (2018) is used.

A multiple regression analysis was conducted to assess each component of the proposed mediation model. The results are presented in table 8, 9 and 10. The results consist

(35)

35 of the association between trust and the exploitation of technology (c-path), the effect of trust on knowledge transfer (a-path) and the association between trust and the exploitation of technology, through knowledge transfer (c’-path).

The basic model indicates that trust is a significant predictor and is positively associated with exploitation of technology (b = 0.386, t = 5.412, p = .000). This indicates that hypothesis 1 (c-path) is supported (see Table 8). It is also found that Trust is positively related to knowledge transfer (b = 0.599, t = 8.549, p = .000), which indicates that the hypothesis 2 (a-path) is supported (see Table 9). Lastly, the results indicate that the mediator, knowledge transfer, is positively related to the exploitation of technology, controlling for the independent variable (trust) (b = 0.431, t = 5.419, p = .000). Thus, hypothesis 3 (b-path) is supported.

In this research, the 95% confidence interval of the indirect effects is obtained with 5000 bootstraps samples (Hayes, 2018). Table 10 indicates the c’-path of the association between Trust and Exploitation of Technology is non-significant (b = 0.1277, t = 1.551, p = .123) when controlling for Knowledge Transfer, so not suggesting any mediation. Furthermore, the output provides the 95% bias corrected bootstrapped confidence interval in which is checked whether zero (0) lies within the interval range. It has been found for the direct effect of trust on exploitation of technology that LL (Lower Limit): -0,035 and UL (Upper Limit): 0,2903. From this it follows that the true direct effect is not significant. When checking for the total indirect effect of Trust on Exploitation of Technology it is found that the effect is .258 and LL: 0,1349 and UL: 0,4141. From this it follows that zero (0) does not occur between the LL and UL, in which we can conclude that the indirect effect (mediation via knowledge transfer) is significant different from zero.

In all regressions in this simple mediation analysis it has been found that the control variables Firms Size and Industry have no significant effect on the variables.

(36)

36 Consequently, Trust was a significant predictor for the dependent variable, Exploitation of Technology, and the mediator variable, Knowledge Transfer. However, Trust is no longer significant in the presence of the mediator variable, which confirms the mediation effect of Knowledge Transfer. The outcomes for the paths are indicated in figure below (Figure 2).

Figure 2: Conceptual Model with outcomes

(37)

37

Table 9: Mediation analysis, KT dependent

Table 10: Mediation Analysis, EoT dependent

4.1.3 – Moderation analysis

To check for moderation, model 21 from PROCESS macro by Hayes (2018) is used. From the first regression (Table 11), which predicts the mediator variable using the independent variable Trust and the moderating variable Knowledge Asymmetry. From this regression it follows that Trust is not a significant predictor of Knowledge Transfer (b = 0.5403, t = 1.2336, p = .2194). Furthermore, Knowledge Asymmetry has been found to not

(38)

38 significantly influence Knowledge Transfer (b = 0.2203, t = 0.4078, p = .5899). Also, the interaction effect between Trust and Knowledge Asymmetry has no significant effect on Knowledge Transfer (b = -0.01, t = -0.0892, p = .9290). From this it follows that hypothesis 4 is not supported.

The next regression output (Table 12) predicts the dependent variable using the independent variables Trust and Knowledge Transfer, as well as the moderating variables Knowledge Asymmetry and Informal Relationship. From the regression it follows that Trust, controlling for Knowledge Transfer, is not a significant predictor of Exploitation of Technology (b = 0.0564, t = 0.6638, p = .5079). Next, Knowledge Transfer, controlling for Trust, is not a significant predictor of the Exploitation of Technology (b = 0.1185 t = 0.4478, p = .655). Furthermore, The effect of Informal Relationship on the Exploitation of Technology has no significant effect on the Exploitation of Technology (b = 0.1002 t = -0.346, p = .7298). Also, the interaction effect between Knowledge Transfer and Informal Relationship on the Exploitation of Technology has been found to have no significant effect (b = 0.0724, t = 0.9873, p = .3252). From this it follows that hypothesis 5 is not supported. When checking for the conditional indirect effects of Trust on the Exploitation of Technology with a 95% bias corrected bootstrapped confidence, it has been found that zero (0) lies not within the LL and UL (Table 13). From this it follows that the conditional indirect effects of Trust on Exploitation of Technology, mediated by Knowledge Transfer and moderated by Knowledge Asymmetry and Informal Relationship are significantly different from zero. Furthermore, in all regressions in this analysis it has been found that the control variables Firms size and Industry have no significant effect.

(39)

39 The outcomes for the path coefficients are indicated in the figure below (Figure 3).

Table 11: Mediation Analysis with mediator variable Figure 3: Conceptual Model with outcomes, including moderators

(40)

40

Table 12: Mediation Analysis with dependent variable

Table 13: Condition Indirect effects

It is shown that in a simple mediation analysis the hypotheses 1, 2, and 3 are supported. However, when checking for moderation none of the hypotheses are supported. The results of the analysis in comparison with the hypotheses are indicated in Table 14.

(41)

41 4.2 – Results post-hoc analysis

This part covers the qualitative results of this research, the post-hoc analysis. The collected data comes from the qualitative research and all findings are based on this collected interview data. All elements of the proposed conceptual model will be covered in this part. The first part consists of a matrix with the most important findings. Second, the main results will be discussed. Last, surprising factors from the interviews are discussed.

4.2.1 – Matrix

Below, in Table 15, the most important quotations are indicated from the qualitative dataset (e.g. the nine cases). All other important findings that relate to this research are indicated in appendix VII, which consists of the cross-case analysis and is also used for the post-hoc analysis.

The role of supplier-buyer relationships on the organisation’s ability to exploit emerging technologies : a mixed methods study in the context of artificial intelligence based robotic process automation

Master’s Thesis Digital Business