Does innovation have a moderating effect on the relationship of network embeddedness on firm performance within a niche market?

(1)

Master Thesis

Student: Danielle Bitter 11414189

MSc. Business Administration, Strategy track

University of Amsterdam, Faculty of Economics and Business Supervisor: dr. N. E. Betancourt

University of Amsterdam, Amsterdam Business School Date: 23th of June 2017

(2)

2

Statement of originality

This document is written by Danielle Bitter 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 study examines the relationship of network embeddedness on firm performance and tries to determine if innovation has a moderating effect on this relationship all within a niche market. This research is conducted in the pharmaceutical niche for manufacturing HIV AIDS medicines. I predict that network embeddedness will have a positive effect on firm

performance within this niche market accordingly to the findings of Echols and Tsai (2005). Furthermore, I posit that innovation will negatively moderate the effect of network

embeddedness on firm performance. The data consists of the number of the solely filled patents, jointly filled patents, and quarterly turnover, number of products sold multiplied by the price of the product, generated by HIV AIDS medicines for companies manufacturing HIV AIDS medicines. Using this data I found no significant relationship between the level of network embeddedness, the number of jointly filled patents in a given year by a company divided by the total number of jointly filled patents of a company, on firm performance. Moreover I was unable to determine a significant relationship of the moderating effect, innovation x network embeddedness, on firm performance. These findings are inconsistent with my predicted findings. Limitations in the data availability could be a viable explanation for the insignificant results found in this research.

(5)

5

Introduction

Networks have been a widely researched subject in the scientific literature. Especially the managerial contributions originating from this subject is what attracts the attention of many researchers. Studies conducted by Baum et al. (2000) and Uzzi (1997) have linked networks to firm performance. More specifically, Ozcan and Eisenhardt (2009) state that “although a single tie of a firm might be useful a firm’s portfolio[egocentric network] of ties is more likely to be crucial to firm performance” (p. 246). Overall, it is apparent that network research has a significant importance for managerial research.

Moreover, the degree of how affiliated a firm is within a network, network embeddedness, also has some implications for firm performance for companies within a niche (Echols & Tsai, 2005). Network embeddedness is defined as the routinization and stabilization [strength] of linkages among members as a result of a history of exchanges and relations within a group or community (Gulati, 1998). They state that network embeddedness has a positive effect on firm performance within a niche. A niche implies that it is

distinctively different than the rest of the industry (Echols & Tsai, 2005).

However, not only networks and network embeddedness can lead to an enhanced firm performance. Innovation has also been considered to be strongly correlated with firm

performance. Although, research is not conclusive of what the precise relationship between innovation and firm performance is for firms of different sizes (Rosenbusch et al., 2011). Maula et al. (2013) state that having different types of ties, e.g. ties within a network and ties outside of that network, can lead to timely attention to recognize disruptive technologies. This argument complements the study of Cohen and Levinthal (1990), which states that “accessing knowledge from across boundaries is an important driver of innovative performance for organizations” (Cohen & Levinthal, 1990 in Tortoriello & Krackhardt, 2010). It is evident that having different types of ties, ties within a network and ties outside of

(6)

6 that network, and thus having different information sources are an important source of

innovation itself (Cohen & Levinthal, 1990) and reduces the chances of becoming obsolete due to radical innovations (Maula et al., 2013).

Thus, it seems that there is an apparent paradox at play when innovation and network embeddedness are considered in relationship to firm performance within a niche. On the one hand, the degree of network embeddedness leads to an enhanced firm performance within a niche on the assumption that the network is relatively homogeneous. While on the other hand innovation performance, a source of firm performance, is mainly driven by accessing

different information sources which implies a sense of heterogeneity. It would seem impossible for a company to maintain a high degree of network embeddedness and a high degree of innovativeness due to physical restraints in maintain all these different ties. Accordingly, a company that is embedded within a niche market has to become a bit less embedded for it to be innovative based on the statement Cohen and Levinthal (1990). The reason for the positive relationship of network embeddedness on firm performance is the close interactions within that niche. Hence, innovation influences the relationship of network embeddedness on firm performance within a niche market.

The scientific literature has not addressed this paradox so far and has therefore left a gap in the literature with significant managerial implications left unhandled. For that reason, this study aims to address and fill this gap in the literature by examining how innovation affects the relationship of network embeddedness on firm performance within a niche market.

The study will be conducted within a pharmaceutical niche market for the

manufacturing of HIV AIDS medicines. This way the statement made in the article of Echols and Tsai (2005) about the positive relationship of network embeddedness can be tested and subsequently if, and what for effect, innovation might have on this relationship within this niche market. I posit that network embeddedness will have a positive effect on firm

(7)

7 performance accordingly to the findings of Echols and Tsai (2005). Additionality I predict that innovation will negatively moderate the effect of network embeddedness on firm performance.

The data for this study was gathered by combining the information about the solely and jointly filled patents, found in Google advanced patent search, and the turnover, number of HIV AIDS specific medicines sold multiplied by the price of the product, found at the publicly available quarterly fillings (10-Q).

As mentioned above this research will contribute to the existing scientific literature by extending current knowledge and integrating the knowledge about niches, networks,

innovation, and firm performance, by closing a gap in the literature. Additionally the results of this research might have significant managerial implications for firms operating within a niche market. A negative moderating effect of innovation on the relationship of network embeddedness on firm performance requires a different strategy to optimize firm

performance than a positive moderating effect.

This paper will first discuss the main theoretical insights about niches, network embeddedness, and innovation which will lead up to the research question and hypotheses in the literature review. In turn, the methodology used to conduct this research will be

addressed. Afterwards, the results found in the data analysis will be discussed. Followed, is the discussion regarding the results. And as last, the conclusion of the overall research question, hypotheses, and research will be given.

(8)

8 Conceptual model

Niche market

Innovation

(9)

9

Literature review

The following section discusses the main insights in the existing literature on niche markets, firm performance, network embeddedness, and innovation, which will lead up to the gap in the literature and consequently my research question and hypotheses. First of all I will explain what niche markets are, next I will elaborate on what drives firm performance within a niche. Subsequently I will explain network embeddedness. And finally, I will introduce innovation and how this might influence the relationship of network embeddedness on firm performance. Afterwards I will present the literature gap, research question, and my

hypotheses. What is a niche?

Echols and Tsai (2005) state that a niche represents the distinctiveness of that niche relative to the rest of the industry. Firms in a niche (niche-firms) therefore, have either distinctive products or processes that are significantly different from their competitors. Moreover, a niche can be further divided into either a market niche or a niche market. There is a significant but subtle difference distinguishing a niche market from a market niche. Therefore a niche market is not the same as a market niche and can therefore not be used interchangeably.

According to the definition given be Abernathy and Clark (1985) in King and Tucci (2002), a market niche occurs when a new technology creates a new market opportunity but this new technology does not obsolete the capabilities of the incumbent firms in the already existing market. In other words, a new technology arises which creates a niche within a market, making it possible for incumbent firms to enter that niche and abstain from becoming obsolete. An example of a market niche could be phones for older people who have bigger buttons that have an easier operation system on their phone. There is a new technology, a simpler operation system, but this technology does not obsolete incumbent firms as the

(10)

10 technology could be copied by the incumbent firms. They just might choose not to invest in this niche due to any cost restraints.

Parrish et al. (2006) on the other hand have more difficulty giving an explicit definition that defines niche markets. They state that there is no consistent definition for niche markets, so therefore they use Kotler’s (2003) 5 key characteristics to describe niche markets. Namely, the customers in the niche have a distinct set of needs; they will pay a premium price to the firm that best satisfies their needs; the niche is not likely to attract competitors; the niche marketer gains certain economies through specialization; and the niche has size, profit, and growth potential. An example of a niche market is the market for shock-, water-, and dustproof cameras, adventure camera for short, for example a go pro. It is a market where firms have a distinct product, which is significantly different from its

competitors, to serve the demand of the customer. The customer will pay a premium price for that camera compared to a regular camera. It is not likely to attract competitors due to the specialization while gaining a certain economy due to specialization and the market has size, profit and growth potential. Mostly what defines the adventure camera market as a niche market instead of a market niche, is that its technology will make incumbent firms obsolete for this specific group of customers. Incumbent firms do not have the technology to produce such cameras to join this niche, therefore making it a separate niche market.

What drives firm performance in a niche?

According to Echols and Tsai (2005) firms in a niche experience a positive firm performance as a result of network embeddedness. They explain the positive relationship between network embeddedness and firm performance by stating that firms in a niche can acquire the necessary detailed and fine-grained information, which is required to operate in a distinctive market, from their network. In addition, it also allows them to keep their

(11)

11 distinctive knowledge within the network and avoid any knowledge spillovers to other

companies in the industry.

The study done by Ingram and Roberts (2000) acknowledges that competing managers who are embedded in a cohesive network of friendships, can improve

organizational performance due to the mechanisms of enhanced collaboration, mitigated competition, and better information exchange.

Moreover Burt (2001) states that social capital is a social structure, a network, that can create a competitive advantage for certain individuals or groups who are pursuing their ends. The better connected you are, the higher the return is. This is in line with the arguments of Echols and Tsai (2005) and Ingram and Roberts (2000) that the degree of network

embeddedness will have a positive effect on firm performance. What is network embeddedness?

Gulati (1998) describes network embeddedness as the routinization and stabilization [strength] of linkages among members as a result of a history of exchanges and relations within a group or community. In addition, the definition of structural embeddedness is also used to further the understanding of network embeddedness (Grewal & Mallapragada, 2006). Namely, structural embeddedness, as explained by Zukin and DiMaggio (1990), is the contextualization of economic exchange in the pattern of ongoing interpersonal relations. Uzzi (1996) defines structural embeddedness as the relational quality of interactor exchanges and the architecture of network ties. Structural embeddedness, the exchanges of ongoing interpersonal relations (Zukin & DiMaggio, 1990), and the quality of interactor exchanges and how these are positioned (Uzzi, 1996) can therefore be seen as the quality of ongoing interpersonal exchanges. This definition of structural embeddedness helps to explain the definition of network embeddedness, given by Gulati (1998), by getting a better

(12)

12 The definition of network embeddedness that will used in this thesis will be the routinization and stabilization [strength] of linkages among members as a result of a history of exchanges and relations within a group or community following the definition of Gulati (1998).

A way to visualize network embeddedness is to look at different cliques within a high school. A person who has reciprocal ties with every person within the clique is viewed as more embedded in that clique than a person who has a few reciprocal ties within that clique. This logic also applies for reciprocal ties versus nonreciprocal ties. Thus, reciprocal ties are stronger than non-reciprocal ties. The same goes for frequency, someone who has frequent interactions with every person within the clique is viewed as more embedded in that clique than a person who has a few interactions with everyone within that clique. Hence, the degree of network embeddedness depends on the strength, and frequency of the linkages between ties.

Innovation as moderator on network embeddedness and firm performance.

Innovation consist of a large body of research which covers multiple disciplines. Due to the many definitions and uses for innovation in the literature it is necessary to have a clear understanding of the definition regarding innovation used in this thesis. One way to look at innovation is to divide innovation into incremental innovation and radical innovation.

What is innovation?

Innovation can be divided into two types of innovation namely, incremental innovation and radical innovation. Dewar and Dutton (1986) state that incremental and radical innovation refer to different types of technological process innovation. They describe incremental innovation, based on the definition of Munson and Pelz’s (1979), as minor improvements of simple adjustments in current technology. On the other hand, Dewar and Dutton (1986) define radical innovation as fundamental changes, that represent revolutionary changes in technology, that represent clear departures from existing practice (Duchesneau,

(13)

13 Cohn and Dutton, 1979; Ettlie, 1983 in Dewar & Dutton, 1986). The degree of new

knowledge embedded in the innovation accounts for the main difference between incremental and radical innovation. (Dewar & Dutton, 1986).

According to Maula et al. (2013) technological discontinuities, or radical innovations are fundamental shifts from one dominant technology to another. They often change the existing industry, e.g. new rules of competition, and new market leaders. These technological discontinuities or radical innovations, happen mostly at the fringes of the industry.

The definition of Maula et al. (2013) of radical innovations lends itself to the comparison of a niche, as defined by Kotler’s (2003) 5 key characteristics or as defined by Abertnathy and Clark (1985) in King and Tucci (2002). Radical innovation implies a fundamental shift from a dominant technology to another technology and changing the industry (Dewar & Dutton, 1986; Maula et al., 2013). A niche implies the rise of a new technology, which may or may not obsolete the existing incumbent firms. This implies a shift in dominant technology within an industry and thereby creating a niche. Hence, radical innovation could be used to explain the existence of niches. The focus in this study will therefore lie on the incremental type of innovation within a niche market, as it accounts for minor improvements of current technology.

Innovation itself will be defined by Zaltman et al.’s (1973) definition of innovation namely, it is an idea, practice, or material artefact perceived as new by the relevant unit of adoption (Zaltman et al., 1973, in Calantone, Cavusgil, Zhao, 2002). This way the definition of innovation lends itself to represent either a tangible adjustment of some sort or an

intangible adjustment, for example a process, as incremental innovation is the focus of this research.

(14)

14 Innovation is believed to be the key to survival (Kim and Maubourgne, 2005 in

Rosenbusch et al., 2011; Kline & Rosenberg, 1986) which implies a correlation to firm performance. Additionally, innovation is also linked as a source for competitive advantage (Peteraf, 1993). According to the Resource Based View (RBV), a requisite for competitive advantage is heterogeneity. Without heterogeneity, every firm will have the same access to the exact same resources. Making it thus theoretically impossible to make a profit according to Peteraf (1993) since all firms will end up in an equilibrium, due to the access of exactly the same resources. Heterogeneity of resources across firms however, allows firms to use certain bundles of resources in a different and more efficient way which can lead to earn more rent (Peteraf, 1993). The definition of Zaltman et al. (1973), innovation as an idea, practice, or material artefact perceived as new by the relevant unit of adoption, could be used to describe the different types of usage of bundles of resources as innovation. It is important to note that not all resources are innovations or that not all innovations are resources. Nonetheless, if there would be no heterogeneity in resources across firms everybody would come up with the same innovation, but that innovation would not lead to a competitive advantage. Hence innovation can lead to competitive advantage however, it is not a given.

This complements the statement made by Cohen and Levinthal (1990) who state that “accessing knowledge from across boundaries is an important driver of innovative

performance for organizations” (Cohen & Levinthal, 1990 in Tortoriello & Krackhardt, 2010). Firm performance will refer for future references to the financial performance of a firm.

The assumption of heterogeneity as a source of innovation leads to an interesting point regarding network embeddedness and its effect on firm performance. A high degree of network embeddedness implies a high degree of strong ties (reciprocity) and high degree of frequency. In other words, a company within a niche with a high degree of network

(15)

15 embeddedness would frequently interacts with the other companies in that niche. As pointed out by Echols & Tsai (2005) this can be beneficial to the company due to the fact that it allows the company to get detailed and fine-grained information with regards to the market and avoid any potential knowledge spill-overs. Yet, innovation calls for a degree of

heterogeneity of information sources [ties].

Within a niche there would be a degree of heterogeneity among the firms since

companies are not exact duplicates of each other. Thus a firm that is innovative doesn’t imply that the firm is not embedded within a network. However, the degree of innovativeness does have some implications of the degree of network embeddedness of that firm and will

therefore have an impact on the relationship of network embeddedness on firm performance. Since a firm can only maintain a number of ties at once it is impossible for a firm to have a high degree of frequent interactions with the other companies within a niche and attend to a high number of ties across different boundaries for a source of information as suggested by Cohen and Levinthal (1990) in Tortoriello and Krackhardt (2010). I therefore posit that innovativeness and embeddedness are opposite sides of a spectrum in which a firm balances the degrees of embeddedness and innovativeness. Whereas, network embeddedness has a positive effect on firm performance resulting from the interactions with other members within that niche, it would seem that innovation, keeping the spectrum logic in mind, effects this relationship in a negative way based on Cohen and Levinthal’s (1990) statement that different sources of information are an important driver behind innovation.

Literature gap, research question, and hypotheses

As stated above it is evident that network embeddedness has a positive effect on firm performance within a niche. Additionally, it is also clear that there is a relationship between innovation and network embeddedness and a relationship between innovation and firm performance. However, the literature lacks to explain the relationship of innovation on the

(16)

16 (positive) relationship of network embeddedness on firm performance within a niche market. This leaves a gap in the literature.

The motives to investigate this from a scientific and managerial perspective are considerable. First of all, it will fill up a gap in the literature and extend our knowledge of innovation, network embeddedness, firm performance, and how these concepts are all related to each other within a niche market. Second of all, it has a managerial implications, as it will help practitioners in the field understand if and how the relationship of innovation might benefit firm performance in regards to network embeddedness within a niche market.

Therefore, the research question will be as follows: What is the moderating effect of innovation on the relationship of network embeddedness on firm performance within a niche market?

The research question is going to be answered on the basis of two hypotheses. First of all, a hypothesis about the relationship of network embeddedness and firm performance within a niche market needs to be established. According to Echols & Tsai (2005) there is a positive relationship between network embeddedness and firm performance within a niche. Despite the fact that they researched this relationship in a product and a process niche. Where a product niche is a niche where firms differ from the rest of the industry through products. And where companies in a process niche differ from the rest of the industry by processes. The difference between a niche and a niche market would not be of that magnitude, that it would influence the relationship of network embeddedness on firm performance. Hence, I expect that network embeddedness is positively related to firm performance within a niche market. Therefore my first hypothesis will be:

H1: Network embeddedness has a positive effect on firm performance within a niche market.

(17)

17 The second hypothesis will examine the role of innovation on the relationship of network embeddedness on firm performance. As mentioned by Echols and Tsai (2005) the reason why network embeddedness has a positive effect on firm performance is due to the fact that companies can acquire the necessary detailed and fine-grained information from their network and avoid knowledge spillovers. As mentioned above, it is not feasible for a company to maintain strong ties within a network and outside of that network as a result of physical limitations. However, innovation requires ties across different boundaries (Cohen & Levinthal, 1990). I predict that innovation will have a moderating effect on the relationship. Moreover I predict this to be a negative effect. Therefore my second hypothesis will be:

H2: Innovation will negatively moderate the effect of network embeddedness on firm performance.

(18)

18

Methodology

This section will elaborate on the research approach and design used to conduct this study. First the strategy used to gather the sample will be discussed. Additionally, the

operationalization of the variables will be explained. Finally, the methods used to analyze the data will be discussed.

Sampling strategy

The pharmaceutical industry was selected to provide a single-industry study of the effect of innovation on network embeddedness on firm performance in a niche market. The niche that this research is set to study, is the market for HIV AIDS medicines. This niche meets Kotler’s (2003) 5 key assumptions for it to be defined as a niche market. Firstly, the customers have a distinct set of needs, in this case the need for HIV AIDS medicines. Secondly, the customers will pay a premium price to the firm that best satisfies their needs. Thirdly, the niche is not likely to attract competitors due to the high costs of developing new medicines. Fourthly, the companies in the niche can gain certain economies through

specialization. Lastly, the HIV AIDS niche has size, profit and growth potential namely, there is an increase in people who are accessing HIV AIDS treatment (“Global HIV statistics”, 2016). Nevertheless the most important characteristic of this niche is the fact that the

technology required for this niche prevents incumbent firms to enter and therefore making it a niche market as supposed to a market niche (Kotler, 2003; King and Tucci, 2002).

The sample was further restricted to pharmaceutical manufacturing companies who produce HIV AIDS medicines and sell the products in the United States. This ensures that the research investigates a niche market were the same type of innovation will be measured. Archival data methodology was used to gather the data for the independent, dependent, and moderating variable.

(19)

19 Thirty-three companies were identified as manufacturing firms of HIV AIDS

medicines on the basis of the United States Food and Drugs Administration (FDA) approved list of HIV AIDS medicines with the corresponding manufactures. This included

manufactures of brand name medicines and generic medicines for HIV AIDS medication. The sample was further reduced from thirty-three companies to five companies, by including companies who had publicly available quarterly fillings, a 10-Q form, that specified the revenue generated by HIV AIDS medicine separately from the rest of the pharmaceutical revenue. This allows the study to only examine the effect of network embeddedness, or any moderating effects of innovation, on firm performance generated by HIV AIDS medicine and thus specific to this niche market. The drastic reduction of the number of companies in the sample is due to the fact that all the generic brand manufactures of HIV AIDS medicine were private companies who had no publicly available records of the quarterly revenues. This immediately reduced the sample size from thirty-three companies down to twelve companies. Nevertheless, out of the twelve remaining companies, only five companies had publicly available records of the quarterly results specifying the revenue generated by HIV AIDS medicine.

Google advanced patent search was used to gather data about the independent variable, network embeddedness, and moderating variable, innovation. This way the search could be adjusted for the niche market, HIV AIDS medicine, a search command that included HIV along with the company name was used to search the patent information in the database. This ensured that the data gathered for these variables were exclusively applicable for the HIV AIDS niche market and would further reduce any noise in the sample. Data about the dependent variable, firm performance, was gathered by assessing the quarterly reports (10-Q form) distributed by the companies on their websites.

(20)

20 The sample was reduced to a total of five companies, with a total of a hundred

observations, with the independent, dependent, and moderating variable data spanning from 2012 to 2016, due to limitations in data availability of the quarterly fillings. Although the sample size might seem quite small it represents the real data as well as possible by eliminating as much noise as feasible. In addition, this study investigates five out of the thirty-three companies present in the niche market which represents roughly 15% of that niche market.

Independent variable

The independent variable of this study is network embeddedness. As mentioned above network embeddedness represent the routinization and stabilization [strength] of linkages among members as a result of a history of exchanges and relations within a group or

community (Gulati, 1998). This variable will be measured by the frequency, number, of joint patent filings of a company with other companies in a given year. When a firm forms a collaboration with another company through a joint filling this is regarded as an exchange and implies a relationship. The more joint fillings a company pursues, the more embedded a company gets within that network. Thus frequency can be regarded as the routinization of linkages.

Due to the fact that this research in conducted within a niche market, all the variables have to specifically address, and only address, this niche market and have no overlap with adjacent niche markets. Patents can assure this exclusivity of measurement, given the fact that each patent was gathered using a search for HIV in the patents description.

Each independent joint patent filing will represent a collaboration. In other words, if company A files for two patents with company B these joint fillings will be regarded as two separate collaborations. This is in line with Echols and Tsai (2005) who, based on Burt’s (1992) network redundancy, have operationalized network embeddedness as the number of

(21)

21 times each firm coinvested in IPOs with partner venture capital firms during a given time. In this study both application and granted patents were included, given the fact that each of these patents represents a formed collaboration. To be consistent throughout, the publication date is used as measure for both applications and grants.

As stated by Scott (2000) in Echols and Tsai (2005) network redundancy can be biased by size and can therefore not be compared across networks of different sizes. To control for the different sizes of networks, network embeddedness will be measured by dividing the frequency of joint patent fillings of a company in a given year by the total number of jointly filled patents of that company in the period of 2012-2016.

Dependent variable

The dependent variable in this research is firm performance. Firm performance refers to the financial performance of the company. Due to constraints of researching within a niche market, limited data availability, firm performance will be measured as the turnover produced by HIV AIDS medicine sales of a given quarter instead of the usual return on assets (ROA) (Hitt et al, 1997). Turnover will refer to the number of products sold, HIV AIDS medication, multiplied by the price of the product. The turnover will be reported in millions of U.S. dollars ($). Turnover that was originally reported in a different currency was converted into U.S. dollars with the appropriate exchange rate for that corresponding quarter. The turnover was not limited to the U.S. market but represents the global sale of HIV AIDS medicine of that company in a given quarter.

Moderating variable

The moderating variable in this research is innovation. Innovation in this research refers to incremental innovation. Incremental innovation is referred to as minor

improvements of simple adjustments in current technology (Munson & Pelz, 1979).

(22)

22 year. Patents are a widely used measurement for innovation throughout the scientific

literature (e.g. Mansfield, 1986; Ejermo, 2009; Hagedoorn & Cloodt, 2003; Nesta & Saviotti, 2005). Hence, patents seem to be a logic choice as a measurement of innovation. As

mentioned before, the difference distinguishing incremental innovation from radical innovation is the degree of new knowledge embedded in the innovation (Dewar & Dutton, 1986). However, radical innovations would also be patented, as is the same for incremental innovations. Due to time constrictions the amount of patents the five companies have filled with regards to the HIV AIDS medicines, and a lack of knowledge in the field of

pharmaceutical patents, it is impossible to sort through all the patents identifying them as a radical innovation or as an incremental innovation. It would be reasonable to expect more incremental innovations among the patent fillings than radical innovations, therefore all the patents will be treated as incremental innovations.

Statistical model

Due to the fact that the dependent variable is count data, a normal linear regression is insufficient in analyzing the data. For this reason, models equipped to analyze count data need to be considered. A Poisson regression or a negative binomial regression is fit to analyze count data. However, the dependent variable data is overly dispersed, and thus a Poisson regression is not adequate for analyzing the data. Hence, a negative binomial regression is used to analyze the data.

(23)

23

Results

The following section discusses the results of this research. First of all, an examination of the dataset is conducted to account for all the constructs used in the analysis. Secondly, an overview of the descriptive statistics for the variables of the study is presented, a correlation analysis is performed and the significant correlations are reported. Finally, an negative binomial model is used to conduct a regression analysis.

Preparing the data

First of all, the data set is checked to see whether there are any missing values and if all the variables/constructs are available in the dataset to conduct the analysis. The construct of network embeddedness has to be computed before it can be used in any analyses. To do so, NE_Frequency is divided by NE_Total_Ties and multiplied by 100 to compute the variable NE_overall. If NE_overall wouldn’t be multiplied times a 100 it would result in a very small number compared to the rest of the dataset. This is also in line with the article of Echols and Tsai (2005), who also multiplied the number of network embeddedness by a 100 to correct for the otherwise small number.

Descriptive statistics and correlation analysis

The sample used to run the descriptive statistics and bivariate correlation analyses, contains a 100 observations in the time period 2012-2016. These observation were gathered in the niche market of HIV AIDS medicines and represents the changes of the variables listed in this study of 5 manufacturing companies within this time period. Table 1. presents an overview of the descriptive statistics which includes the mean, standard deviation and

variance for all the separate variables. This table also includes a bivariate correlation analysis of the variables.

(24)

24 As reported in table 1. the moderating variable, Innovation, has a weak (r = 0,24) positive correlation with the dependent variable, Performance. In addition, the interaction variable, NE_overall_Innovation_Interaction, had a slightly stronger, medium (r = 0,30), positive correlation with Performance. NE_overall_Innovation_Interaction also has a highly positive correlation with NE_overall (r = 0,69) and Innovation (r = 0,70). However, this is to be expected as the interaction term is comprised out of the two variables and therefore should not be addressed any further.

(25)

25 Table 1 Descriptive statistics and correlation matrix

(26)

26 Regression analysis

When analysing the data it is evident that the dependent variable, Performance, consists of count data. Namely, it represents the quarterly sales, the price per product, HIV AIDS medicine, multiplied by the number of products sold, of each company. A Ordinary Least Squares (OLS) regression assumes that the dependent variable is continuous and can assume negative and positive values. However, count data can only be measured in whole and is bounded by zero and can therefore not assume a negative value (Knudsen et al., 2007). Therefore a Poisson regression or a negative binomial regression would be a better model fit for the data than a OLS regression.

A Poisson regression provides the best fit as a model when the mean of the count is higher than the variance of the count. While a negative binomial regression provides a better fit when the variance of the count is higher than the mean of the count. Thus, a Poisson regression is not an appropriate model to analyse count data in the case of overdispersion of the dependent variable (Long & Freese, 2006; Barron, 1992). As summarized in table 1. it is evident that the variance of the count (= 717481) greatly exceeds the mean of the count (m = 826). Therefore the model used to analyse the data, is a negative binomial regression, this analysis will be conducted in SPSS package 21.

To ensure this model was appropriate for the data, the goodness of the fit was checked by assessing the Pearson Chi-Square of the negative binomial regression. A value below 0,05 implicates that the model does not fit the data well and other analyses should be considered. A value above 0,05 indicates that the model does fit the data well. As seen in appendix A the Pearson Chi-Square (value/df = 0,76) is above the 0,05 threshold and therefore the negative binomial regression fits the data well.

Moreover, the Omnibus Test was assessed by checking its significance level to see whether the model is statistically significant. A p-value above 0,05 indicates a non-significant

(27)

27 model, whereas a p-value below 0,05 implies that the model is statistically significant. As reported in appendix B the p-value of the Omnibus Test ( p = 0,011) is below the threshold of 0,05 and therefore the model is statistically significant.

Table 2. represents the negative binomial regression conducted in SPSS package 21. The effect of the interaction term, NE_overall_Innovation_Interaction, is not significant (effect= <0,01, p= 0,39). Thus, the effect of network embeddedness, NE_overall, on firm performance does not depend on innovation. However, network embeddedness is also nonsignificant (effect= <-0,01, p= 0,75). This implies that network embeddedness does not have an effect on firm performance. Hence, none of the predictors tested in this regression have an effect on firm performance.

Table 2 Negative binomial regression with dependent variable Performance

B SE Test Statisticsa _df _p

Intercept 6,20 0,41 227,40 1 0,00**

NE_overall <-0,01 0,02 0,10 1 0,75

NE_overall_Innovation_Interaction <0,01 <0,01 0,75 1 0,39

Innovation 0,01 -0,03 0,38 1 0,54

a Wald chi squares are used for the chi square values. *. p-value is significant at 0.05 level.

**. p-value is significant at 0.01 level.

(28)

28

Discussion

This research was conducted to find out if and how innovation could moderate the

relationship of network embeddedness on firm performance. Unfortunately, the results of the data analyses are not significant and are therefore not congruent with the proposed

hypotheses. In this discussion section, I will elaborate on the findings of this study and relate them to the literature. Afterwards, I will discuss the limitations of this study and make

recommendations for future research. Finally, I will present the contributions of this research.

Findings

The question this research was set out to answer was: What is the moderating effect of innovation on the relationship of network embeddedness on firm performance within a niche market? As mentioned before, the results of the negative binomial regression used to analyze the data turned out to be not significant. This implies that neither network embeddedness nor innovation had an effect on firm performance within a niche market. This contradicts my first hypothesis which assumes a positive relationship of network embeddedness and firm

performance. Additionally, these results also imply that innovation does not moderate the relationship of network embeddedness on firm performance within a niche market. These findings contradict my second hypothesis, which assumes that innovation will negatively moderate the effect of network embeddedness on firm performance.

The findings of this study are inconsistent with the literature, as Echols and Tsai (2005) did find a positive relationship of network embeddedness on firm performance within a niche. The slightly different definitions of niches could not have been the reason for this incongruence. The study of Echols and Tsai (2005) investigated the market of venture

capitalists, this market does not obsolete incumbent firms and can therefore not be defined as a niche market (Parrish et al., 2006). Despite the difference in the definition of the niche, the

(29)

29 positive relation of network embeddedness Echols and Tsai (2005) found was due to the fact that the companies can acquire the necessary detailed and fine-grained information from other companies within the niche to operate within that niche. Hence, the definition of the niche has no impact on the relationship of network embeddedness and firm performance. Limitations and recommendations

A limitation of this research is, as mentioned before, the limited availability of the financial performance data of the companies specific to this niche market. Therefore, only a small subgroup of the niche market (15%) could be investigated and limits the

generalizability of the results to the rest of the niche market. Regardless, the sample is inadequate for generalizing the results to other niche markets.

Another limitation of this research is the fact that it didn’t take the mixture of ties within the network into account. Research has pointed out that networks where the composition consists of a good mix of ties can lead to an enhanced firm performance. Networks were there is a mixture of strong and weak ties are likely to have a high firm performance (Rowley et al., 2000; Uzzi, 1997). “Networks that are composed of many

diverse partners (Baum et al., 2000) provide information, flexibility and resource benefits that are likely to enhance firm performance” (Ozcan & Eisenhardt, 2009, p. 247).

For future research purposes the compositions of the networks of firms should be taken into account. Namely, the finding of Echols and Tsai (2005), that network

embeddedness has a positive effect on firm performance, is in sheer contrast with the findings presented in the study of Baum et al. (2000), which posit that networks with mixed,

heterogeneous, ties are more likely to have an effect on enhanced firm performance. Firms operating in a niche can be viewed as relatively homogeneous, compared to firms in the rest of the industry, due to the specific products or processes required for that niche. This implies that firms in a niche would have a homogenous network of ties within that niche.

(30)

30 Contributions

The contributions of this research are of significant importance, they might not be of much managerial importance in hindsight. They do however have great significant scientific importance. This research uncovers the fact that the scientific literature falls short in

explaining network embeddedness and innovation in relation to firm performance with regards to niches. This study also attempts to confirm the findings of the study of Echols and Tsai (2005), which it does not. Even though the results of the study were not significant, this study attempts to enhance the understanding of innovation in relation to network

embeddedness and firm performance within a niche market. All and all, this research sheds light upon a not so commonly researched area which implies that further investigation of this topic within the scientific literature is desperately needed.

(31)

31

Conclusion

In this research it became apparent that none of the predictors, network embeddedness and innovation had any effect on firm performance within a niche market. Moreover, there was no significant effect of innovation on the relationship of network embeddedness on firm performance to be found. These findings do not support the findings of Echols and Tsai (2005) that network embeddedness has a positive effect on firm performance within a niche. The results of this study answer to the research question regarding what the moderating effect of innovation on the relationship of network embeddedness on firm performance within a niche market is, by demonstrating that there was no significant relation to be found. This research does however emphasizes the importance of future research to be conducted in this field. It becomes apparent that there is little understanding in the scientific literature of niches to how network embeddedness and innovation are related with regards to firm performance. This research attempts to further the understanding of network embeddedness and innovation in relation to firm performance within a niche market. However, future research should be conduct within this field to get an even better understanding of niches and how innovation, network embeddedness and firm performance are related to niches and to each other.

(32)

32

References

Barron, D. N. (1992). The analysis of count data: Overdispersion and autocorrelation. Sociological methodology, 179-220.

Baum, J., Calabrese, T., & Silverman, B. 2000. Don’t go it alone: Alliance network composition and startups’ performance in Canadian biotechnology. Strategic Management Journal, 21: 267–294.

Burt, R.S. (1992). Structural Holes: The Social Structure of Competition. Harvard University Press: Cambridge, MA.

Burt, R. S. (2001). Closure as social capital. Social capital: Theory and research, 31-56. Calantone, R. J., Cavusgil, S. T., & Zhao, Y. (2002). Learning orientation, firm innovation

capability, and firm performance. Industrial marketing management, 31(6), 515-524. Cohen, W. M., & Levinthal, D. A. 1990. Absorptive capacity: A new perspective on learning

and innovation. Administrative Science Quarterly, 35: 128–152.

Dewar, R. D., & Dutton, J. E. (1986). The adoption of radical and incremental innovations: An empirical analysis. Management science, 32(11), 1422-1433.

Echols, A., & Tsai, W. (2005). Niche and performance: the moderating role of network embeddedness. Strategic Management Journal, 26(3), 219-238.

Ejermo, O. (2009). Regional Innovation Measured by Patent Data—Does Quality Matter? Research Paper. Industry and Innovation, 16(2), 141-165.

Global HIV statistics. (2016, November). Retrieved from

(33)

33 Grewal, R., Lilien, G. L., & Mallapragada, G. (2006). Location, location, location: How

network embeddedness affects project success in open source systems. Management Science, 52(7), 1043-1056.

Gulati, R. (1998). Alliances and networks. Strategic management journal, 19(4), 293-317. Hagedoorn, J., & Cloodt, M. (2003). Measuring innovative performance: is there an

advantage in using multiple indicators?. Research policy, 32(8), 1365-1379. Hitt, M. A., Hoskisson, R. E., & Kim, H. (1997). International diversification: Effects on

innovation and firm performance in product-diversified firms. Academy of Management journal, 40(4), 767-798.

Ingram, P., & Roberts, P. W. (2000). Friendships among Competitors in the Sydney Hotel Industry 1. American journal of sociology, 106(2), 387-423.

King, A. A., & Tucci, C. L. (2002). Incumbent entry into new market niches: The role of experience and managerial choice in the creation of dynamic

capabilities. Management science, 48(2), 171-186.

Kline, S. J., & Rosenberg, N. (1986). An overview of innovation. The positive sum strategy: Harnessing technology for economic growth, 14, 640.

Knudsen, H. K., Ducharme, L. J., & Roman, P. M. (2007). Job stress and poor sleep quality: data from an American sample of full-time workers. Social science &

medicine, 64(10), 1997-2007.

Kotler, P. (2003), Marketing Management, 11th ed., Prentice-Hall, Upper Saddle River, NJ. Long, J. S., & Freese, J. (2006). Regression models for categorical dependent variables using

Stata. Stata press.Mansfield, E. (1986). Patents and innovation: an empirical study. Management science, 32(2), 173-181.

Maula, M. V., Keil, T., & Zahra, S. A. (2013). Top management’s attention to discontinuous technological change: Corporate venture capital as an alert mechanism. Organization

(34)

34 Science, 24(3), 926-947.

Munson, F. C., & Pelz, D. C. (1979). The innovating process: A conceptual framework. Univ. Michigan, Ann Arbor, Working Paper.

Nesta, L., & Saviotti, P. P. (2005). Coherence of the knowledge base and the firm's

innovative performance: evidence from the US pharmaceutical industry. The Journal of Industrial Economics, 53(1), 123-142.

Ozcan, P., & Eisenhardt, K. M. (2009). Origin of alliance portfolios: Entrepreneurs, network strategies, and firm performance. Academy of Management Journal, 52(2), 246-279. Parrish, E. D., Cassill, N. L., & Oxenham, W. (2006). Niche market strategy for a mature

marketplace. Marketing Intelligence & Planning, 24(7), 694-707.

Peteraf, M. A. (1993). The cornerstones of competitive advantage: a resource‐based view. Strategic management journal, 14(3), 179-191.

Rosenbusch, N., Brinckmann, J., & Bausch, A. (2011). Is innovation always beneficial? A meta-analysis of the relationship between innovation and performance in

SMEs. Journal of business Venturing, 26(4), 441-457.

Rowley, T., Behrens, D., & Krackhardt, D. 2000. Redundant governance structures: An analysis of relational and structural embeddedness in the steel and semiconductor industries. Strategic Management Journal, 21: 369–386.

Scott, J. (2000). Social Network Analysis: A Handbook. 2nd edn SAGE Publications.

Tortoriello, M., & Krackhardt, D. (2010). Activating cross-boundary knowledge: The role of Simmelian ties in the generation of innovations. Academy of Management

Journal, 53(1), 167-181.

Uzzi, B. (1996). The sources and consequences of embeddedness for the economic performance of organizations: The network effect. American sociological review, 674-698.

(35)

35 Uzzi, B. (1997). Social structure and competition in interfirm networks: The paradox of

embeddedness. Administrative Science Quarterly, 42: 36–67.

Zaltman, G., Duncan, R., & Holbek, J. (1973). Innovations and organizations (Vol. 1973). New York: Wiley.

Zukin, S., & DiMaggio, P. (Eds.). (1990). Structures of capital: The social organization of the economy. CUP Archive.

(36)

36

Appendix A

Goodness of Fita Value df Value/df Deviance 70,624 96 ,736 Scaled Deviance 70,624 96 Pearson Chi-Square 73,233 96 ,763

Scaled Pearson Chi-Square 73,233 96

Log Likelihoodb _-766,165

Akaike's Information Criterion (AIC)

1540,330

Finite Sample Corrected AIC (AICC)

1540,751

Bayesian Information Criterion (BIC)

1550,751

Consistent AIC (CAIC) 1554,751 Dependent Variable: Performance

Model: (Intercept), NE_overall, NE_overall_Innovation_Interaction, Innovation

a. Information criteria are in smaller-is-better form.

b. The full log likelihood function is displayed and used in computing information criteria.

(37)

37

Appendix B

Omnibus Testa Likelihood Ratio Chi-Square df Sig. 11,168 3 ,011

Dependent Variable: Performance Model: (Intercept), NE_overall, NE_overall_Innovation_Interaction, Innovation

a. Compares the fitted model against the intercept-only model.

Does innovation have a moderating effect on the relationship of network embeddedness on firm performance within a niche market?

Statement of originality

Table of contents

Abstract

Introduction

Literature review

Methodology

Results

Discussion

Conclusion

References

Appendix A

Appendix B