0
‘’The influence of governmental regulations
and patenting on R&D expenditures by
European companies within the energy supply
sector‘’
D. Evers, June 2014
Abstract
This study examines the determinant factors of R&D expenditures by European firms within the SBI’35 sector (Electricity, gas, steam and air conditioning supply). It is assessed that either governmental regulations and patenting by targeted companies will have a positive effect on the firms’ R&D expenditures. To test the formulated conceptual model, 9 firms based in the European Union within the targeted sector where analysed over a 5 year period (2008-2012). The results show that the determinant patenting does have a significantly positively effect on a firms’ R&D expenditures over the years. The amount of R&D incentives and the amount of patent citations does appear to have not a significantly effect on R&D expenditures, while the control variables size and age have respectively both a significant positive effect on the R&D expenditures of firms within the European energy supply sector.
1
Master Thesis Research (MSc BA-SIM)
University of Groningen
‘’The influence of governmental
regulations and patenting on R&D
expenditures by European companies
within the energy supply sector‘’
Name: Daan Evers
Student number: S2417944
Study: MSc BA -SIM
Supervisor: Dr R.A. van der Eijk
Second supervisor: Ms I. Estrada
Word count: 14.219
2
Preface
3
Executive summary
In recent years the European Commission expressed its concern about the Energy sector. As a reaction to these concerns the European Commission proposed further liberalisation of the electricity sector on 19 September 2007 after the establishment of the Energy policy by the European Commission, following a consultation process, on 10 January 2007. Together with this reaction measures were coupled with new guidelines and incentives on state support to foster environment-friendly activities and international corporation, such as renewable energy production by companies which are within the energy supply sector.
R&D expenditures, and the factors which influence this phenomenon, can be put into a very wide concept and therefore have been analysed in many different ways. The term R&D expenditures refers to activities which are designed to increase the stock of knowledge, or to devise new applications of knowledge in a particular field of interest. The activities of R&D include three factors which are (1) basic research, (2) applied research and (3) experimental development in order to improve products or services.
Many authors in this field of research have studied the determinants of R&D expenditures and especially the effect governmental interventions. Some authors suggest that governmental regulations- and incentives, as well monetary credits, do have a positive impact on R&D expenditures of firms as a proxy. Many other authors do fail to find significant evidence that credits and governmental incentives do increase the R&D expenditures of companies. In other studies the evidence that incentives stimulate R&D expenditures is statistically significantly found. As an addition to these findings on grants other authors point out other determinants which are a proxy for R&D expenditures. These are, for example, the number of applied patents and backward citations of patents by different companies as well as the number of R&D efforts conducted by R&D employees with a Masters or a PhD degree and furthermore competition- and export rates and lastly the concentration of equity among investors. These factors should all positively influence R&D expenditures by firms.
Because of the wide range of possible determinants for R&D expenditures only some factors that influence firms R&D expenditures are examined. This study combines perspectives from the fields of governmental programs and patents to examine which factors determine R&D expenditures within the European Union. Based on the existing literature, a theoretical framework was developed that included governmental regulations, patents and backward patent citations as determinants for R&D expenditures. The empirical analysis was performed by using a sample of firms within the European Union and energy supply sector because this sector is high capital- and R&D intensive and international markets and international knowledge flows are an increasingly significant source of energy innovation.
4
showed that the amount of governmental regulations - and backward patent citations do not significantly influence R&D expenditures. Lastly, another result which should be mentioned is that the control variables size and age seemed to have a significant and positive effect on the R&D expenditures of the analysed firms.5
Table of contents
Preface ... 2 Executive summary ... 3 1. Introduction ... 7 1.1 Introduction ... 7 1.2 Research question ... 81.3 Scope and domain of research ... 8
1.4 Contribution and theoretical interest ... 9
1.5 Outline of the study ... 10
2. Literature review ... 11
2.1 Definition research and development... 11
2.2 Underlying theories ... 11 2.2.1 Governmental regulations ... 11 2.2.2 Patents ... 12 2.3 Hypotheses ... 15 2.3.1 Governmental regulations ... 15 2.3.2 Patents ... 15 2.3.3 Patent citations ... 16 2.4 Control variables ... 17 2.5 Conceptual framework ... 18 3. Research methodology ... 19
3.1 Design of the study ... 19
3.2 Data collection ... 19
3.3 Measures ... 20
3.4 Analysis ... 22
3.5 Generalizability, validity & reliability ... 24
4. Results ... 26
4.1 Means, standard deviations and correlation ... 26
4.2 Results of regression ... 27
5. Discussion ... 30
6. Conclusion, limitations & further research ... 32
6.1 Conclusion ... 32
6.1.1 Definition R&D expenditures... 32
6
6.1.3 Conclusion ... 32 6.2 Theoretical implications ... 33 6.3 Practical implications ... 34 6.4 Limitations ... 35 6.5 Further research ... 36 References ... 37 Websites ... 41 Appendices ... 43Appendix A: Key indicators ... 43
Appendix B: Line graphs, scatterplots and trend lines ... 44
7
1. Introduction
1.1 Introduction
In recent years the European Commission expressed its concern about the energy supply sector. As a reaction to these concerns the European Commission proposed further liberalisation of the electricity sector on 19 September 2007 after the establishment of the Energy policy by the European Commission, following a consultation process, on 10 January 2007 (European Commission, 2007a; 2013). On 23 January 2008, a further and deeper proposal was presented which incorporates more environmental rules. This includes a revised emissions trading scheme, greater energy efficiency and ambitious targets for renewable energy (European Commission, 2007b). These targets will have an important effect on the energy sector in this age of time. These measures were coupled with new guidelines on state support to stimulate environment-friendly activities, such as renewable energy production. In addition to these findings about state support, as an example can be shown that the Dutch government is willing to achieve that 14% of the generated energy is sustainable and this needs to be 100% in 2050 (Rijksoverheid, 2012). To achieve these new set of goals the European Commission installs relaxed rules and tax benefits are given. A governmental policy has the potential to improve (social) welfare by encouraging research and design (further in this study mentioned as R&D) investments of companies. But all too often, gains from these development policies fall short to promises (Russo, 2004). Policy-makers assume that more R&D is socially preferable to less and furthermore assume that R&D incentives do stimulate research efforts.
Many economists have responded to the earlier mentioned challenge with series of empirical contributions to analyse the measurements of the effectiveness of R&D credits. In studies which focus on credits authors sometimes fail to find significant evidence that the amount of incentives does increase the R&D expenditures of companies (Goolsbee, 1998; Billings, Glazunov & Houston, 2001). In other studies, the evidence that credits stimulate R&D expenditures is statistically significantly found (Mamuneas & Nadiri, 1995; Hall & Van Reenan, 1999; Bloom, Griffith, & Van Reenan, 2002). Whether grants stimulate R&D expenditures by firms will be an important concern for both innovation policy and theory. As an addition to the earlier mentioned findings on grants, Clausen (2007) states that grants will lead to a significant increase upon the number of R&D efforts conducted by R&D employees with a Master or a PhD degree and will lead to a higher level of idea generation while implementing R&D. According to Dietzenbacher and Los (2002) competition- and export rates do influence the R&D expenditures of firms, while Baysinger et al. (1991) state that a high concentration of equity among investors will positively affect the firms R&D expenditures. Furthermore, it is pointed out in the study of Russo (2004) that the relationships between credits on R&D, benefits and social costs are difficult to capture empirically.
8
for example in the computer, energy, petroleum, chemicals- and drug industry. Pakes and Grinliches (1980) suggest that the amounts of patents are a fairly good indicator of the development output by the R&D department of firms within these industries.The factors which influence R&D expenditures of a firm, which are mentioned by Gustavsson & Poldahl (2003), Clausen (2007), Dietzenbacher & Los (2002) and Baysinger et al. (1991), will not be analysed in this particular study. This is done because of the wide variety of data (both qualitative and quantitative), that need to be analysed to determine all factors which influence R&D expenditures, will be difficult to capture in terms of scope. While data on the variables of patents and governmental regulations is widely available. The retrieved data from the amounts of governmental regulations and obtained patents will be used while studying and analysing the different hypotheses. Additionally, a subject of particular interest is forming a strong theoretical foundation for the hypotheses which will be analysed within this study.
1.2 Research question
The objective of this study is to contribute to the already existing literature in this field of interest is outlined in the introduction section. Not much particular research has been conducted for European companies in this field of interest and the relationship between patents and R&D expenditures as a proxy are not yet empirically studied widely within the academic world. With this in mind the following research question will be addressed;
‘’In what way do governmental regulations and patenting influence the R&D expenditures of European companies which are in the sector ‘Electricity, gas, steam and air conditioning supply’’? To answer the research question several sub research questions are created. This will lead to further deepening of the subject and a more precise and valid answering of the research question. The created sub questions are formulated as follows;
How can R&D expenditures be defined?
Which factors does the literature suggest as determinants for R&D expenditures?
To what extent do regulatory incentives by the European Commission influence the firms’ R&D expenditures?
To what extent does the amount of patents influence the firms’ R&D expenditures?
1.3 Scope and domain of research
9
2008). In line with this findings the OECD (2008) states that governmental regulations may act not only as a generator of new knowledge but also increase absorptive capacity and increase the pace of adoption of new technologies by firms within this sector. Also patenting is common in the energy sector, while the energy industry constitutes for about 20% of the total patent cases according to the findings by Morgan Lewis (2014). Furthermore, data will be available through the use of quantitative datasets, for analysing the different incentives for R&D expenditures, as presented by Orbis and the European Commission (2013). Additionally, the data according the R&D expenditures will be analysed in a quantitative way as presented by Orbis (2013).1.4 Contribution and theoretical interest
It becomes clear that the amount of R&D expenditures, in comparison with the effects of patenting as proxy for R&D expenditures, are not briefly tested and analysed in the European Union (further in this study mentioned as EU) and the Netherlands while this sector is capital intensive, R&D intensive, patenting intensive, and of high political- and boundary-crossing interest due to climatic issues (CBS, 2012; Morgan Lewis, 2014; OECD, 2008; Fischer & Newell, 2008; Garrone & Grilli, 2010). These determinants are tested for Canadian companies, in the US and only briefly within ‘Organisation for Economic Co-operation and Development’ (OECD)-countries. In order to extend this research topic, and create a comparative research design in the form of a theory testing model, use is made of data from European companies within the energy supply sector. By analysing different literature on the topic it becomes clear that the effects of patens on R&D expenditures by firms are difficult to capture when it comes to empirical testing (Cohen et al. 2002; Hall et al. 1986; Stoneman, 1983).The literature gaps can be formulated as (1) the lack of knowledge about the influence of governmental regulations- and incentives on R&D expenditures of European companies within the energy supply market and (2) the amount of applied patents by firms within the dataset. The relationship between these determinants and R&D expenditures of these particular firms will be tested.
As mentioned earlier, the literature gap lies in the assumption that not much particular research is conducted with respect to this sector in the EU for these particular determinants while this sector is worth to be examined. The already existing literature on this topic, which is carried out in different countries, should lead to new insights on this topic for particular the European energy supply sector. In this study also the relationship between the amount of patents as a proxy for R&D expenditures by targeted firms will be analysed.
10
1.5 Outline of the study
11
2. Literature review
Firstly, in this section the definitions of the relevant determinants, which effect R&D spending by European firms in the energy supply sector, will be mentioned and analysed through the use of different studies addressing these subjects. After the different variables for R&D expenditures by firms are outlined next hypotheses, which are in line with these variables, are structured and these hypotheses will lead eventually to the conceptual framework that will be used within this study.
2.1 Definition research and development
According to Russo (2004) R&D refers to activities which are designed to increase the stock of knowledge, or to devise new applications of knowledge in a particular field of interest. The activities of R&D include (1) basic research which is designed to advance scientific knowledge, without regard to a specific practical implication. Furthermore, R&D is (2) applied research designed to advance knowledge within a specific practical implication in a particular field. And lastly, it is (3) designed is an experimental development to create or improve products, devices, materials or processes. Because the study is focusing on the EU the European definition of R&D will be used, which is similar to the definition of the OECD (2002). This definition of R&D constrains that R&D comprises creative work undertaken on a systematic basis in order to increase the stock of knowledge, including knowledge of the people within an organization, culture and society. This amount of knowledge will be used to devise new applications (OECD, 2002). The definition, which is mentioned in the ‘Frascati Manual’ by the OECD (2002), states also that this new retrieved knowledge will result in practical implications and solutions for different types of problems. Malecki (1997) states that development in R&D by a company could lead to new knowledge, which will foster a competitive advantage within a particular industry. This will eventually lead to a more positive balance and positive growth numbers for the company involved in the R&D process.
As earlier addressed in this study, it needs to be made clear that R&D incentives of a firm do not only depend on firm-specific characteristics. Also environmental, industrial characteristics and many more factors are affecting the firms’ R&D expenditures and organizational objectives (Gustavsson & Poldahl, 2003; Goolsbee, 1998). In the following section the determinants, which are relevant for this study, will be analysed and explained in depth.
2.2 Underlying theories
To eventually come to a comprehensive framework for the determinants R&D expenditures, several underlying theories from the relevant literature need to be addressed. These theories will form the basis of the theoretical framework and will draft the different determinants as used within this study. 2.2.1 Governmental regulations
12
Researchers will more and more follow the enactment of the credit that was estimated. But also other incentives for R&D expenditures are addressed by different authors. According to Dietzenbacher and Los (2002) competition within in industry and export rates do influence the R&D expenditures of firms. These two factors could be stimulated by the government when annual demand is promoted and competition is stimulated within different industries. Baysinger et al. (1991) state that a low percentage of outside directors, in a corporation’s board of directors, and a high concentration of equity among investors will positively affect the firms’ R&D expenditures. Clausen (2007) states that grants will lead to an significant increase upon the number of R&D FTE’s conducted by R&D workers with PhD or Master’s degree and will lead to a higher level and standard of idea generating while implementing R&D. Due to the broad variety of measurements and sequentially widening of scope these variables will not be incorporated within this study.Analysis by different authors on grants and its relationship with R&D spending showed an increase in the R&D expenditures of particular firms (McCutchen, 1993; Mamuneas & Nadiri, 1995; Hall & Van Reenan, 1999; Bloom, Griffith, & Van Reenan, 2002). In addition, R&D credits appear to make a significant contribution towards increased R&D spending among different firms (McCutchen, 1993). Additionally, Goolsbee (1998) states that in the United States government funds provide often more than half of the entire nations R&D spending. As mentioned by Diao, Roe, & Yeldan (1999) an increased R&D spending of a company could possibly lead to an increase in the companies welfare and the total economy. Diao, Roe, & Yeldan (1999) analysed a six per cent comprehensive R&D credit financed by a lump sum tax, obtaining a long-run increase in welfare of about 36% for Canadian companies.
The desired targets, as outlined by the European Commission (2007a), stress the desire that 20% of the generated energy needs to be sustainable in 2020 and this needs to be up to 95%, compared to the results of 1990, by 2050. Also governments of the different countries within the EU set up targets. For example, the Dutch government states that in 2020 14% of the energy needs to be sustainable while this needs to be 100% in 2050 (Rijksoverheid, 2012). In addition, these objectives of the European Commission towards more sustainable energy will eventually lead to a reset of improving goals and other, mentioned above factors, for firms in the direct affected sector. Regulations which are made by the government will allocate a responsibility to the firms, which are subject of these regulations, to fulfil these targets. Additionally, the capability of governments to grant credits to influence the R&D expenditures of different firms is important (Hall et al. 2009). Hall (1996) argues that governmental funded R&D will raise the return on the firms’ private R&D and in effect will lower the total costs of R&D for a specific firm. It is stated that the response of a firm on private R&D to governmental funded R&D is in the order of seven per cent. This means that for every one dollar governmental R&D funding the private spending of the firms’ R&D will raise by seven cents. Hall (1996) concludes that this can be seen as a small complementary effect of governmental R&D funding towards private R&D spending.
2.2.2 Patents
13
counted from the date that the patent was submitted by an authorized agency (OESO, 2009). For this study, which will focus on the European energy supply sector where the authorized agency is the European Commission. In line with the study of Griliches (1987) it is stated that firms who make use of a R&D department and program are for more likely to apply for patents. While only 20% of the studied firms by Griliches (1987), with zero or missing R&D, make use of patenting. This number is 90% for firms which have a R&D expenditure rate larger than 10 million US dollar.Patents can foster innovative activities, carried out by R&D business-units within in company, in different ways. Firstly, patents are capable of doing this because they make new knowledge public to other companies in the sector (Cohen et al. 2002; Li, 2008). In other words, patents diffuse information which otherwise would be kept secret by the patenting company and will spill over to other firms. In comprehension with this statement it can be expected that other innovative firms as a result of shown patents, are capable to come up with new innovations with use of a R&D department. This faster diffusion of new technologies and innovations by companies will have a positive effect on the welfare of the whole society. In this way patents will prevent an unnecessary overlap of R&D efforts by different companies in the same industry and will stimulate the different R&D departments to only seek for solutions in homo- or heterogeneous fields of research. To interpreted and react to new patents, companies and R&D departments should be capable to absorb this patent information and create new innovative activities for their own seek. Companies can enhance their absorptive capacity by investing in their own R&D activities. Therefore Fung (2005) states that the impact of knowledge spill overs is dependent on companies own investments in the R&D department. Furthermore, it is not only information about the amount of patents which is given by the formal agencies. Also the number of citations about an applied patent is given according to Hall et al. (2001). By measuring these citations it can be analysed how valuable the obtained patents are (Blazsek & Escribano, 2010).
14
Table 1: Determinants of R&D expenditures by firms, extracted from existing literature
References Type of industry Specific
drivers/incentives
Incentive R&D expenditures
Goolsbee, A. (1998) Mining, civil, industrial- and chemical engineering Governmental regulations (grants); Patents are supportive No
Billings, B. A., Glazunov, S.N., & Houston, M. (2001) Non-service related industries Governmental regulations No Mamuneas, T. P., & Nadiri, M. I. (1995) US manufacturing industries Governmental regulations Yes
Hall, B. H., & Van Reenan, J. (1999)
Country level Governmental regulations (credits)
Yes
Bloom, N., Griffith, R., & van Reenan J. (2002) Manufacturing sector in 9 OECD countries Governmental regulations (credits) Yes
Cohen, W. M., Goto, A., Nagata, A., Nelson, R. R., & Walsh, J. P. (2002)
Manufacturing sectors US and Japan
Patents Yes
Russo, B. (2004) N.A. Patents; governmental regulations
Yes
Hall, B., Griliches, Z., & Hausman, J. (1986) US manufacturing sector Patents Yes Griliches, Z. (2007) Manufacturing sector Patents Yes Li, X. (2008) Software industries Patents Yes
Blazsek, S., & Escribano, A. (2010)
High-tech and non-high-tech sectors
Patents Yes
Gustavsson G., & Poldahl, A. (2003) Swedish firm-level Governmental objectives Yes Dietzenbacher, E., & Los,
B. (2002)
32 US industries Competition; export sales Yes Clausen, T. H. (2007) 1074 Norwegian firms with a positive R&D budget Governmental grants ; PhD employees
“Far from the market subsidies”
15
2.3 Hypotheses
By creating a framework, and defining the determinants of R&D expenditures, several underlying theories from existing literature within this field of study are addressed in the previous chapter. These theories will form the basis for deciding which variables are used in this study. In order to answer the research question these underlying theories, according to the R&D expenditures of a firm, will be further operationalized and finally will lead to different hypotheses addressing the subject with the outcome to answer the research question of this study.
2.3.1 Governmental regulations
As earlier mentioned in this study, credits and incentives cause an increase in the R&D expenditures of particular firms (McCutchen, 1993; Mamuneas & Nadiri, 1995; Hall & Van Reenan, 1999; Bloom, Griffith, & Van Reenan, 2002). In addition, R&D credits appear to make a significant contribution towards increased competitive R&D spending among different firms due to R&D spill overs (McCutchen, 1993). Governmental regulations in this study could be seen as rules which are relaxed and tax benefits which will be given by the authorized agency (Energeia, 2012). A way to study this phenomenon is to see every introduction of a governmental regulation as an event. This event could also be the introduction of a grant given by the government (Hall et al. 1999). Incentives of a company toward R&D expenditures could be stimulated by the government (Levy & Terleckyj, 1983; Energeia, 2012; European Commission, 2013). One of the findings by the article of Russo (2004) is that incremental and comprehensive R&D credits given by the government produce relatively large increases in research effort and welfare by the companies affected. In comparison with the findings by Russo (2004), a time lag for one year is included on this determinant because possible grants are granted in a given year to a particular firm and accordingly will lead to R&D expenditures by firms in the next calendar year(s). Furthermore, the energy supply sector can be formulated as a capital intensive market with high R&D expenditures made by firms and the sector stresses much attention by policy-makers within governments or agencies and is therefore subject to many incentives (CBS, 2012; Garrone & Grilli, 2010). According to Garrone & Grilli (2010), for this particular sector policy-makers should point out that international markets and international knowledge flows are an increasingly significant source of energy innovation. In comprehension with these findings it will be hypothesized in this study that;
H1: The amount of governmental regulations will have a positive influence on the R&D expenditures made by a firm.
2.3.2 Patents
16
patent intensive sector while this sector constitutes about 20% of the total patent cases as presented by Morgan Lewis (2014).As already can be concluded according to earlier findings in this section, patents are not the only output of R&D. Patens can be seen as only a fraction of this R&D output. This fraction may vary possibly considerably over time and also over industry (Hall et al. 1986; Pakes & Griliches, 1980). As discussed by Stoneman (1983), patents can be strongly seen as an input to the R&D process rather than an output of it. To put this in perspective, the application of the patent occurs at an early point in the development process of a product or process and the gross part of the expenditures, which would be associated with the patent, occur after the application is made. In line with the statement made by Stoneman (1983), patents will be categorized as a proxy for R&D expenditures within this study, while using a one year time lag in relation with R&D expenditures. With these implications in this study will be hypothesized that;
H2a: The amount of patents has a positive influence on the R&D expenditures of a company. 2.3.3 Patent citations
As previously mentioned in this section, not only information about the amount of patents is given by the formal agencies, also the number of citations about an applied patent is often given (Hall et al. 2001). An advantage of measuring these citations is that the number of citations stands for how valuable the obtained patents are (Blazsek & Escribano, 2010). Earlier studies have showed that the more a patent uses citing the more valuable the new structured patent is and that and that backward citations are positively correlated with the eventually monetary value of the applied patent (Lanjouw & Schankerman, 2011; Harhoff et al., 1999).This reveals that backward citations can be seen as a barometer for determining the worth of the applied patents. Citations within a patent application could be forward and backward. Forward citations are described as the number of citations received by a given patent and backward citations are described as citations about previously issued patents (Marco, 2006). According to Sanyal & Vancauteren (2013) a patent which cites previous patents, which belong to a narrow set of technologies, will have a low originality score. Similarly to these findings, citing patents which make use of a wide range of fields would gain a higher score. The number of citations could also be used to measure the success of a spill over and, as following due to the absorption capacity of a firm as mentioned by Fung (2005), the amount of R&D expenditures made by a firm in the year after the patents application was filed. Fung (2005) also argues that the R&D expenses made by a company are commonly treated as an input measure for R&D activities and that counting of patents is a better measure for successful innovative output. While patenting can be seen as a requirement for the existence of patent citations these patent citations increase productivity and accordingly the impact of the firms own patents could therefore also be used to measure the innovative output. When interpreted these findings, it will be hypothesized in this study that;
H2b: The amount of backward citations of a patent does have a positive influence on the R&D expenditures of a company.
17
only governmental regulations have impact on the organizational objectives of a company, which influences the R&D expenditures of this company, but also other factors need to be kept in mind (Goolsbee, 1998).2.4 Control variables
To test the relative impact of the independent variables in this study three relevant control variables will be addressed. A problem where other studies have to deal with is the specification of the error term in the tested model of the size of the companies’ analysed (Hall, 1986; Griliches, 1987). These difficulties arises from two, which are sometimes related, causes; the presence of a large number of zeros in the independent variables and the large size range of the firms in the different samples within this studies. Griliches (1987) states that when there is a large range in the dataset of targeted firms in terms of size the relationship between the independent variables and R&D expenditures will be obscured. Firms of large size are more able to use internally generated funds within their own R&D process. Younger firms will be more cash constrained. But innovation of small firms appears to benefit from the presence of resources like grants and external institutions (Feldman, 1994). Furthermore, larger firms are more likely to benefit from economic advantages in terms of scale. Studies within this field of research suggest that relative large firms are more likely to be subsidized by agencies and that older firms likely will get more access to grants than relative small firms (Almus & Czarnitzki, 2003; Czarnitzki & Hussinger, 2004). In conclusion, large companies are able to invest in R&D and do make use of grants. According to Grouve et al. (2011) this will lead to a higher success rate of innovation and so the economic value of the company increases and thus the company will be able to innovate again.
18
2.5 Conceptual framework
The conceptual framework which is shown in figure 1 is based on the previously analysed literature and the resulting hypotheses as mentioned earlier in this section. As stated in the literature review three determinants may affect firms’ R&D expenditures based on governmental regulations and patenting. The control variables, which also could influence the firms’ R&D expenditures, are also shown in the conceptual model. A visual representation of the expected relationships between governmental incentives and patenting with R&D expenditures can be found in figure 1;
Figure 1: Conceptual framework
19
3. Research methodology
3.1 Design of the study
The aim of this study is to determine the influence of governmental regulations and patenting as a precursor of R&D expenditures of companies in the European energy supply sector, within the time period 2008-2012. To test the theory this paper will be structured as a theory-testing study (Hair et al. 1998). This academic interest had led to a conceptual model and hypotheses which are analysed in previous parts of this study. This analyses will be carried out through using and analysing quantitative data in a panel data format and tested through multiple statistical analyses. The data will be analysed in a quantitative way to test data in a reliable and objective way. According to Galliers (1991) a quantitative way of analysing data will lead to positivist or scientific methodology rather that interpretivist or anti-positivist methodology paradigms. In this study quantitative data will be analysed only through existing and relevant databases. This type of data collection will be most reliable and accurate. A database gives true command of relevant data and enables to retrieve, sort, analyse, summarize, and report results in different moments. A database can combine data from various files and makes data entry more efficient and accurate and makes it possible to analyse different variables. Quantitative testing allows also for the opportunity to generalize findings and to analyse how many one variable is related to another variable. Finally, this way of analysing the different hypotheses will lead to results and, according to these results, to theoretical- and practical implications for companies within the energy supply sector.
3.2 Data collection
In this study companies, which are in the European sector for energy supply, are analysed. Firstly, in this study is chosen for analysing European companies due to the availability of many data relating to these companies through official agencies and data websites like Orbis (2013) and the European Commission (2013). The companies which are within the SBI’08 code 35, (Electricity, gas, steam and air conditioning supply) as presented by UNSD (2014), and if established in an European country, will be inserted in the dataset. As earlier mentioned in this study, this sector is R&D- and capital intensive, has high political interest and thus subject of many R&D stimulation programs. Furthermore, 20% of all patents applications could be accounted to the energy sector. The use of curtain companies will lead to internal consistency within this study and these restrictions also makes this study feasible in terms of (theoretical) scope. According to Bureau van Dijk, within this study referred as Orbis (2013), 11 different companies meet all determinants and are subject of the energy supply sector with a relevant country code and will be the companies used within the dataset. Additionally, data will be available through the use of quantitative data in search for the number of governmental regulations and patents (Orbis, the European Commission and CORDIS) and also quantitative data for the dependent variable of R&D expenditures (data sets and usage of the websites Eurostat, Orbis and the European Commission). For the collection of patent data the relevant databases (1) Espacenet, which makes use of data retrieved from the European Patent Office (EOB), the (2) United States Patent and Trademark Office (USPTO) and (3) Orbis, will be used. In conclusion, different data will be used when studying the different hypotheses. Also a subject of particular interest is forming a strong theoretical foundation for the hypotheses, which are mentioned before.
20
or more variables should be tested at different stages in time according to Pelz & Andrews (1964). In this particular study we will analyse R&D data of 2008 to 2012, because in this time span the innovation project called the ’Energy policy project’ was established (European Commission, 2007a). As outlined earlier in this study, the independent variables will be analysed as a proxy for R&D expenditures. In conclusion, the data stressing the independent variables will be measured form the time period 2007 to 2011.Despite research already have been carried out in this field of research, it is rare that patents are seen as in input for R&D expenditures. In conclusion, the aim to this study is to complement already existing papers in this field of research and extend theatrical implications. This had eventually led to the conceptual framework which is earlier mentioned in this study. Another aim of this study is to enhance the statistical validity of the conceptual model due to (statistical) data analysis and to reach a better understanding and a redesign of the relationships between the factors determined in the different hypotheses. In this study this will be the relationship between the independent variables (1) governmental regulations, (2) patenting and (3) patent citations with the dependent variable R&D expenditures. The second reason is to enhance the construct validity and external validity (to extract it towards the other sub-energy sectors like ‘Gas’ and ‘related activities’) within the whole European energy supply sector of this particular subject. Also there should be made use of control variables (e.g. firm age and firm size of analysed companies).
3.3 Measures
When analysing R&D expenditures with a sample size which is meeting standards, autocorrelations and partial autocorrelations can be estimated for each year in the period of 2008-2012 according to Hall (1986). The R&D intensity of a firm is measured by R&D expenditures according to Barry (2005). Therefore, this study uses R&D investments (in Euro’s) by companies in the sustainable energy sector for the period 2008-2012 as the dependent variable in this study. The year 2008 is the year when the ‘FP7-energy calls’ are widely established and of that year could have influence on the R&D expenditures of the analysed firms.
21
To collect data stressing the amount of patents, obtained by firms within the ‘Electricity, gas, steam and air conditioning supply sector, these amounts should be measured within the time period of the years 2007-2011 because patents are categorised as a precursor to R&D expenditures. When a company applies for a patent, different information is given about the request. One kind of information given (1) is the year when the patent is granted by the authorized agency, (2) a second is the year when the company applies for a curtain patent and lastly, the (3) number of citations generated by a curtain patent (Hall et al. 2001). In this study patens will be measured when these are applied for by the company. This is done due to several reasons. Firstly it is not the case that all patents within the dataset will be granted to the company by the authorised agencies involved. Furthermore, Hall et al. (2001) are pointing out that the year the patents is applied stands closer to the year that the invention is commercialized and R&D expenditures are involved, than the year a patent is granted. This is obviously what will be studied and analysed within this study. Hall et al. (2001) conclude that, if available, applications for patents in a curtain year should be used to measure the amount of patents. If granted patents should be analysed as relevant data the problem occurs that a part of the patents applied for are not included in the analysis because they never are approved. When this way of data collection is followed lots of information could be missed. In line with the amount of patens the backward citations, which are described as the number of citations used to construct an applied patent as described by Marco (2006), will be measured from 2007 to 2011 on the given criteria to analyse whether patents citations are a precursor of R&D expenditures and whether the amount of citations foster R&D expenditures made by firms.Within this study the sector ‘Electricity, gas, steam and air conditioning supply’ will be analysed. The 'OECD Patent Statistics Manual‘ indicates that there are three ways to allocate a patent to curtain sectors (OECD,2009). The first way is called direct mapping. With direct mapping an expert is used to make a judgment to which sector the patent belongs. The second way is to use the industry code of the requesting company and to take over the industry code of the company which requested for the patent. The last way is to make use of a concordance table. This is done by using experts to propose an a priori agreement between IPC classes and industries. These relationships will later be displayed in a table. By the data selection in this study use is made of, for the allocation of patents, according to the industry code of the requesting company and accordingly the second option according to the OECD (2009). The assignment is done by the using the SBI’08 codes (the first two digits). The control variable size is measured by counting and dividing the number of salaried employees (FTE’s) as in 2012 (Orbis, 2013). In this study, an interaction effect is measured for firms which are under (0) or above (1) 20000 employees in 2013 for the companies within the database and will be used as a dummy variable within the regression model. This interaction is inferred to analyse the effect of one independent variable on the dependent variable which depends on the magnitude of another independent variable (Ai & Norton, 2003). The interaction is performed by multiplying the explanatory variable by the dummy variables for the units (size). The variable age of the firms within the database is measured by the years’ from incorporation as in the period 2008-2012, as presented in Orbis (2013). The control variable years will be measured within a nominal scale (0 = ‘2008’, 1= ‘2009’, 2= ‘2010’, 3= ‘2011’, 4= ‘2012’).
22
Table 2: Relevant measures for defining variables R&D expenditures
Category Sub category Measurement Source Measurement scale Dependent variable R&D expenditures N.A. Amount of R&D expenditures from 2008 to 2012 Orbis Ratio Independent variables Governmental regulations: Patents: (1) Governmental incentives (2) Amount of patents applied for (3) Number of patent citations Amount of governmental incentives in FP7 in 2007-2011 Amount of applied patents and backward citations in 2007-2011 FP7-Energy projects as given by CODIS Orbis, EPO and the USPTO. Ratio Ratio Ratio Control variables Year: Size: Age: N.A. N.A. N.A. Data available in 2008-2012 Number of salaried employees in 2012 Year of establishment to 2012 Orbis Orbis and website companies Orbis and website companies Nominal Ratio Ratio Interaction variables Patents*Size Citations*Size Calls*Size
N.A. N.A. N.A. Ratio
3.4 Analysis
23
companies were removed from the dataset because of outlier data.A statistical test of R-square will be used to evaluate the importance of the added variables into the model in the different stages and provide information about the quantity of which the determinants explain R&D expenditures of firms.Concerning the analysis in this study, the dependent R&D expenditures variable has been undertaking a square root-transformation. Because of the skewed distribution of data this transformation was necessary to control for the skewed distribution for the amount of R&D expenditures that are carried out by the targeted firms. After this transformation by square-rooting the R&D expenditures, the variable was normally distributed and could be entered in the analysis. As earlier mentioned in this study, an one year time lag is included in this study to analyse the effect of governmental regulations on R&D expenditures. Hall, Jaffe and Trajtenberg (2001) do indicate that, when working with patent applications, a lag should be recorded for the time span of three years. This is because of the truncation problem, which implies that it will take time before all applications are processed, approved and made known to third parties. This lag will be inapplicable in this study, because applied patents are incorporated in the dataset and not only granted patents will be analysed. Furthermore, some data on the patent variable of one company was missing within the dataset. This problem is overcome by interpolating patent data for missing data within the years 2007 and 2010 and extrapolating missing patent data in the year 2011. The process of interpolating could be described as the determination of the value of a function between two points at which these two points have prescribed values (Dyn et al. 1990). The act or process of estimating the value of a variable or function outside the observed range is called extrapolating. Within the analysis, on patents within this study, this should be kept in mind. As mentioned earlier in this study, Pakes and Griliches (1980) state that patents are a fairly good indicator, with a strong simultaneous relationship between R&D expenditures and patent applications, of the inventive output for the R&D department of the firm.
The data concerning the dependent variable does have a range from the years 2008 to 2012, while the explanatory variables, which act as a proxy for R&D expenditures, do have a range from 2007 to 2011. Considering the multiple-regression model for N observations and T time periods accordingly 45 observations will be incorporated within this dataset for 9 firms (i), with several unknown quantities or β-values. The following equation is structured;
24
The meaning of the different variables will be explained below;√Yi = dependent variable: √ R&D expenditures (in $) α₀ = constant B coefficient
χ1ᵢ = control variable: year (T) χ2ᵢ = control variable: firm size χ3ᵢ = control variable: firm age
χ4i = independent variable: amount of applied patents by targeted firms χ5ᵢ = independent variable: amount of backward citations applied patents χ6ᵢ = independent variable: amount of governmental regulations (in €) χ7ᵢ = interaction variable: dummy size*
ɛᵢ = error term
i = respondent
3.5 Generalizability, validity & reliability
The causal relationship between the capacity for innovation and firm size can be defined as a Schumpeterian hypothesis, while R&D activities of firms rise disproportionably with a more large size (Levin et al. 1985). Accordingly, larger firms will be more able to use internally generated funds within their own R&D process, while younger firms will be more cash constrained. Furthermore, larger firms are more likely to benefit from economic advantages in terms of scale. Studies within this field of research suggest that large firms are more likely to be subsidized and that older firms likely will get better access to grants than smaller firms (Almus & Czarnitzki, 2003; Czarnitzki & Hussinger, 2004). Although some studies have found evidence that age does not matter for the probability of getting access to grants (Almus & Czarnitzki, 2003). In addition to these findings, it is stated by Clausen (2007) that young firms never have been a significant positive predictor variable. Because of differences in size between targeted firms, the relationship between both determinants could be obscured.
Furthermore, if variables are measured in different ways, outcomes of the standardized coefficients show which variables have a greater relative effect on the dependent variable (Field, 2005). Since variables in this research are measured and scaled in different ways (like Dollars, Euro’s, patents amounts and numbers) it is useful to include the standardized coefficients. Additionally, within this study the standardized coefficients of the variables are displayed as shown in table 4.
While only a few companies within this sector made their R&D expenditures public, more than 5 % of the sample is not incorporated within the analyses and should constitute as a source of selection bias (Berk, 1983). An interaction variable is inferred within the analysis to measure the effect of one independent variable on the dependent variable which depends on the magnitude of another independent variable (Ai & Norton, 2003). This is done by multiplying the explanatory variable in question by the dummy variables for the units (firm size). The consequence of this easy way out is that the generalizability of this data to the population of these units is lost (Snijders, 2005).
25
within an regression analysis (Field, 2005). Multicollinearity occurs when multiple factors are correlated not only with the dependent variable, but also to each other within a regression model (Field, 2005). Balanced data, which is made use of within this dataset, is preferred over unbalanced panels with respect to multicollinearity (Jaeger, 2008). Balanced data allows an observation of the same unit in every time period (e.g. month or year), which reduces the noise introduced by unit (individual or groups) heterogeneity. To control for multicollinearity, ‘Tolerance’ and ‘VIF’ are interpreted as being collinearity diagnostics. These diagnostics can be found in table 4. To rely on the absence of multicollinearity, VIF rates should be <10 and tolerance rates should be <1. After deleting the interaction variables size*patents and size*citations, because of VIF rates higher than 10, in table 4 all diagnostics for VIF are below 10 and Tolerance levels are smaller than 1. This guarantees the absence of multicollinearity.Field (2005) states that Heteroscedasticity stands for the distribution of numbers for one variable in relation to the distribution of numbers for another variable. To test for heteroscedasticity a scatterplot is created. Figure 2 shows the standardized residuals for observations from 2008 to 2012. The figure shows an acceptable evenly divided cloud of observations and concluded can be that there is no reason to suspect heteroscedasticity.
Figure 2: Scatterplot of standardized residuals
To test for normality of residuals a normal probability-plot and histogram were created in order to provide evidence for expected normality of the error term. Figure 3 indicates that the standardized residuals follow the shape reasonable. Figure 4 reveals that the standardized residuals fit the straight line within an acceptable range.
26
4. Results
4.1 Means, standard deviations and correlation
As a starting point for the analysis the means and standard deviations of the variables within this dataset are measured. The average size of firms in the sample was 44222,67 salaried employees (FTE), with a minimum of 1205 and a maximum of 236.156 employees in 2012. The average age of firms in the sample is 53,89 years, with a minimum of 15 year and a maximum of 115 years in 2012. When looking at patenting activity, the 9 firms participating in the dataset, obtained for 161 patents and these patents made use of 554 backward citations from 2007-2011. The average amount of ‘FP7- energy calls’ within the sample in 2007-2011 was €179.3 million per year. While only a few companies within this sector made their R&D public this is a possible source of selection bias.
The correlation analysis in table 3 shows that some variables correlate significantly with the dependent variable R&D expenditures, while the amount of governmental regulations and the amount of citations do not correlate significantly with R&D expenditures. In addition, the correlation matrix shows that several independent variables correlate with each other. Firstly, the variable R&D expenditures significantly correlates with the variables amount of patents (0.484; p< 0.01), size (0.766; p< 0.01) and age (0.746; p< 0.01). Since R&D expenditures are often an outcome for these factors, as outlined by different authors, these correlations are not surprising. Also size affects R&D expenditures while it is plausible that larger firms invest more in R&D due to economies of scale. Furthermore, the correlation matrix shows that the variable age also correlates with other variables. Age correlates with the amount patents (0.299; p< 0.01) and size (0.344; p< 0.05). These correlations could be explained while older firms have possibly more in-house know-how and monetary possibilities to patent and do extensive R&D research. Lastly, a correlation between the amount patents and citations is shown (0.828; p< 0.01). This could be explained due whether often use is made of patent citations by the application of a patent.
Table 3: Means, standard deviations and correlations
Variable Means SD 1 2 3 4 5 6 1. √R&D expenditures 8369,36 4943,09 2.#governmental regulations 179300000 22261187,17 ,108 3. #patents 3,58 3,026 ,484** ,305* 4. #citations 12,31 14,168 ,273 ,372* ,828** 5. Size 44222,67 35,775 ,766** ,000 ,390** ,178 6. Age 53,89 35,775 ,746** ,037 ,299** ,126 ,344* 7. Year 2,00 1,430 ,105 ,916** ,389** ,394** ,000 ,040
27
4.2 Results of regression
For the model p<0.05 is significant and use is made of standardized coefficients. The regression analysis is performed in a hierarchical manner. The independent variables are entered in 3 stages within this study. In the first stage, the control variables are entered into the regression. In the second stage the independent variables, which relationships are examined, are entered and in the third model an interaction variable is added. Furthermore, R² (R² =0.898) indicates that the predictors, or in this study called the determinants, do highly explain the firms’ R&D expenditures. The difference in R-squared between the different hierarchical models is not large with only 4%. The inclusion of the explanatory variables into model 2 and the interaction variable into model 3 resulted in an additional 4% of the variance being explained (R² change = 0.040), which is not many.
Hypotheses 1 is related to the governmental regulations- and incentives of a firms’ R&D expenditures. The hypotheses predicts that the amount of governmental incentives (in Euro’s) positively affect a firms’ R&D expenditures. The regression model results show that the relationship between these two variables is not significant, but positive (0.064; p> 0.05). Adjusted for the influences of the amount of patents- and citations, the relationship between the amount of incentives and R&D expenditures is not significant (p= 0.654). However, by performing a scatterplot with trend line a positive slope is recognized between the rise of R&D incentives and years as showed in Appendix B (R²= 0.838).
Hypotheses 2a and 2b are related to the amount of patents and the amount of backward patent citations on the firms R&D expenditures. Hypothesis 2a predicts that the amount of patents positively affects the R&D expenditures of a firm, as a proxy. Results show that the relationship between both variables is significant (0.326; p< 0.05). Hypothesis 2b predicts that the amount of backward citations positively affect a firms R&D expenditures within this particular sector. Nevertheless, results show that the relationship between the variables amount of citations and R&D expenditures is not significant (0.011; p> 0.05). Accordingly with the determinant governmental regulations, this determinant is not significantly related with R&D expenditures (p= 0.921). Nevertheless, by performing a scatterplot, as shown in Appendix B, with trend line a positive slope arises for the amount of patent citations during the analysed time period (R²= 0.155).
28
Table 4: Regression resultsVariables Model 1ₐ Beta √R&D expenditures Model 2ₐ Model 3ₐ Beta Beta Tolerance VIF Year Size Age 0,083 0,579** 0,543** -0,048 0,550** 0,529** -0,043 0,461** 0,356** 0,138 0,350 0,257 7,231 2,859 3,887 #patents #citations #calls 0,077* 0,027 0,100 0,326* 0,011 0,064 0,131 0,243 0,146 7,605 4,111 6,819 size*#calls 0,405** 0,159 6,304 R² Adjusted R² F-test F-change Sig. F-change n ,858 ,848 82,579 82,579 0,000 41 ,866 ,845 40,911 1,387 0,529 38 ,898 ,871 34,112 ,241 0,022 35
ₐ Standardized regression coefficients are reported. *p <0.05; **p<0.01
To analyse whether the independent variables explain and predict the increase in time of R&D expenditures, a single OLS-regression is performed for all three determinants separately as in line with the study of Knüppe (2012). This regression is performed to determine whether these determinants were significantly correlated with the time factor (Hall et al. 1999).
Results show that there is only one large change has occurred for the explanatory variables. This change is measured for the difference of the amount of patents in the period 2007-2011. The results, as shown in table 5, predicts that within a time period of 5 years (2007-2011) the difference and increase for the amount of patents equals 5*0,184 = 0,92 while the amount of governmental incentives increased 0,005*5= 0,025 and the amount of citations increases by 0,040*5= 0,2.
Table 5: Correlation coefficients and explained variance for time taken independent variables
Variables Years (T=5)
B (s.e.) Sig. R²
#patents 0,184 (0,066)** 0,008 0,151 #citations 0,040 (0,014)** 0,007 0,155 #calls 0,005 (0,002)** 0,003 0,091
29
The most important determinant factor in predicting R&D expenditures, within this conceptual model, is the amount of patent applications. Furthermore, the amount of governmental incentives and the amount of patents citations showed less importance and were not significantly corralled and related with R&D expenditures. Nevertheless, a trend is watched that a rise in amount of governmental incentives will affect R&D expenditures positively. Additionally, firms’ age and size do affect the amount of R&D expenditures made by firms within the European Energy sector. In conclusion, table 6 will summarizes the key findings of the multiple-regression model used within this study.Table 6: Key findings
30
5. Discussion
The findings of this paper shed light on the determinants of R&D expenditures. It is based on a dataset of 9 firms with in total 45 (ixT=N) observations, operating in the Electricity, gas, steam and air conditioning supply industry in the EU. The aim of this paper is to analyse the influence of regulatory-, patent-regulatory-, and patent citing determinants on firms’ R&D expenditures as a proxy. Results show that for a positive and significant effect on the amount of patents on the firms’ R&D expenditures and non-significant and positive effects for the amount of governmental regulations and backward citations on R&D expenditures. Control variables that appeared to have a significant positive effect were both firm size and age in terms of full-time-employees and years from incorporation of the firm.
Firstly, the findings of this paper indicate that only the amount of patents respectively does significantly- and positively influence R&D expenditures of firms with the dataset over time. In addition with these results it needs to be stated that, as earlier outlined in this study, R&D expenditures of a firm do not only depend on firm-specific characteristics. Also environmental, industrial characteristics and many more factors are affecting the firms’ R&D expenditures and their organizational objectives (Gustavsson & Poldahl, 2003; Goolsbee, 1998).
31
findings some firms do not patent their R&D research and outcomes. Cohen et al. (2002) state that problems with demonstrating the novelty of the R&D outcome, the ease of inventing around the patent, the disclosed information of a patent application, and the costs of defending or applying for a patent withhold firms for making use of patenting.Another result from the statistical analysis within this study is that backward citations of patents cannot be seen as a direct proxy for innovative behaviour and does not have a significant influence on R&D expenditures. Also the relationship between the amount of backward citations and R&D expenditures of the targeted firms was not significant. An explanation for the findings can be that citations do measure the quality of a patent (Blazsek & Escribano, 2010). Earlier studies showed that the more a patent makes use of citations the more valuable the new patent is and that backward citations are positively correlated with the monetary value of the patent (Lanjouw & Schankerman, 2011; Harhoff et al., 1999). In contradiction with this finding, Fung (2005) states that the number of citations could be used to measure the success of a spill over and, as following due to the absorption capacity of a firm, the amount of R&D expenditures made by a firm. Furthermore, a possible shortcoming of these observations could be that these were incomplete because no use was made of the amount of forward citations. Forward citations are described as the number of citations received by a given patent as backward citations are described as citations about previously issued patents (Marco, 2006). To measure the potential value of R&D expenditures carried out by a company and to measure the amount of forward citations, in comparison with R&D expenditures, should be an addition within this field of research.
As for the control variables, firm size has a significantly positive influence (0,461, p; <0,05) on R&D expenditures. The empirical results suggest that small firm innovation appears to benefit less from the presence of external institutions and resources than large firm, which take advantage of the benefits from the presence of knowledge resources within their companies. A potential concern of using the number of employees is the assumption that firms which receive grants may hire R&D staff and thus their employment increases during the execution of the project where the grant is granted. This would cause some endogenous shortcomings among the receipt of governmental incentives and firm size according to Almus & Czarnitzki (2003). Other empirical studies suggest that small firm innovation appears to benefit from the presence of external institutions and resources (Grilicnhes, 1987). Although, large firm innovation benefits from the presence of knowledge resources where location factors appear to be especially beneficial for small firm innovative activities.
32
6. Conclusion, limitations & further research
In this section the conclusion of this study will be given by answering the formulated research question and outline theoretical- and practical implications. To put this answer in perspective the limitations of this study will also be addressed and as following recommendations for further research.
6.1 Conclusion
The goal of this research was to examine what determines the relationship between governmental regulations, patenting and R&D expenditures. This goal was translated into the research question as formulated in chapter 1;
‘’In what way do governmental regulations and patenting influence the R&D expenditures of European companies which are in the sector ‘Electricity, gas, steam and air conditioning supply’’? First, conclusions with regard to the definition and determinants’ of R&D expenditures will be defined in this section. After these factors are defined, the conclusion about the research question will be outlined according to the influence of governmental incentives, patenting and patent citations on R&D expenditures with respect to the analysed firms.
6.1.1 Definition R&D expenditures
R&D refers to activities which are designed to increase the stock of knowledge, or to devise new applications of knowledge in a particular field of interest. The activities of R&D include (1) basic research which is designed to advance scientific knowledge, without regard to a specific practical implication. Furthermore, R&D is (2) applied research designed to advance knowledge within a specific practical implication in a particular field. And lastly, R&D is (3) designed is an experimental development to create or improve products, devices, materials or processes.
6.1.2 Determinants R&D expenditures
Many authors in this field suggest that governmental incentives and credits do have a positive impact on R&D expenditures of firms. But many authors do fail to find significant evidence that credits and governmental incentives do increase the R&D expenditures of companies. In other studies the evidence that incentives stimulate R&D expenditures is statistically significantly found. Also the amount of patents seems to affect R&D expenditures and can also be seen as a proxy for this phenomenon. As an addition to the findings concerning governmental grants, other authors point out other determinants for R&D expenditures. Grants will lead to an significant increase upon the number of R&D efforts conducted by R&D employees with a Masters or a PhD degree and will lead to a higher level of idea generation while implementing R&D. Competition- and export rates do influence the R&D expenditures of firms while a high concentration of equity among investors will positively affect the firms’ R&D expenditures.
6.1.3 Conclusion
