The true innovative firm : micro, small and medium-sized firms in the Netherlands

Universiteit van Amsterdam

Amsterdam Business School

The true innovative firm:

micro, small and medium-sized firms in the Netherlands

Master thesis MSc. Business Studies

Entrepreneurship and Innovation track Supervisor: dr. W. van der Aa Second supervisor: dr. T. Vinig

Maarten Laurens van Bemmel Student number: 10658440

August 27, 2014 Word count: 13.567

2 Preface

In this research we have used an existing dataset called “MKBbeleidspanel”, which is computed and maintained by a company called Panteia. They conduct comprehensive research on various topics and offer these data to third parties for further research purposes. Further information about Panteia and their research can be found on their websites www.panteia.nl and www.ondernemerschap.nl. As this data is owned and maintained by Panteia am I not able to share this. For this reason can it be obtained by contacting them via their a-mail address: info@panteia.nl.

I would like to thank Panteia for disclosing their data, because I would never have been able to create such a rich and detailed dataset by myself. Secondly, I would like to thank Rob in ‘t Hout for his assistance and guidance with this data and dr. Wytze van der Aa for his guidance and advice regarding the thesis.

Abstract

This research examines the differences in innovativeness of different firm sizes in the Netherlands. It is found that Micro firms are significantly less innovative than Small and Medium-sized firms. The results show that firm size correlated quite weak to “innovative output” R = . 10 to R = .17, which is consistent with the results of Camisón -Zornoza et al., (2004) who found a positive R = .15 between firm size and innovativeness. The correlation between firm size and “process innovations” was moderately strong R = . 20 to R = .33, which suggest that with an increasing firm size an increase in “innovative output” and especially “process innovations” is expected. SME’s are advised to invest in process innovations, as these yield more success than investing in product or service innovations. Answers to the question why micro firms are less likely to innovate then small and medium-sized firms were found in the fact that they show a lower probability to conduct R&D, serve new customer segments, offer additional education to employees and have active

collaborations with other companies. Financial constraints were the main reasons for micro firms to not invest in innovative activities, which might have resulted in a lower

3

Index

1. Introduction p. 5

1.2 Theoretic background p. 7

1.3 The current study p. 11

2. Method p. 13 2.1 The data p. 13 2.2 Analyses p. 16 3. Results p. 17 3.1 Innovative output p. 17 3.2 Process innovation p. 20

3.3 Research and development p. 21

3.4 Serving new customer segments p. 22 3.5 Investments and constraints p. 24

3.6 Education and knowledge p. 27

3.7 Collaboration with other companies or institutions p. 28

4. Discussion p. 30

4.1 Outcomes of the hypotheses p. 31

4.2 Discussion of the outcomes p. 32

4.2.1 Innovative output and process innovations p. 32 4.2.2 Constraints to innovate p. 34 4.2.3 Serving new customer segments p. 35

4.2.4 R&D p. 36 4.2.5 Collaborations p. 36 4.3 Implications p. 37 4.4 Limitations p. 40 4.5 Future research p. 42 4.6 Conclusion p. 43 5. References p. 44 6. Appendix p. 48

4 6.1 Innovative output p. 48 6.2 Process innovations p. 51 6.3 R&D p. 55 6.4 Investments p. 56 6.5 Education p. 58 6.6 Collaborations p. 59

6.7 Serving new customer segments p. 63

6.8 Constraints p. 65

5

1. Introduction

Since the eighties a scientific discussion arose about the innovativeness of different firm sizes within an industry. The big unknown in this discussion is who has the innovative advantage; smaller or larger firms? Almost thirty years later there is still no scientific consensus on this topic. For this reason, the main objective of this research is to examine if there is a significant difference in the innovativeness of different firm sizes situated in the Netherlands. Where “innovative output” is seen as innovations developed by a firm, whether they are a product, process or service innovation. Two of the pioneers in this scientific field tossed the term “innovation rate” (IR), which is the amount of innovations per thousand employees. These researchers found a significantly higher IR in small companies (<500 employees) than in large companies (>500 employees), which makes small firms more innovative (Acs & Audretsch, 1987). Tether (1998) disputes this statement by claiming that large firms produce more valuable innovations than small firms, which makes large

companies more innovative. However, Acs and Audretsch (1988, p. 681) counter-argue this by stating that there doesn’t seem to be a great difference in the “quality and significance of the innovations between large and small firms”.

A great deal of literature claims that larger firms are more innovative than small firms. It is, for instance, said that smaller firms are less innovative than larger ones due to the entry barriers created by accumulated knowledge by the larger firms (Love & Roper, 2001). It is also indicated that larger companies innovate more than smaller companies, however this effect is only present up till 1800 employees. When a company has more employees this relation becomes negative (Love and Roper, 1999). Damanpour and Evan (1984) claim that larger firms have broader and more diversified resources and capacities. Secondly, are they equipped with a better technical knowhow, which enables larger firms to innovate more than smaller firms (Nord & Tucker, 1987). As larger firms are better able to handle losses when innovations fail to meet their expectations, they are capable to take on higher risks than smaller firms, which gives them more opportunities to innovate

(Damanpour, 1992).

Likewise, have many articles been written about findings where small firms are more innovative than large firms. Small and medium-sized enterprises (SMEs) are said to be more

6

flexible than large firms, which makes them more able to adapt and improve, and they are more easily accepting and willing to implement changes (Damanpour, 1996). A disadvantage for larger firms is the fact that they usually have a more formalized structure and a higher degree of bureaucracy. This can lead to a negative effect on the innovative culture, as decreases in the commitment towards innovation can be found (Hitt et al., 1990). Scherer and Ross (1990), therefore claim that due to an excessive bureaucratic control or a decrease in management control, larger firms will have a lower R&D efficiency.

Lastly, there are studies which suggest that different firm sizes do not differ

significantly in their innovativeness. The research by Yin & Zuscovitch (1998), for instance, remains inconclusive regarding this question. They state that large and small firms have different innovation incentives, which makes large firms better process- and small firms better product innovators. It is also said that large and small firms are more innovative than firms that have an intermediate size (Bertschek & Entorf, 1996), which indicates that only medium-sized firms differ from the other firm sizes. These researchers explain this with the fact that medium sized firms are more established than small firms and are therefore less inclined to innovate. On the other hand, do medium sized firms have less capacity to innovate than large firms.

It is clear that this field of research is quite fragmented and many different opinions exist about which firm size is most innovative. For this reason have Camisón-Zornoza et al., (2004) conducted a meta analysis from 53 empirical studies where the researchers tried to estimate the magnitude of the average effect of the relationship between firm size and innovativeness. They found a significant and positive correlation between size and innovation. Therefore, according to this meta analysis, does an increasing firm size goes paired with an increase in innovativeness as well. However was the effect size .15, which is quite low. The current research expects that innovativeness increases with an increasing firm size. For this reason is the article of Camisón-Zornoza et al., (2004) used to build our current research on. Due to the lack of consensus in this field is the relationship between innovativeness and firm size investigated in the Netherlands. The following research question is tried to answer: Do micro, small and medium-sized firms in the Netherlands differ from each other in their product-, service- and process innovativeness?

7

1.2 Theoretic background on differences in innovativeness in different firm sizes

To investigate the differences in innovativeness in different firm sizes, is it important to discuss and define the construct of innovation. An entrepreneurs job is to find profitable opportunities to upset any equilibrium, whether it is in equilibrium or disequilibrium (Baumol, 2005). This means that an entrepreneur wants to obtain a supernormal return by creating a competitive advantage (Barney, 1991) in a market that used to be in perfect competition. Upsetting such an equilibrium is done with innovations and results in a

temporary monopoly position (Schumpeter, 1911). Schumpeter called the constant process of innovations that replace the old ones “creative destruction”, and said that to innovate means to create novel outputs, a new good or quality of a good, a new method of

production, a new market, a new source of supply or a new organizational structure, which in other words is doing things differently (Schumpeter, 1911). Due to the fact that it is virtually impossible to do things exactly the same, is any change an innovation by this definition (Hansen & Wakonen, 1997, p. 350). For this reason have Crossan and Apaydin (2010)implemented a new comprehensive definition of innovation that captures more important aspects of innovation. “Innovation is the production or adoption, assimilation, and exploitation of a value-added novelty in economic and social spheres; renewal and

enlargement of products, services, and markets; development of new methods of production; and establishment of new management systems. It is both a process and an outcome” (Crossan & Apaydin, 2010, p. 1155). This definition includes not only the

internally conceived but also the externally adopted innovations, it claims that innovations are more than a creative process and draws attention to the two roles of innovation as a process and an outcome.

The innovation as a process answers the question “how” an innovation is produced and the innovation as an outcome answers the question “what” the innovation is. According to these researchers are “innovation as an outcome” and “innovation as a process” not equally important. The role of innovation as an outcome is “necessary and sufficient” for a successful innovation, however the role of innovation as a process is “necessary but not sufficient” for a successful innovation (Crossan & Apaydin, 2010, p. 1169). For this reason is innovation as an outcome very often the dependent variable in academic research towards

8

innovation, and will it be the dependent variable in the current paper as well. To keep the span of this thesis within boundaries will the current research involve three different types of innovations; 1) service innovation, which is “a new service experience or solution that consists of one or several of the following dimensions: new service concept, new customer interaction, new value system/business partners, new revenue model, new organizational or technological service delivery system” (Den Hertog, van der Aa & de Jong, 2010, p . 494.), 2) product innovation, which is the introduction of a meaningful new and or modified product (Wang & Ahmed, 2004), and 3) process innovation, which is the implementation of a new or modified production or delivery method, which can be used to improve production and management methods (Wang & Ahmed, 2004).

Research has indicated that innovative firms grow faster, experience higher profits and are more productive than less innovative firms (Geroski & Machin, 1993; Roper & Hewit-Dundas, 1998). For this reason, it is desirable to be innovative as a firm, however this is not as simple as it seems. Innovations are not produced by a single element in a company, industry or market, which means that “innovation as a process” includes many variables that influence this process. For this reason it is difficult to determine what variable is responsible for a specific innovation. Many studies are trying to explain what factors are responsible for innovations in different kinds of industries. One of the most studied variables that positively influences innovativeness is R&D spending (Shefer & Frenkel, 2005; Acs & Audretsch, 1988; Wagner & Hansen, 2005; Hewit-Dundas, 2006). If no money is spend on R&D it is highly unlikely that a company is going to innovate within their industry. However, when a

company has a budget to spent on R&D, the chance on innovating increases. The innovative output however increases with a “less than proportional rate” (Acs & Audretsch, 1988, p. 679; Cohen & Klepper, 1996, p. 946). This means that R&D spending influences the innovative output, but the amount spent does not directly reflect the amount of innovations.

It is commonly known that larger firms are able to invest more in R&D than small and medium sized firms. Research has indicated four general patterns regarding this R&D

spending; 1) the probability of conducting R&D grows with the firms size, 2) R&D and firm size are highly positively related in industries, 3) the R&D grows proportionally with firm size in almost all industries, 4) the amount of innovations and patents produced per dollar of

9

R&D spending declines with the size of the firm (Cohen & Klepper, 1996). These researchers explain this with a theory named “cost spreading” (Cohen & Klepper, 1996, p.926). This theory indicates that when a firm is large, it has a bigger R&D output over which it can spread the costs of R&D. So not only has the larger firm a bigger output from R&D than smaller firms it is also more likely that this R&D will realize a greater return. For this reason have larger companies a higher incentive to spend money on R&D than smaller firms. This however, only implicates that large firms have an innovative advantage towards small firms. It does not mean that they will produce more or better innovations. Small firms might profit from cost spreading as well when they collaborate in R&D with large firms.

A second variable that is intertwined with innovation is a highly skilled labour force that is capable of dealing with technological problems (Shefer & Frenkel, 2005; Acs & Audretsch, 1988; Hewitt-Dundas, 2006; Karlsson & Olsson, 1998). It is said that knowledge inputs are linked to innovative output (Audretsch & Feldman, 2004). Other research states that humans differ in their abilities to be productive, and even more importantly, how they interpret problems and apply their cognitive abilities to solve these problems. These researchers even consider this to be the source of the relationship between the

heterogeneity of the work force and innovation or productivity (Alesina & La Ferrara, 2005). This means that a highly educated and diverse labour force can contribute to the

innovativeness of a firm. When the R&D department does not have the appropriate, highly skilled employees, it is unlikely that innovative output will originate from this activity. For this reason it is important to keep searching for qualified personnel and offering current personnel extra education and schooling.

Another issue that has to be taken into account is the fact that many companies collaborate with other companies by interchanging technologies and information, which can lead to firm inter-dependencies (Smith, Dickson & Smith, 1991). Collaborating companies might be able to create innovative output that, because of a lack of resources, could not have been created independently (Cohen & Klepper, 1996). Research has shown that small companies are more flexible and autonomous than their bureaucratic counterparts, which makes them more able to accept and execute change (Damanpour, 1992; Rothwell &

Zegeveld, 1982). However, smaller firms experience more constraints to innovate than larger firms. Examples of these constraints are; raising capital, incapability of attracting sufficient

10

manpower and/or skilled labour, a lack of market intelligence, and the marketing ability to launch the products (Smith et al., 1991. p. 460). Next to that is the growth of small firms frequently constrained by the amount of internal finance (Carpenter & Petersen, 2002), which is more frequently present in large firms. Small companies might need the expertise, finances and technological recourses of larger companies to develop and exploit their own innovations. Other research has indicated that small firms experience constraints in finances, legislations, networking and marketing strategies to innovate, however larger firms

experience constraints from the high risk of development and a lack of internal expertise (Hewitt-Dundas, 2006). Research like this, questions if it is possible that not the size, but the organizational structure, which creates less constraints, foster the innovative environment (Stock et al., 2002). It is important to keep in mind that smaller firms have constraints that larger firms don’t have to deal with, and vice versa.

As markets are dynamic and customers change behavior, it is important for the success of firms to detect these changes, interpret and react to them (Böttcher et al., 2009). These researchers developed a technique to identify different customer segments and are able to analyze their developments. Customer segmentation is the act of grouping

customers in homogenous groups based on common attributes. When the market changes, firms have to react to this by adjusting their product or service to the new demands or by serving new customer segments. This can be an already existing segment or a newly

originated one, as the changing market can induce segmentation instability (Blocker & Flint, 2007). It is said that one way to innovate is to “discover unmet customer needs or identify underserved customer segments” (Sawhney, et al., 2006, p. 31). By serving these new

customer segments will the companies be challenged to innovate and offer a new product or service.

Research has indicated that large and small firms have different innovation incentives (Yin & Zuscovitch, 1998). These researchers claim that large firms tend to invest more in process innovations due to their dominance in the current market. However, smaller firms invest more in product innovations, because they cannot invest as much in process

innovations as large firms can. For this reason it is hard to compete in process

innovativeness with larger firms and is it more profitable to invest in new products. This result has been found in other research as well. Wagner & Hansen (2005) state that the size

11

of the firm is not of influence on the innovation rate, but on the type of innovation. They concluded that large firms produce more process innovations, which consist of improved processing or manufacturing. Smaller firms have an advantage in product and business systems innovations. These are “new product development or improvement” and “new and/or improved business and marketing practices”, respectively (Wagner & Hansen, 2005, p. 838). Small firms have to allocate their attention to product and/or business systems innovations, because these require less investment. Large firms however can gain a competitive advantage by investing in process innovations. This shows that the firm size influences the type of innovation pursued.

1.3 The current study

It is clear that innovations do not originate on their own and that innovations can be seen as a process, which means that many variables are responsible for innovations to be created. In the current study is the innovativeness of different firm sizes in the Netherlands

investigated. Innovations, which can be a product, process or service innovation, have been indicated by the owner or director/manager of the firm. This “innovation as an outcome” was measured in two ways. A first question was asked whether the company had produced a new product or service in the last six months, which is the dependent variable “innovative output”. Unfortunately were service and product innovations assessed in the same question. For this reason are we not able to discuss the separate constructs. The second question informed us whether the company had implemented new or better company processes in the last six months, which is the second dependent variable “process innovation”. These innovativeness are expected to be influenced by many predictors that these individual companies do, or do not have.

It is expected that with an increasing firm size the dependent variables “innovative output” and “process innovations” increase as well, which is in line with the findings of the meta-analyses of Camisón-Zornoza et al., (2004), and other research which found that with an increasing firm size the innovativeness increases as well (Love & Roper, 2001; Love & Roper, 1999; Damanpour & Evan, 1984; Nord & Tucker, 1987; Damanpour, 1992). This expected relationship would refute previous findings, which claim that small firms are more

12

innovative than larger firms or have an innovative advantage (Acs & Audretsch, 1987;

Damanpour, 1996; Hitt et al., 1990; Scherer & Ross, 1990), and that large and small firms are more innovative than firms with an intermediate size (Bertschek and Entorf, 1996). Secondly, is it expected that this relationship is also found in the different types of innovations. An increasing firm size is expected to go paired with an increased probability to have created a process-, service-, and or product innovation, which would refute the findings of Yin & Zuscovitch (1998), which claim that large firms are better process- and small firms better product innovators.

Predictor variables that are believed to influence the innovativeness of firms are also investigated. Firstly is it expected that with an increasing firm size the probability to have an active R&D department increases as well, which is in line with Cohen and Klepper (1996). Companies that have an active R&D department are expected to create more innovations than companies that do not have this. Secondly, is it expected that the larger the firm, the more difficulties exist in finding qualified personnel, which is in line with Hewitt-Dundas (2006) who found that larger firms have a lack of internal expertise. It is believed that the education level is of influence on the innovative output, which is in line with previous studies (Shefer & Frenkel, 2005; Acs & Audretsch, 1988; Hewitt-Dundas, 2006; Karlsson & Olsson, 1998). More importantly it is expected that medium sized companies are better able to appropriate the benefits of the knowledge of their employees than small and micro companies. It is an organizational capability to integrate individual specialized knowledge (Grant, 1996). Smaller firms do not have the finances (Smith et al., 1991) to acquire enough high educated employees and larger firms are too bureaucratic to appropriate this

specialized knowledge (Damanpour, 1992; Rothwell & Zegeveld, 1982). As larger firms demand an even higher educated and skilled employee than smaller firms, will the search for the right personnel be more difficult. For this reason is it expected that there is a positive relationship between firm size and highly educated personnel.

To support the innovative process, finances must be in order and investments in R&D must be made. Smaller companies, for instance, have more trouble raising capital than larger companies (Smith et al., 1991. p. 460), which is expected to be found in the current data as well. Cohen & Klepper (1996) said that collaborating companies might be able to create innovative output that, because of a lack of resources, could not have been created

13

independently. Due to the fact that smaller companies have less resources than larger companies, and are for this reason more inclined to collaborate, is it expected that there is a negative relationship between firm size and collaborations. Lastly, it is expected that firms who are able to serve new customer segments are more likely to innovate than firms who do not serve these segments, as this is one of the 12 dimensions to innovate according to

Sawhney, et al., (2006). With the current study is tried to answer the following research question: Do micro, small and medium-sized firms in the Netherlands differ from each other in their product-, service- and process innovativeness? Next to that are the following eight hypotheses being investigated.

H1: There is no significant difference in innovative output between small and micro companies.

H2: Medium-sized companies have a significantly higher innovative output than small and micro companies.

H3: Small and micro companies do not differ in their process innovativeness.

H4: Medium-sized companies produce more process innovations than small and micro companies.

H5: Medium-sized firms have a higher probability to serve new customer segments than small and micro firms.

H6: Medium sized companies experience less financial constraints to innovate than small and micro companies.

H7: There is a positive relationship between firm size and highly educated personnel. H8: There is a negative relationship between firm size and active collaborations.

2. Method

2.1 The Data

The data used in the current research was obtained at the Economisch Instituut voor het midden- en klein bedrijf (EIM), which is managed and maintained by a company called

14

Panteia1. Some of their own conclusions and results from these data are publically available on the internet2. Their data is of great quality because multiple external parties control the data collection and research, which resulted in an ISO 9001 certificate3 (a certificate that warrants the quality management in an organization). The dataset provided by Panteia is called the “MKBBeleidspanel” and was conducted via a survey throughout the years 2005 until 2011. These surveys were conducted through telephone conversations with the owner or director/manager of the company. This resulted in a total dataset comprised of 12

subsets, in which the year 2005, 2008 and 2009 have two, three and three sets respectively. Due to the fact that the EIM started to conduct the MKBBeleidspanel in 2005 is the amount of respondents in 2005 and 2006 not as high as later years. However, these samples are large enough to create reliable results.

The surveys consist of open and multiple choice questions and were assessed at more than 2000 companies per year. These companies were randomly selected from all SME’s in the Netherlands, which makes it a good reflection of the industry. Many companies are represented in multiple years, which makes the data longitudinal, however a part of the companies are only present in some, or even one year. For this reason is every year analyzed individually. Two example questions are “How many persons - including yourself- are

currently working for your company” and “Has your company introduced new products or services to the market in the last six months”? It is clear that the first question is an open (numerical) question and the second a multiple choice (categorical) question. The second could be answered with four possible answers 1) Yes, 2) No, 3) I don’t know, and 4) I don’t want to tell. These surveys remain largely the same over the years, however some questions were deleted or added throughout the years.

As the goal of the current research is to investigate the differences in innovativeness of different firm sizes, have the firm sizes been recoded into different variables. These new codes satisfy the standards of the European Commission of Enterprise and Industry (ECEI) (see Table 1). This means that we have computed four groups; micro, small, medium-sized, and large companies depending on the amount of employees. Other research computed

1

More information about this company can be found on their website www.panteia.nl

2 _{Some scientific articles and results which are based on their data are freely accessible on the internet}

www.ondernemerschap.nl

3

With an ISO 9001 certificate companies are able to provide prove of quality management that satisfy international standards. http://www.lrqa.nl/normen/4305-iso9001kwaliteitsmanagement.aspx

15

their groups differently, which may be the reason that they found different results. Kim (1980) computed four groups to distinguish between firm sizes; a group with less than 100 employees, 100 up till 499 employees, 500 up till 1000 employees and more than 1000 employees. Acs & Audretsch (1987) only used two firm sizes; the firm was considered to be large when it had more than 500 employees, however it was considered to be small when it had less than 500 employees.And other research used the number of employees without grouping them as an independent variable (Blau & McKinley, 1979; Glisson & Martin, 1980; Ettlie, 1998; Sengupta, 1998). Using an international standard for firm size is the only way to accurately compare results from different researches.

The current dataset unfortunately contained too few large companies which affected the results too much (see 2.2 Analyses for a detailed explanation). For this reason only companies with less than 250 employees were used in the analysis, which means that micro, small and medium-sized firms will be compared. Secondly, were the responses “I do not know” and “I do not want to tell” reported as missing values via Data > Define variable properties, because these options were seldom answered. These doubters affected the results too much. For this reason only companies with the answers “yes” and “no” were used in the analysis. Two surveys were conducted in 2005, however only one had the amount of employees as a variable in it. To be sure no data was lost, the datasets were merged via Data > Merge Files. These datasets were merged on the variable

“respondentnummer”, as the key variable must be the same in both datasets and have the exact same name.

Table 1

Firm sizes determined by the amount of employees, turnover and balance sheet total

Firm size Employees Turnover Balance sheet total Large >250 > € 50 million > € 43 million Medium-sized <250 ≤ € 50 million ≤ € 43 million Small <50 ≤ € 10 million ≤ € 10 million Micro <10 ≤ € 2 million ≤ € 2 million Note: These firms sizes are determined by the European Commission of Enterprise and Industry (ECEI)

16

2.2 Analyses

In the current research, groups were computed to distinguish between firm sizes. This means that both the dependent (innovative output and process innovations) and

independent (firm size, serving new customer segments, offering additional education to employees, R&D, investments, collaborations and constraints) variables are categorical. When analyzing two categorical variables, the Pearson Chi-Square (X²) test using cross

tabulations is most instructive. A X²-test measures whether there is a significant difference in the effect of a categorical independent variable on a categorical dependent variable. A significant p-value of α = .05 in such a test means that at least one group differs significantly from another group in regard of the dependent variable. That is, it tells if the two

categorical variables in the cross tabulation are independent of one another or correlated to each other. Since the test only tells that “at least one group differs from another group” the results can only be interpreted as and inferences about causal relationships must be done very cautiously. This test is also able to tell if the observed value in each cell differs from their expected value. A large deviation from one of the observed values to the expected value increases the X²-value and enables you to reject the null hypothesis. This deviation from the expected value is indicated by the standard residual (R) for that specific cell. When the R is higher than two, it has a large effect on the statistical significance of the X²-test and consequently on the correlation between the categorical variables.

The first assumption that must be satisfied to execute a X²-test is that all observations are independent. As the survey from one company has no influence on a second companies survey outcome, this assumption is satisfied. Secondly, must all row and column variable categories be mutually exclusive and include all observations. Since there is no overlap in categories with categorical variables this assumption is also met. The third and last assumption is the large expected frequencies. This means that the values of the

expected values must be ≥ 5 and all individual cells must be at least 1, except for 2*2 Tables, where all values must be > 5 (Moore & McCabe, 2008, p. 414). If this assumptions is not satisfied, these cells will affect the test results disproportionately and nothing can be said about the significance of the test. Due to the fact that large companies (>250 employees) and the answers “I do not know” and “I do not want to tell” were frequently responsible for

17

not meeting this assumption, were these respondents not taken into account for these specific analyses. If there are still X²-test that do not meet this assumption, it is however still possible to say something about the general pattern in the data. When the results over the years all show the same direction but lack statistical significance due to low scores in individual cells, inferences about relationships can only be made with great care.

As these results only have an interpretative character, correlation (R) matrices will be computed to find support for the results of Camisón-Zornoza et al., (2004). These expected results would mean that we find a positive relationship between firm size and the two dependent variables. Secondly, with these correlations it is tried to find the best predictors for innovative output and process innovations. These predictor variables, which are

discussed in paragraph 1.3, are taken from the surveys as well. Due to the extensive size of the dataset and a limited time span, only two correlation matrices per year will be discussed, one for both dependent variables. In other words only the first dataset of every year will be investigated.

3. Results

This section is divided into seven parts, in which results of the different hypotheses are being discussed. Firstly, will the results of the dependent variables “innovative output” and

“process innovations” be discussed. These results are compared in regard of the different firm sizes. Secondly, will the predictive power of the independent variables be discussed and are the firm sizes compared in regard of their deployment of these variables.

3.1 Innovative output

The results indicate that at least one firm size differs in their innovativeness from the other firm sizes, which can be inferred due to the fact that all X²-tests show a significant p-value at α =.05 level (see Table 8 and 11-18). When looking at Figure 1, we see a pattern where micro firms continuously indicate a lower probability to innovate than small and medium-sized firms. These results reject hypothesis 1, because micro firms seem to differ in their

18

question if small and-medium sized firms differ from each other as well. Medium-sized firms only indicated a lower probability to innovate than small firms in the third dataset of 2008, which means that most of the time they have indicated a higher probability to innovate. However due to the small differences in probabilities to innovate, is it hard to make inferences about a significant difference. For this reason nothing can be said about

hypotheses 2.The medium-sized group is the only firm size that shows a positive trend over the years. This would mean that they have been innovating more than previous years, which is not found in micro and small firms. However, this positive trend is most certainly found due to the fact that the indicated probability of innovative output for medium-sized firms seems to be an outlier.

Next to that, there seems to be a relationship between the “continuous renewal as company strategy” and firm size, X² (2, N=724) = 62.54, p <.05 (see Table 9). Even though that micro firms have indicated a 50,7% probability to have continuous renewal as the company

strategy, do the other firm sizes show a significant higher probability of 74,6 and 86%. As the micro, small and medium-sized group show a standard residual of -3,1, 2,2 and 3,0

respectively, a positive relationship between firm size and renewal as a strategy was found. On the other hand, all firm sizes show a very high probability to have a strategy to offer excellent service to customers (see Table 10), which is probably indicated due to the socially

0 10 20 30 40 50 60 70 05 08 08 08 09 09 09 10 11 Per ce n tage re sp o n d e d " ye s" Year

Figure 1: Innovative output

Micro Small Medium sized Linear (Micro) Linear (Small) Linear (Medium sized) Firm size

19

desirability of this answer. These different strategies do however agree with the results of Yin and Zuscovitch (1998) and Wagner & Hansen (2005), who claim that firm size influences the strategy and type of innovation pursued.

From these tests results it is clear that micro firms differ from small and medium-sized firms in regard to their innovative output. This general pattern is present in all seven years and was also found in the correlation matrices (see Table 2). A positive correlation was found between firm size and innovative output in every year this question was asked. These correlations range from .10 to .17, which is almost the same as the correlation of .15 that Camisón -Zornoza et al., (2004) found. These correlations are quite low, however they do show a consistency in the relationship, which means that with an increasing firm size the probability to have an innovative output increases as well. However due to the fact that the differences between small and medium-sized firms are so small, we have to reject

hypothesis 2. Medium-sized firms do not seem to innovate more than small firms.

Table 2

Correlations between Firm size and Innovative output

Year 2005 2007 2008 2009 2010 2011

R .17 .13 .12 .10 .16 .12

The current research shows some contradicting results compared to previous work. Yin and Zuscovitch (1998) claimed that smaller firms produce more product innovations and larger firms more process innovations. The current research, however, indicates that there is a positive relationship between firm size and innovative output, which is contradicting to the results of Yin and Zuscovitch (1998). Secondly, have Bertschek and Entorf (1996) found that small and large companies are more innovative than medium-sized companies. As there were no large firms in the current analysis are we not able to disprove that they are more innovative than the medium-sized firms. However, it was found that small and medium sized firms innovative equally as much, which contradicts their results.

20

3.2. Process innovation

A general pattern regarding the process innovations was found as well. As all X²-tests show a significant p-value at an α =.05, we can infer that at least one firm size differs from the others in their ability to produce process innovations (see Table 19-28). When taking a look at Figure 2, a pattern where micro firms continuously indicate a lower probability to have created a process innovation than the other firm sizes is found. These results reject hypothesis 3, because it seems to be that small firms have produced more process innovations than micro firms. The medium-sized firms consistently indicated a higher

probability to have created a process innovation than small firms. However, due to the small differences in probabilities, are these X²-tests not able to give any evidence regarding the question if small and-medium sized firms differ from each other as well. This pattern where an increasing firm size is positively correlated to process innovations was found in the correlation matrix as well (see Table 3). These correlations range from .20 to .33, which means that the chance to create a process innovation is weak to moderately correlated to firm size.

Table 3

Correlations between Firm size and Process innovations

Year 2005 2007 2008 2009 2010 2011

21

These results are in line with the results of Yin & Zuscovitch (1998). They claimed that large firms are better process innovators than small firms. The positive correlations between firm size and process innovations show that larger firms are better process innovators than smaller firms. This also refute the findings of Bertschek and Entorf (1996), who said that large and small firms are more innovative than firms that have an intermediate size. As the current research found that the probability to produce a process innovation increases with firm size have we found contradictory results from Bertschek and Entorf (1996). Hypothesis 3 is rejected due to the fact that small firms produce more process innovations than micro firms. As all groups show a negative trend, which means that they are producing less process innovations than previously, and the differences in small and medium sized firms are too small, we have to reject hypothesis 4. This means that small and medium sized firms produce equally as much process innovations.

3.3. Research and development

R&D is one of the most investigated predictors of innovative output. If no money is spent on R&D it is highly unlikely that a company is going to innovate within their industry. However, when a company has a budget to spend on R&D, the chance on innovating increases. It is commonly known that larger firms are able to invest more in R&D than small and medium

0 10 20 30 40 50 60 70 80 90 100 05 07 08 '08 08 09 09 09 10 11 Per ce n tage re sp o n d e d " ye s" Year

Figure 2: Process innovations

Micro Small Medium sized Linear (Micro) Linear (Small) Linear (Medium sized) Firm size

22

sized firms. In the current study were differences found in the probability to conduct R&D, X² (2, N=740) = 38.09, p <.05 (see Table 29). A standard residual of -3.0, 1.3 and 4.0 were given for the micro, small and medium-sized firms, respectively. When taking a look at the

residuals it seems that the probability that a firm applies R&D increases with firm size. Positive correlations of R = .23 and R = .16 were found between firm size and R&D activity. These results are consistent with the results of Cohen and Klepper (1996), who found that the probability of conducting R&D grows with firms size. Next to that are small and medium-sized firms more inclined to apply R&D systematically and are micro firms applying R&D more incidental, X² (2, N=214) = 6.38, p <.05 (see Table 30), which might indicate that larger firms have a more established R&D department than smaller firms. Lastly, R&D is also positively correlated to process innovations with a range of R = .22 to R = .27 and innovative output R = .35 to R= .36, which means that it is effective to have an active R&D department. In other words, when R&D activity grows the probability to create an innovation grows as well.

3.4. Serving new customer segments

The results indicate that at least one firm size differs from the others in their ability to serve new customer segments, which can be inferred do to the fact that all X²-tests were

significant at the α =.05 level (see Table 53-60). Figure 3 shows a pattern where micro firms are less inclined to serve new customer segments than the other firm sizes. However, small and medium-sized firms only show minor differences in their ability to serve these new segments. For this reason hypothesis 5, where we expected that medium-sized firms were best able to serve new segments, is rejected. The different firm sizes recognize that with changing markets it is necessary to react with an adjusted product, service or by serving new customer segments. For this reason have 25% to 30% of micro companies, and 40% of small and medium-sized companies been serving new customer segments in the last years.

23

Sawhney, et al., (2006, p. 31) said that to discover new or unmet customer needs or identify underserved customer segments is one of the 12 ways to innovate. The current data shows that serving new customer segments is moderately to strongly correlated to innovative output and process innovations. Table 4 shows that the correlations between serving new customer segments and innovative output range from R = .31 to R = .46, which is quite moderately to strong.

Table 4

Correlations between Serving new customer segments and Innovative output Year 2008 2009 2010 2011

R .31 .36 .40 .46

However, the act of serving new customer segments is weaker correlated to process

innovations. These correlations range from R = .24 to R = .30 (see Table 5), which is weak to moderately positive. It seems that to serve new customer segments can lead to innovations, whether they are a product, process or service innovations.

0 10 20 30 40 50 60 08 08 08 09 09 09 10 11 Per ce n tage re sp o n d e d " ye s" Year

Figure 3: Serving new customer segments

Micro Small Medium sized Linear (Micro) Linear (Small) Linear (Medium sized) Firms size

24

Table 5

Correlations between Serving new customer segments and Process innovations Year 2008 2009 2010 2011

R .24 .29 .28 .30

To serve new customer segments is not correlated to firm size. These correlations range from R = .06 to R = .12 (see Table 6), which means there is no or a negligible relationship.

Table 6

Correlations between Firm size and Serving new customer segments Year 2008 2009 2010 2011

R .12 .06 .10 .07

3.5. Investments and constraints

The constraints experienced by the different firm sizes do not show many unexpected results. An average of 25% had difficulties in realizing their investments (see Table 61). As all firm sizes indicated a probability between 24% and 26% no differences were found between these firm sizes. Secondly, no difference were found in the experienced laws and legislations constraints (see Table 68 and 69), and did almost all companies experience the same

constraints from working conditions demands and environmental requirements (see Table 70 and 71).

A linear relationship between firm size and the plan to invest in new products and markets was found (see Table 33). This same pattern in indicated probabilities is seen in 2006 where an increasing firm size is paired with an increased probability to invest in the renewal or an improval of company processes (see Table 34). The X²-test results indicate that at least one firm size differs in their ability to invest over the years (see Tables 35-37), which can be inferred due to the fact that all X²-tests show a significant p-value at α =.05

25

level. The ability to invest is weakly positive correlated to firm size with correlations ranging from R = .19 to R = .27. When looking at Figure 4, a pattern is found where micro firms continuously indicate a lower probability to invest. This can be the result due to the fact that 80% of medium-sized firms are able to obtain funding from banks, versus 40% and 30% for small and micro firms, which differs significantly (see Table 65). Strangely, were no finances appropriated from family, friends and via venture capital firms (see Tables 66 and 67), which generally are the financial providers for small businesses (Ang, 1992).

These companies also indicated the expected difficulties to obtain funding (see Tables 62 and 64). Micro firms have a significantly lower ability to acquire outside capital, and see this as a constraint to innovate more often than small and medium-sized companies (see Figure 5). These results do not agree with hypothesis 6, where we expected that

medium-sized firms would experience less financial constraints to innovate then micro and small companies. It seems to be that only micro firms differ from the other firm sizes

regarding this constraint, which is in line with Smith et al., (1991, p. 460) and Hewitt-Dundas (2006). These researchers claim that smaller firms have more financial constraints than larger firms. 0 20 40 60 80 100 120 06 07 08 Per ce tage re sp o n d e d " ye s" Year

Figure 4: Invested

26

Lastly, did the data show that there is a relationship between open vacancies and firm size (see Figure 6). It seems that with an increasing firm size the amount of open vacancies increases as well, which can be inferred due to the fact that all X²-tests show a significant p-value at α =.05 level (see Tables 72, 73, 75 and 76). Secondly, does firm size seem to be related to difficulties in finding qualified personnel as well (see Table 74). This might be due to the fact that with increasing firm size, more qualified personnel is needed.

These result give mixed perspectives on this topic. It seems to be that companies up till 250 employees experience somewhat the same constraints. However, micro firms experience more difficulties in obtaining funding, and larger firms experience more difficulties in finding qualified personnel.

0 10 20 30 40 50 60 70 05 '09 Per ce n tage re sp o n d e d " ye s" Year

Figure 5: Funding as constraint

Micro Small Medium sized Linear (Micro) Linear (Small) Firm size

27

3.6. Education & Knowledge

As it is very important to companies to have highly trained and educated employees, it is

common to offer extra education to employees. The data shows that medium-sized firms have been offering education to their employees almost without exception (see Figure 7). Around 80% of small companies offer this education, while micro companies offer this 40 to 70% of the time. These X²-test all show a significant p-value at a α =.05 level, which means that at least one group differs from the others (see Tables 38-40). Offering additional education to employees is weak to strongly positively correlated to firm size with

correlations ranging from R = .25 to R = .43, which means that with increasing firm size the probability to offer additional education to employees increases as well Obviously, when companies offer this extra education to employees, results towards innovation are expected by the firms. However, extra education for employees is weak to moderately correlated to process innovations with correlations ranging from R = .20 to R = .30, and it is not or

negligibly related to innovative output with correlations ranging from R = .09 to R = .12. This means that with increasing firm size, more additional education is offered to the employees

0 10 20 30 40 50 60 70 80 90 100 07 08 09 09 Per ce n tage re sp o n d e d " ye s" Year

Figure 6: Open vacancies

Micro Small

Medium sized Linear (Micro) Linear (Small)

Linear (Medium sized)

28

and this extra education has a positive effect on process innovations but not on innovative output.

These results are not in line with Audretsch and Feldman (2004), who claim that knowledge inputs are linked to innovative output. Our results also contradict the results of Alesina and La Ferrara (2005), who found that a highly educated and diverse labour force can contribute to the innovativeness of a firm. A conclusion regarding hypothesis 7 cannot be given due to the fact that we do not know what kind of education the employees (on average) have. It is, however, clear that larger firms demand higher educated personnel than smaller firms and are willing to invest in this.

3.7. Collaboration with other companies or institutions

Cohen and Klepper (1996) already stated that collaborating companies might be able to create innovative output that, because of a lack of resources, could not have been created independently. Especially the smallest firms that lack the resources to innovate should be able to benefit from these collaborations. It seems that firm size is of influence on the intentions to collaborate. As almost 40% of micro firms say they do not collaborate to obtain knowledge they might collaborate for other reasons than small and medium-sized firms. Due to the fact that all X²-tests show a significant p-value at α =.05 level (see Tables 41, 43 and 44) we can state that at least one firm size differs from the others in their probability to

0 20 40 60 80 100 120 05 06 07 Per ce n tage re sp o n d e d " ye s" Year

Figure 7: Invest in education, knowledge and

schooling

Linear (Medium sized)

29

collaborate with other companies. Figure 8 shows a pattern where 40% of micro firms and 50 to 60% of small and medium sized firms collaborate with other companies. Surprisingly do 80% of micro, 63% of small, and 50% of medium-sized firms indicate to only collaborate with small firms (in this case <100 employees). Secondly, does 6% of micro, 9% of small and 12% of medium-sized firms indicated to only collaborate with large firms (in this case >100 employees, see Table 42). The correlations between firm size and active collaborations range from R = .08 to R = .12, which means that we found no or an negligible relationship.

Hypothesis 8 is rejected as there is clearly not a negative relationship between firm size and collaborations.

To collaborate is weakly correlated to innovative output, where the correlations range from R = .09 to R = .18, and process innovations, where the correlations range from R = .15 to R = .21. These are contradicting results from Cohen and Klepper (1996), who stated that

collaborating companies might be able to produce innovative output that could not have been created independently. The fact that firm size does not influence these results does not mean that these collaborations are useless in the eye of the companies. As a matter of fact almost all companies say that the result of short-term, and long-term collaborations at least yield something (see Tables 48 and 52). 50 to 55% of the firms claim to have produced new

0 10 20 30 40 50 60 70 08 09 10 Per ce n tage re sp o n d e d " ye s" Year

Figure 8: Collaborations

30

products and services due to short-term collaborations, versus 40 to 49% of long term collaborations (see Tables 45 and 49). Around 56% of short-term and 50% of long-term collaborations were indicated to have enabled the companies to penetrate new markets (see Tables 46 and 50). And around 40% of the firms claim to have developed new knowledge via these collaborations (see Tables 47 and 51).

4. Discussion

Many researchers have tried to give an answer to the question which firm size is most innovative and which variables or processes are most predictive of this innovativeness. Unfortunately is this field of research scattered and do many opinions about this topic exist. Camisón -Zornoza et al., (2004) conducted a meta analysis with 53 empirical studies and found a positive R = .15 between firm size and innovativeness, which suggests that innovativeness increases as firm size increases. The current research tried to give a clear picture of this field of study and tries to find support for the findings of Camisón -Zornoza et al., (2004). This was done by examining the differences in innovative output of different firm sizes situated in the Netherlands. With this knowledge is tried to answer the following research question: Do micro, small and medium-sized firms in the Netherlands differ from each other in their product-, service- and process innovativeness?

The discussion is divided in six subparts. Firstly will the results regarding the different hypotheses and their outcomes be discussed. Secondly are the implications of our results regarding former research and the current market addressed. Lastly are some limitations of the current study, suggestions for future research and a conclusion given. As there were very little companies with 250 employees or more, were these large firms reported as missing values in the dataset. For this reason were differences in innovativeness between micro, small and medium-sized firms in the Netherlands investigated. It is said that these differences are even more interesting due to the fact that 99% of the companies in the European Union (EU) are SME’s. Next to that are they responsible for more than half of the total created value in the EU (Verheyden & Goeman, 2013; European Commission, 2014). The current findings suggest a pattern where micro firms differ from the other firm sizes in innovative output, process innovations, the ability to serve new customer segments, the ability to invest, experienced financial constraints, and active collaborations.

31

4.1 Outcomes of the hypotheses

Micro firms indicated a significantly lower probability than small and medium-sized firms on the dependent variables innovative output (see Figure 1) and process innovations (see Figure 2), which means that hypotheses 1 and 3 are rejected (see Table 7 for an overview of the hypotheses). However no differences between small and medium-sized firms were found, for which we cannot reject hypotheses 2 and 4. Positive correlations were found between firm size and innovative output (R = .10 to R = .17) and firm size and process innovations (R = .20 to R = .33), which suggests an increase in innovativeness with an

increasing firm size. Secondly, do the results suggests that firm size is not of influence on the type of innovation produced, because an increasing firm size goes together with an increase in both dependent variables.

Table 7

Overview of the hypotheses

H1: There is no significant difference in innovative output between small and

micro companies.

Rejected H2: Medium-sized companies have a significantly higher innovative output

than small and micro companies.

Rejected H3: Small and micro companies do not differ in their process innovativeness. Rejected H4: Medium-sized companies produce more process innovations than small

and micro companies.

Rejected H5: Medium-sized firms have a higher probability to serve new customer

segments than small and micro firms.

Rejected H6: Medium sized companies experience less financial constraints to innovate

than small and micro companies.

Rejected H7: There is a positive relationship between firm size and highly educated

personnel.

Cannot be answered H8: There is a negative relationship between firm size and active

collaborations.

32

In the current research was “Innovation as an outcome” the dependent-, and “innovation as a process” the independent variable. From the “innovation as a process” were five variables investigated. Firstly, do the results indicate that micro firms indicated a lower probability to have served new customer segments than the other firm sizes, however small and medium-sized firms are equally likely to have served new segments. For this reason is hypothesis 5 rejected. A positive relationship between firm size and R&D activity was found. These R&D departments were also related to innovative output and process innovations, which indicates that having an R&D department seems to be productive. It was expected that medium-sized firms experienced the least financial constraints to innovate, however, only micro firms seem to experience more constraints than the other groups. For this reason hypothesis 6 was rejected. As we do not know the average educational level of all employees, we cannot accept or falsify hypothesis 7. It is however clear that there is a positive relationship between firm size and additional education offered to employees. Lastly, have we found a positive relationship between firm size and active collaborations. This was not expected due to the fact that smaller firms have less resources and might benefit most from these collaborations. For this reason hypothesis 8 was rejected as well.

4.2 Discussion of the outcomes

4.2.1 Innovative output and process innovations

The current research found a comparable relationship between firm size and innovativeness as Camisón -Zornoza et al., (2004), however this resemblance is debatable. Their meta analysis was conducted over 53 empirical studies, which were used to estimate the

magnitude of the average effect between firm size and innovativeness. The fact that these studies also contained large firms (>250 employees) is a big difference from the current study. As other research claims that the general discussion is usually between SMEs and large enterprises (Karlsson & Olsson, 1998), differs the current from previous research. There is a possibility that the relationship between firm size and innovativeness is different when large firms are added to the dataset. This could influence the correlation positively or negatively. However, due to the fact that no articles were found that mention this inverted negative pattern, it is believed that this positive correlation between firm size and

33

innovativeness spans the entire firm size range. As the R between firm size and innovative output (R = .10 to R = .17) is quite low and the R between process innovations and firm size moderately (R= .20 to R = .33), have we found some evidence towards previous research (Yin & Zuscovitch, 1998; Wagner & Hansen, 2005). These researchers claim that large and small firms have different innovation incentives and that the size of the firm is of influence on the type of innovation pursued, respectively. Our results show that with an increasing firm size more process innovations originate, however this relation is less strong towards innovative output. This does not imply that larger firms pursue more process- and smaller firms more product innovations, however it does suggests that larger firms are significantly better in producing process innovations.

The fact that the R between firm size and innovative output was quite low might be impacted by the way our groups were configured, as the process used to measure firm size exercises a significant effect on the relation between firm size and innovativeness (Camisón -Zornoza et al., 2004, p. 351). It is unclear if Camisón et al., (2004) controlled for the way firm size was computed, however it is clear that the articles used in their research differed tremendously in the method of grouping the firm sizes. A second explanation might be the fact that the dependent variable “innovative output” was configured out of two innovations. If smaller firms are more innovative on product innovations and larger firms on service innovations, would they both have answered yes, even though they have different

innovations. When a larger percentage of both groups answered yes on this question would the correlation be lower than if only one group would have answered yes. In other words, the way that the question was postulated might have influenced the data.

The fact that firm size is more correlated to process innovation than innovative output can be explained by the better financial situation of larger firms (Yin & Zuscovitch, 1998; Wagner & Hansen, 2005) and the dominance in the market (Yin and Zuscovitch, 1998). Due to the fact that smaller firms have less finances available, it is said that they are better off investing in product innovations than process innovations, as these are less capital intensive. A second explanation for the stronger relationship between firm size and process innovations is the fact that large firms have more market intelligence (Smith et al., 1991 , p. 460). With increasing market knowledge, underperforming or bad processes can be

34

to produce these process innovations.

Former research claimed that due to the fact that medium-sized firms are more established, they would be less inclined to innovate than small firms (Bertschek & Entorf, 1996). On the other hand, a lower capacity to innovate would make medium sized firms less innovative than large firms. The current analysis did not find such a pattern where medium-sized firms are the least innovative firm size. The probability to have a strategy to

continuously renew increases with firm size (Table 9), which refutes the idea that medium-sized firms are less inclined to innovate. Secondly, does the data imply that with an

increasing firm size the capacity to innovate increases as well, which is in line with the results of Bertschek & Entorf (1996). As the innovative capacity increases with firm size are larger firms more inclined to spent money on R&D than smaller firms.

4.2.2 Constraints to innovate

The different firm sizes experienced mostly the same, or equal, constraints. 25% experienced difficulties in realizing their investments and most firms had difficulties with working

condition demands, environmental regulations and other laws and legislations. Micro firms indicated to experience more financial constraints than small and medium-sized firms. They are less able to attract outside capital, and see this as a significant constraint to innovate. These results are in line with previous research, which claims that smaller firms have more financial constraints than larger firms (Smith, et al., 1991; Hewitt-Dundas, 2006). On the other hand have the medium-sized firms indicated to experience more difficulties in finding qualified personnel, and was there a positive relationship found between firm size and open vacancies. In other words, larger firms have more difficulties finding the right man for the job than smaller firms. These results are in line with Hewit-Dundas (2006), who claimed that larger firms experience constraints due to a lack of internal expertise. However they counter argue the results of Smith, et al., (1991), who said that smaller firms experience constraints from a lack of sufficient manpower and/or skilled labour. A possible explanation for these different theories from previous research might be the fact that larger companies demand an even higher educated employee than small and micro firms. The fact that we have found a positive relationship between firm size and additional education offered to employees ground this line of thinking. This could mean that the employee that is wanted by the micro

35

and small firms might not even be on the radar for the medium-sized and large firms. It is said that knowledge inputs are linked to innovative output (Audretsch & Feldman, 2004) and that a highly educated and diverse labour force can contribute to innovativeness (Alesina & La Ferrara, 2005). However the current research found a moderate R between additional education offered and process innovations (R = .20 to R = .30), but no, or a negligible R between additional education offered and innovative outcome (R = .09 to R = .12). This implicates that investing in extra education for employees pays off in regard of process innovations, however no or almost non innovative output results from this. This might be explained by the fact that process innovations are different in nature as product and service innovations (Den Hertog et al., 2010; Wang & Ahmed, 2004). A second explanation might be that it is easier to innovate on an innovation as a process than on innovation as an outcome. Refining the process towards the end product, or service, is expected to require less finances, time and skills than producing a product or service innovation, which makes it more lucrative to pursue. However, firm size is weak to strongly correlated to additional education offered to employees (R = .25 to R = .43). This implies that larger firms are more able to offer this extra education and they are appropriating

innovations from this activity.

4.2.3 Serving new customer segments

As the act of serving new customer segments is positively correlated to innovative output (R = .31 to R = .46) and process innovations (R = .24 to R = .30), can it be said that this is a good predictor for innovativeness. Due to a lack of finances was it expected that micro and small firms would enter new customer segments less frequently than medium-sized firms. However, it was found that only micro firms differ in the probability to have served new segments. An explanation for this might be that these companies serve new segments that are close to the old segments they have served. For this reason less finances are needed to make a deliberate switch. This can explain that only micro firms, who have the least finances available, differ from the other groups in regard of their ability to serve new segments. A second explanation is the simple fact that micro firms are the smallest and have less manpower to direct their attention to other segments. For this reason are they not able to make a deliberate choice to serve new segments. These groups do however, recognize that