Version classiqueVersion mobile
OpenEdition Books

The role of German universities in a system of joint knowledge generation and innovation

 | 
Mirja Meyborg

8. Empirical Results regarding the Knowledge Generation, Innovation and Collaboration Potential of the German Universities

Texte intégral

1Chapter eight starts with the empirical results of this PhD thesis, beginning with the comprehensive analysis regarding the publication, patenting and collaboration behaviour of the German universities. Further, it points to proximity patterns in times of globalisation and explores how the German universities behave in this regard.

8.1. Knowledge Generation and Collaboration

2The first analysis explores in how far the publication activity of the German universities has increased over the past ten years. Thus, the following figure simply illustrates this development in absolute numbers and simultaneously points to the extensive self-developed data set in this regard:

Figure 11: Publication Activity of all German Universities, absolute Numbers, 2000-2009 (own illustration).

3At the first glance, it is obvious that all analyses regarding knowledge generation and collaboration can make use of around 250,000 single publications where at least one German university is involved with for a time period from 2000 until 2009. Thus, there is no doubt that the German universities are highly involved in scientific papers, here with regard to the Scopus data-base. While they have been involved in around 44,000 publications within the first time period, they could already record about 80,000 publications in 2009. Thereby, the highest increase has occurred from the second to the third time period, making up 30% of the overall change. Generally, the German universities could increase their number of publications from the first to the last time period by around 70%.

4Being aware of the enormous increase of the German universities output in scientific papers, the next step consists of the presentation of the development path of their collaboration activity. Hence, the following figure indicates to what extent the German universities are already involved in close network collaborations which are pictured through co-authorships:

Figure 12: Publication and Cooperation Activity of all German Universities, absolute Numbers, 2000-2009 (own illustration).

5The figure demonstrates that within the first time period, about 23,000 publications have been published by the German universities where only one single person has been involved with. In contrast, at the same time, the German universities have published about 21,000 publications which have been co-authored. Thus, in 2000, the German universities have still published more often on their own than to go into a cooperation. Overall, as Figure 12 already indicates, on average, around 44,000 publications have been published during the first time period.

6But, while single-authorship has even declined from the first to the second time period, co-authorship has risen about 42% during that time. Further, in 2009, the German universities were overall engaged in almost 80,000 publications at all, but only about 28,000 have come from single-authorship. Thus, publication and collaboration activity continuously rose over the time period from 2000 until 2009, whereas it is very striking that single-authorship has become much less important over the past ten years. From the first to the fourth time period, single-authorship has increased by only 18%, while co-authorship has risen by 125%.

7With regard to the structure of the co-authored publications, it is also obvious that the number of bilateral and multilateral networks is almost equal within the first time period. In 2000, there are around 11,000 bilateral networks and about 10,500 networks where more than two authors have participated. But, while the share of bilateral networks has not even doubled from the first to the fourth time period, networks with more than two authors have risen by almost 170% during that time period. Last, networks with more than 10 authors can be rather neglected in this regard, even though they have increased by 250%; their overall share has just made up three per cent of all networks considered within the last time period. However, it can be concluded that not only co-authorship has become much more important during the past ten years, but also all those networks with more than two but less than ten participating authors.

8In order to underpin the finding that single-authorship has become much less important over the past ten years compared to co-authorship, a box plot is illustrated in the following, which visualizes again above findings:

Figure 13: Box plot of the Publication Activity of all German Universities, single-authorship vs. co-authorship, 1999-2010 (created by means of STATA).

9While on the left side, the distribution of single-authored publications is illustrated, the right side presents the distribution of the co-authored publications. Generally, the box plot delivers a variety of robust measures of variation and location; thus, the table on the right hand side illustrates the median, the upper and lower quartiles, the upper and lower whiskers which define what is commonly referred to as the upper inner and lower inner fence values and the outliers of both distributions for a time period from 2000 until 2009.

10Comparing both measures, it is again very obvious that the German universities are much more engaged in cooperation activities in the meantime, as the median of the co-authored distribution is 455 compared to 257 of the single-authored ones. But not only has the median indicated this finding, but also the 0.25- and 0.75-quantiles. While within the group of single-authored publications 25% lie below 92.75 and 25% above 517, both values are much higher in the case of the co-authored publications, namely 25% below 169 and 25% above 952.25. The lower whisker is zero for both distributions, but the upper whisker is for the co-authored publications twice as high as for the single-authored ones. As there are outliners in both data sets, the distance between the highest value of single-authored and co-authored publications is even larger; the highest value of co-authored publications is three times as big as the highest value of single-authored publications. Thus, it is again highly obvious that close network collaborations have much more increased as could be especially seen by the illustration of above box plot.

11Up to this point, it is well established that the publication and cooperation activity of the German universities has highly increased over the past ten years. Overall, they could increase their number of publications from the first to the last time period by around 70%. Further, the collaboration activity of the German universities continuously rose over the time period from 2000 until 2009, as single authorship has only increased by 18%, while co-authorship has risen by 125%. Thus, hypothesis la is confirmed by above findings.

12A second step now explores in how far the role of the German universities has changed from solely knowledge producers towards knowledge mediators, which, in turn, leads to an increasing importance of the German universities as a central node for knowledge transfer. Before coming to this concern, an illustration of the institutional distribution of the German university co-authors is firstly presented by the following figure:

Figure 14: Institutional Distribution of the German University Co-Authors, weighted Numbers, 2000-2009 (own illustration).

  • 48 In this context, universities include all domestic and foreign universities except of the 76 Germa (...)

13As can be seen from the figure, the German universities are either linked to other universities48, to research institutes, to themselves or to enterprises. First of all, it is obvious that all four network constellations have highly increased over the past ten years, even though one group has grown stronger than one another. The strongest network can be found between the German universities and other universities. Even though these collaborations have occupied only second place within the first time period with around 4,300 linkages, they have risen by 166% to around 11,400 linkages. The group of research institutes is also very strong; they have had 4,500 linkages to the German universities within the first time period, and could increase their network collaborations by 102% up to 9,200 linkages within the last time period. However, compared to the group of other universities, they have lost some places in this regard. Thus, knowledge networks with other universities could highly catch up with the research institutes, being nowadays even much stronger than the research institutes. It is also obvious that the German universities have cooperated quite frequently with themselves, too. Within the first period, there have been around 2,100 linkages among the German universities which have risen by 130% to around 4,900 linkages. However, the weakest group of network collaborations seems to be the group of the enterprises, as they have started quite low with only around 800 linkages. But, it is very eye-catching that they could increase their linkages to the German universities by incredible 212% up to 2,400 linkages. Thus, these network collaborations possess the highest rate of increase among all groups.

14In this context, it can be carefully concluded that network collaborations with universities are most important, but, university-industry linkages have had a more dynamic development as they could have been increased by incredible 212% up to 2,400 which has made up the highest increase among all collaboration partners. One reason for this could be that enterprises have sometimes problems to provide themselves with the scientific knowledge they need, and thus, looking for university partners themselves, too.

  • 49 In this context, distinct cooperation partners mean that double counts do not occur, i.e. if a Ger (...)

15Further, being aware of the overall dynamic development of the German university collaborations to other institutional partners in general, it is further interesting to explore how the role of the German universities as knowledge mediators has developed over time. In order to evaluate the increasing importance of the German universities as central nodes for knowledge transfer, the values of normalized degree and betweenness centrality are used. By means of normalized degree centrality, it is shown how the German universities have developed over the past ten years regarding their distinct cooperation partners49. Further, betweenness centrality illustrates how the German universities are already linked in the chains of contacts as it measures how often a German university lies on the shortest way between two other actors.

16First of all, the following table shows the mean values of normalized degree centrality of all German universities:

Table 10: Mean Value of normalized Degree Centrality of the German Universities, 2000-2009 (own illustration).

17The table shows that the German universities have highly increased their distinct cooperation partners over the past ten years, as the value has risen by around 20%. Thereby, the top three German universities in 2009 are Heidelberg (0.091701), the Technical University of Munich (0.0865566) and the LMU Munich (0.796075), while the University of Koblenz-Landau, the University of Weimar and the University of Bamberg are the top three regarding the highest increase over the past ten years. However, the following table now points to the mean value of betweenness centrality of the German universities from 2000 until 2009, thus illustrating how all German universities have developed on average in this regard:

Table 11: Mean Value of Betweenness Centrality of the German Universities, 2000-2009 (own illustration).

  • 50 43 of the German universities have had decreasing values in this regard.
  • 51 28 of the German universities have had decreasing values in this regard.

18At the first glance, it is eye-catching that the value of betweenness centrality has not increased over the past ten years, but decreased from around 0.026 to 0.02450. As the value of normalized degree centrality51 has generally developed the other way round, one may assume that while some of the German universities have gained in importance as knowledge mediators, others might have lost some places in this regard. Another reason can be given by the fact that more German universities have gained in importance regarding their number of distinct cooperation partners, thus, the network has become larger but not automatically tighter so that some of the German universities have received smaller values regarding their importance as mediator. Anyway, the top three universities in 2009 in this regard consist again of the University of Heidelberg (0.097603), the Technical University of Munich (0.086908) and the LMU Munich (0.078098), while the University of Koblenz-Landau, the University of Bamberg and the University of Weimar are the top three regarding the highest increase over the past ten years. A detailed examination of the distinct groups of German universities, pointing to significant differences in their behavioural patterns can be found in chapter ten.

19Hence, as the German universities have indeed highly increased their distinct cooperation partners over the past ten years at the expense of possible loops within the networks, hypothesis 2a is only largely confirmed.

  • 52 Test of multi-collinearity has been made; all mentioned variables do not have a strong correlation (...)

20Finally, bearing in mind that university-industry linkages have experienced the most dynamic development, the OLS regression model explores whether the increasing cooperation activity of the German universities with enterprises has been affected by third-party funds or not. The following table now illustrates the results of the model52:

Table 12: Results of the OLS-Regression Model - Number of Enterprise Interactions (own illustration).

21The table now expresses to what extent the above independent variables affect the publication activity of the German universities, especially with regard to third party funds. It shows that third-party funds and administrative income highly affect their collaboration activities with enterprises. For both variables, the result is highly significant with a standard error less than 1%. Referring to both coefficients, it is obvious that their values are particularly small. However, this finding is not too extraordinary as the size of third-party funds and administrative income is extremely high compared to the number of publications. Hence, in case of an increase of third-party funds or administrative income in the amount of one unit, the number of publications would, of course, only increase very slightly as the coefficients indicate. Reasons for the significance of third party funds can be constituted in the fact that most of those funds have come from the private economy which, in turn, has led to more collaborations and co-authored publications. Besides, the variable administrative income includes all supplied services as any income from hospital treatment, from sales of goods of any agricultural test materials or from the sale of any tangible assets. Thus, it is possible that the German universities have closely worked with enterprises concerning above supplied services which, in turn, has also led to more co-authored publications, too. Besides, it is interesting to notice that there is also a significant result for one dummy variable, namely the size-dummy with a standard error less than 10%. From this it can be concluded that the number of co-authored publications with enterprises increases with the size of the university.

8.2. Innovation and Collaboration

22The second line of thought builds upon the changing role of the German universities as more entrepreneurial institutions, using patent applications to illustrate their changing role within innovation patterns.

  • 53 The classical German university is actually known as a knowledge-intensive institution that is hig (...)

23First of all, it should be mentioned that the scope of the data set is many times smaller than the one of the publications. In total, the amount of the patent data makes up only 1% of the one of the publications53, as the following figure indicates:

Figure 15: Patenting Activity of all German Universities, absolute Numbers, 1999-2008 (own illustration).

  • 54 The decline from the first to the second period has probably been occurred due to the dot.com bubb (...)

24Figure 15 illustrates the overall patenting activity of the German universities from 1999 until 2008 as it has been done for the scientific publications. It shows that after a small decline of patent applications from the first to the second time period54, the German universities have experienced a huge increase regarding their number of filed patents. While they have filed around 281 patents from 1999 until 2000, they could already triple their number of patent applications within the last time period.

  • 55 In a few cases, the German university is applicant and inventor at the same time. Figure 16 anyway (...)

25As a patent does not only include applicants but inventors too (see chapter four), it is further interesting to illustrate how many patents exist in total where any German university has been involved with, either being applicant or inventor55:

Figure 16: Distribution of Applicants and Inventors of the German Universities, absolute Numbers, 1999-2008 (own illustration).

26It can be well observed that while the number of applicants has generally increased over the past ten years, the number of inventors has developed the opposite way. It is striking that the German universities have started quite strong to operate as inventor on a patent. From 1999 until 2000, over 500 inventors could be recorded, while within the last time period, there have only been about 280 patents where at least one German university has been registered as inventor, which makes up a decrease of over 50%. Of course, as the overall patenting activity of all German universities has increased over the past ten years, the development path of the number of applicants looks the other way round as Figure 16 indicates. Thus, one can already conclude that the behaviour of the German universities has strongly changed over the past ten years, as they operate much more often as applicants themselves, hence, being less dependent on other institutional actors anymore.

  • 56 Exclusive applicants can freely dispose of their patent and of course of its possible revenues, to (...)
  • 57 All further analyses with regard to cooperation patterns will refer to co-applicant networks.

27Reasons for this opposite development lie possibly in the fact that the so called German ‘Hochschullehrerprivileg’ prevailed until February 2002. This means that after 2002 employees of universities (professors or scientific assistants) could not anymore dispose freely of their intellectual property rights. Thus, before 2002, it is possible that in case the professor or any scientific assistant cooperated with any other institution possessed a kind of weaker position as private person when it went along the question who should operate as applicant and who as inventor on the patent56. On the other side, from February 2002 onwards, the university itself appears frequently as applicant itself when a professor or scientific assistant invent anything which is worth for a patent.57

28Coming now to the development path of the cooperation activity of all German universities, it can be stated that it is generally unusual to operate as exclusive applicant on a patent, as at least one or two further inventors are regularly also involved within a patent application. For example, from 1999 until 2008, there are only about 100 patents where only one German university is recorded as exclusive applicant, with no other inventors on it except of itself. However, this PhD thesis concentrates on co-applicant networks in order to show the development path of their cooperation activities in this regard. The following figure firstly shows the increasing number of co-applicants of the German universities from 1999 until 2008:

Figure 17: Number of Co-Applicants, absolute Numbers, 1999-2008 (own illustration).

29As can be seen from the figure, the number of the German university co-applicants has highly increased over the past ten years, except of a small decline from the first to the second time period. However, this decrease is not surprising as the number of filed patents has also declined from 1999/2000 until 2001/2002. Overall, the German universities have seen an increase of around 150% during the past ten years, while the largest rise has occurred from the third to the fourth time period by around 80%. Overall, it can be stated that not only the patenting activity of the German universities in terms of filed patents has highly increased over the past ten years, but the number of co-applicants, too.

30Up to this point, it is well shown that not only the number of filed patents has highly increased over past ten years, but also the cooperation activity of the German universities as illustrated by the rising number of co-applicants. Overall, they could even triple their number of patent applications, thus operating much more often as applicants themselves so that they are less dependent on other institutional actors anymore. Hence, hypothesis 1 is also confirmed in terms of patents by above findings.

31Finally, it is also shown how the co-applicants of the German universities are institutional distributed over the past ten years as the following figure illustrates:

Figure 18: Institutional Distribution of the German University Co-Applicants, absolute Numbers, 1999-2008 (own illustration).

32First of all, the aforementioned opposite development can be confirmed through the above figure as a highly increasing trend of almost all possible co-applicants can be observed. Only the development path of the private persons looks differently. While they have made up the highest share of co-applicants from 1999 until 2000, they have strongly decreased over the past ten years (minus 85%). In this context, one can assume that almost all private persons have been working at the same German university as the co-applicant, too. Thus, while the German universities have been working more or less 'alone' on a patent within the first time period, they have really changed their cooperation manner during the past ten years, as they have increasingly worked with other institutional partners.

33Coming to the institutional distribution of the German university co-applicants, it is obvious that they have most often cooperated with enterprises. Within the first time period, they have had 31 enterprises as co-applicants, and from 2007 until 2008, they have already cooperated with 187 enterprises (plus 500%). The highest increase of co-applicants from the first to the last time period can be observed within the group of research institutes (plus 540%), even though the absolute number of those partnerships is always much lower compared to the one of the enterprises. Besides, the group of German universities as co-applicants is also very strong. Within the first time period, they have even made up the highest share of co-applicants, but have developed slower than the enterprises. Thus, within the last time period, they have occupied second place between the enterprises and the research institutes. Overall, the German universities could raise the share of cooperating with themselves by 250%. Last, the share of other universities as co-applicants has also increased by 250%, but anyway, it has just made up 6% of all possible co-applicants within the last time period.

  • 58 It has to be mentioned in this regard that some of the German universities are not at all involved (...)

34Further, being aware of the dynamic development of the German university interactions especially to enterprises, research institutes and to themselves, it is further interesting to examine how the role of the German universities as innovative mediators has developed over time. Thus, as it has been done for the scientific publications, the values of normalized degree and betweenness centrality are used to evaluate the increasing importance of the German universities as central nodes for innovation transfer58. First of all, the following table shows the mean values of the normalized degree centrality of all German universities:

Table 13: Mean Value of normalized Degree Centrality of the German Universities, 1999-2008 (own illustration).

35As in the case of publication data, above table also well documents for the patents that the German universities have highly increased their distinct co-applicants over the past ten years, as the value has enlarged by around 120%. Thus, the dynamics regarding the development of the number of distinct co-applicants is much higher compared to the one of the co-authors. Of course, in terms of absolute numbers, both distributions cannot be compared with each other at all. However, the top three German universities within the last time period are Erlangen-Nurnberg (0.0608496), Würzburg (0.0608696) and the Karlsruhe Institute of Technology (KIT) (0.0521739), while the University of Freiburg, the University of Hamburg and the University of Erlangen-Nurnberg are the top three regarding the highest increase over the past ten years.

36At the first glance, it is obvious that the German universities possess different behavioural patterns regarding their publication and patenting activities. This important finding will be discussed later on within the final conclusion and reflection.

37Further, the following table points to the mean values of betweenness centrality of the German universities from 1999 until 2008, thus illustrating how all German universities have developed on average in this regard:

Table 14: Mean Value of Betweenness Centrality of the German Universities, 1999-2008 (own illustration).

38It can be seen that the value of betweenness centrality has increased over the past ten years, differently from the mean value of the betweenness centrality in terms publications. Anyway, the top three universities within the last time period in this regard consist of the KIT (0.115064), Erlangen-Nürnberg (0.0896877) and the Technical University of Munich (0.0652174), while the LMU Munich, the University of Würzburg and the Technical University of Aachen are the top three regarding the highest increase over the past ten years. A detailed examination of the distinct groups of German universities, pointing to significant differences in their behavioural patterns can be found in chapter ten.

39To exemplarily show how a co-applicant network looks like, the following figure illustrates the network of the best-performing university regarding the value of betweenness centrality; thus, it is the co-applicant network of the KIT:

Figure 19: Co-Applicant Network of the KIT, 2007-2008 (own illustration).

40Overall, a strongly growing importance of the German universities as co-applicants in innovation networks can be generally observed over the past ten years, especially with regard to close network collaborations with enterprises. Hence, hypothesis 2b is also confirmed in terms of patents.

  • 59 Test of multi-collinearity has been made; all mentioned variables do not have a strong correlation (...)

41Finally, bearing in mind that industry as well as research linkages has gained in importance over the past ten years, it is again reasonable to ask whether third-party funds have had a significant effect on the collaboration activity of the German universities in general. Hence, above mentioned OLS regression model is applied. The following table now illustrates the results of the model59:

Table 15: Results of the OLS-Regression Model - Development of the Patenting Activity (own illustration).

42The table demonstrates to what extent the four independent variables affect the cooperation activity of the German universities regarding joint patent applications. Thereby, it shows that third-party funds have highly affected the patenting behaviour of the German universities over the past ten years; the result is significant with a standard error less than 1%. Referring to its coefficient, it is again obvious that the value is extremely small. However, as it has been already discussed within the part of the publication analysis in this regard, this result is not too extraordinary as the size of third-party funds is extremely high compared to the number of patent applications. Hence, in case of an increase of third-party funds in the amount of one unit, the number of patent application would hardly change. Nevertheless, the reasons for the significance of the variable can be explained in the fact that the German universities have strongly cooperated with enterprises which are, for example, external funding sources. Further, it is interesting to notice that there are significant results for two of the four dummy variables, too, namely for the elite-universities as well as for the size dummy. For the elite-universities, the significance is on a 99% level. From this, it can be drawn that all significant results are rooted in this special type of university. Last, the size dummy also possesses a slight significant result with a standard error less than 10%. Thus, the size of a German university positively affects the number of co-applicants itself, too.

43Hypothesis 2b is also confirmed in terms of patents as third-party funds have had a positive significant effect on the patenting activity of the German universities.

8.3. Proximity Patterns in Times of Globalization

44This subchapter deals with the comprehensive analysis of the spatial examination with regard to the publication behaviour of the German universities as already illustrated within the part on the methodology. As already brought up, in times of globalisation it is of high interest to explore whether proximity still matters as much as it did years before. Thus, the following figure gives a first overview of the spatial distribution of the co-authors of all German universities within the 1,000 km radius for 2000 and 2009:

Figure 20: Spatial Distribution of the German University Co-Authors (100 km Corridor) weighted Numbers, 2000 and 2009 (own illustration).

45First of all, this kind of illustration also emphasises the fact that the German university co-authors within the 1.000 km radius have highly increased over the past ten years. Second, at the first glance, it also seems very obvious that local co-authors who are within a radius of 100 km are especially important for the German universities as potential cooperation partners. This finding might be also confirmed by the huge increase of these partners from 2000 until 2009. While in 2000 about 1,655 partners have been located very close to the German universities, in 2009 already 4,233 co-authors have been located within the radius of 100 km which makes up an increase of 155%. The other distance corridors have not been able to show such strong increase rates. Third, while the co-authors did not decline that much from the first to the second distance corridor in 2000, in 2009 only half as much co-authors are in the second corridor compared to the first one. Further, another huge decline can be observed from the fifth to the second corridor which accounts for both time periods. In 2000, there were 986 coauthors recorded within the fifth corridor compared to only 600 within the sixth one which makes up a decline of 40%. The same finding accounts for 2009, too. Thus, it is not surprising that the conducted cluster analysis has found the average cut-off point at around 468 km. However, in general, it seems that the number of co-authors declines depending on the distance. In order to acquire a more valid estimation in this regard, a second distance corridor has been evolved which now covers 25 km each. Thus, the following figure shows again the spatial distribution of the German university co-authors for 2000 and 2009.

46Further, the values of Spearman's rank correlation coefficient are listed for both years which can be used to statistically confirm the finding that geographical proximity still matters when those co-authors are considered that are within the radius of 1,000 km:

Figure 21: Spatial Distribution of the German University Co-Authors (25 km Corridor) weighted Numbers, 2000 and 2009 (own illustration).

  • 60 In this context, locality includes all co-authors that are within a radius of only 25 km.

47Figure 21 emphasises even stronger that especially locality60 still plays an important role for the German universities. It is very apparent that from 2000 until 2009 an increase of around 140% can be observed regarding all co-authors that are inside the 25 km corridor. Also the second stronger decline of co-authors can be still seen at the point of 134 around 500 km. Besides, the values of Spearman's rank correlation coefficient both show a quite strong and negative effect. In 2000, the value is -0.80 and in 2009 even -0.85. This means that the number of coauthors declines with the growth of distance.

  • 61 Figure 21 shows a second decline at approximately 475 km. Besides, the cluster analysis has also p (...)

48However, according to figure 21, it can be assumed that locality highly matters while the German universities do not further consider distance for the choice of their co-authors when they are anywhere located between 25 km and approximately 475 km61. Afterwards, proximity seems to play a role again regarding the choice of cooperation partners as above figure indicates. Bearing this in mind, the following figure now firstly illustrates the spatial distribution of the German university coauthors between 25 km and <475 km:

Figure 22: Spatial Distribution of the German University Co-Authors (25km - <475 km) weighted Numbers, 2000 (own illustration).

49As can be seen from the figure, without the co-authors of the first distance corridor and without the co-authors that are farther away than 475 km, the German universities have not seen distance as a barrier, as spearman's rank correlation coefficient still shows a slight positive effect which means that the number of co-authors has even increased with the growth of distance (with p=0.05). Last, the following figure now illustrates the spatial distribution of the German university co-authors that are anywhere between 475 and 1,000 km located:

Figure 23: Spatial Distribution of the German University Co-Authors (475 km -1.000 km) weighted Numbers, 2000 (own illustration).

50Now, by means of above figure, it can be well seen that the number of cooperation partners has declined with the growth of distance as the value of Spearman's rank correlation coefficient is -0.93.

51To sum up, while locality still seems to highly matter regarding the choice of the cooperation partners, regional distance of 25 km until <475 km does not seem to play a role, but from 475 km onwards the friction of distance have started weight again.

52Further, it is aimed to present an overview of the spatial distribution of all German university co-authors, either being regional or supra-regional located (see again chapter six for the classification). Hence, not only the co-authors of the 1,000 km radius are considered, but also all others beyond. By means of the cluster analysis, the following figure shows the share of regional and supra-regional located cooperation partners for the time period from 2000 until 2009:

Figure 24: Spatial Distribution of the German University Co-Authors, weighted Numbers, 2000-2009 (own illustration).

  • 62 Supra-regional located does not necessarily mean that the co-author is outside Germany.

53As can be seen from the figure, the spatial distribution of the German university co-authors is more or less equally distributed when the classification of the cluster analysis is taken into account. Hence, around half of all co-authors are regionally located, i.e. they are located within a radius of around 468 km on average, and others that are farther in distance are considered to be supra-regionally62 located partners. To sum up, considering all cooperation partners it has been shown that regionally and supra-regionally located partners are more or less equally important, even though today the number of the regionally located coauthors has slightly passed the supra-regionally located ones.

54Finally, it is now shown how the institutional cooperation partners of the German universities are spatially distributed. Thereby, it is demonstrated whether industry, university and research partners have become more regional or supra-regional.

Figure 25: Spatial Distribution of the Enterprises, 2000 and 2009 (own illustration).

Figure 26: Spatial Distribution of the Universities, 2000 and 2009 (own illustration).

Figure 27: Spatial Distribution of the Universities, 2000 and 2009 (own illustration).

55Above figures now show the spatial distribution of the different German university cooperation partners. Thereby, the spatial distribution of the enterprises from 2000 and 2009 is firstly presented. While the share of regionally located enterprises has been 54% in 2000, this share has even increased during the past ten years and has accounted for 57% in 2009. Thus, the German universities have started to collaborate even more often with regionally located enterprises. Second, coming to the spatial distribution of the universities as cooperation partners of the German universities, it can be stated that they have only hardly changed from 2000 until 2009. While the share of supra-regionally located universities has been 48% in 2000, it has been 50% in 2009; thus, universities as cooperation partners have been more or less equally distributed. Finally, the spatial distribution of the research institutes of the German universities from 2000 and 2009 is presented. While in 2000, the German universities have more likely cooperated with supra-regional research institutes, they meanwhile are equally engaged in regional and supra-regional network collaborations.

56Overall, it can be stated that especially research and industry partners have become more regional, while the share of regional and supra-regional located universities has remained quite similar, even though the share of supra-regional located universities has become slightly higher. However, as already demonstrated, the overall distribution has shown that regional and supra-regional partners are more or less equal distributed, even though regional cooperations have generally increased in importance compared to supra-regional ones.

Notes

48 In this context, universities include all domestic and foreign universities except of the 76 German universities that are subject of study.

49 In this context, distinct cooperation partners mean that double counts do not occur, i.e. if a German university is linked to the same enterprise several times, this cooperation is only counted once.

50 43 of the German universities have had decreasing values in this regard.

51 28 of the German universities have had decreasing values in this regard.

52 Test of multi-collinearity has been made; all mentioned variables do not have a strong correlation to each other. Further, the residuals are normally distributed and possess homogeneity of variance.

53 The classical German university is actually known as a knowledge-intensive institution that is highly engaged in scientific papers, but less in innovation output pictured through patent applications. Therefore, the differences in the scope of data are still prevalent and not uncommon.

54 The decline from the first to the second period has probably been occurred due to the dot.com bubble which was a historic speculative bubble covering the end of the nineties (1997-2000). The burst of the bubble took place during 2000-2001.

55 In a few cases, the German university is applicant and inventor at the same time. Figure 16 anyway indicates that the patenting behaviour of the German universities has changed from merely inventors to more likely applicants.

56 Exclusive applicants can freely dispose of their patent and of course of its possible revenues, too.

57 All further analyses with regard to cooperation patterns will refer to co-applicant networks.

58 It has to be mentioned in this regard that some of the German universities are not at all involved in innovation networks, while others still possess null values regarding the value of betweenness centrality.

59 Test of multi-collinearity has been made; all mentioned variables do not have a strong correlation to each other. Further, the residuals are normally distributed and possess homogeneity of variance.

60 In this context, locality includes all co-authors that are within a radius of only 25 km.

61 Figure 21 shows a second decline at approximately 475 km. Besides, the cluster analysis has also pointed to an average cut-off point of around 475 km (468 km).

62 Supra-regional located does not necessarily mean that the co-author is outside Germany.

Table des illustrations

Légende Figure 11: Publication Activity of all German Universities, absolute Numbers, 2000-2009 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-1.jpg
Fichier image/jpeg, 111k
Légende Figure 12: Publication and Cooperation Activity of all German Universities, absolute Numbers, 2000-2009 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-2.jpg
Fichier image/jpeg, 104k
Légende Figure 13: Box plot of the Publication Activity of all German Universities, single-authorship vs. co-authorship, 1999-2010 (created by means of STATA).
URL http://books.openedition.org/ksp/docannexe/image/238/img-3.jpg
Fichier image/jpeg, 125k
Légende Figure 14: Institutional Distribution of the German University Co-Authors, weighted Numbers, 2000-2009 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-4.jpg
Fichier image/jpeg, 131k
Légende Table 10: Mean Value of normalized Degree Centrality of the German Universities, 2000-2009 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-5.jpg
Fichier image/jpeg, 76k
Légende Table 11: Mean Value of Betweenness Centrality of the German Universities, 2000-2009 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-6.jpg
Fichier image/jpeg, 72k
Légende Table 12: Results of the OLS-Regression Model - Number of Enterprise Interactions (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-7.jpg
Fichier image/jpeg, 123k
Légende Figure 15: Patenting Activity of all German Universities, absolute Numbers, 1999-2008 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-8.jpg
Fichier image/jpeg, 124k
Légende Figure 16: Distribution of Applicants and Inventors of the German Universities, absolute Numbers, 1999-2008 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-9.jpg
Fichier image/jpeg, 134k
Légende Figure 17: Number of Co-Applicants, absolute Numbers, 1999-2008 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-10.jpg
Fichier image/jpeg, 123k
Légende Figure 18: Institutional Distribution of the German University Co-Applicants, absolute Numbers, 1999-2008 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-11.jpg
Fichier image/jpeg, 150k
Légende Table 13: Mean Value of normalized Degree Centrality of the German Universities, 1999-2008 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-12.jpg
Fichier image/jpeg, 72k
Légende Table 14: Mean Value of Betweenness Centrality of the German Universities, 1999-2008 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-13.jpg
Fichier image/jpeg, 70k
Légende Figure 19: Co-Applicant Network of the KIT, 2007-2008 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-14.jpg
Fichier image/jpeg, 276k
Légende Table 15: Results of the OLS-Regression Model - Development of the Patenting Activity (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-15.jpg
Fichier image/jpeg, 120k
Légende Figure 20: Spatial Distribution of the German University Co-Authors (100 km Corridor) weighted Numbers, 2000 and 2009 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-16.jpg
Fichier image/jpeg, 180k
Légende Figure 21: Spatial Distribution of the German University Co-Authors (25 km Corridor) weighted Numbers, 2000 and 2009 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-17.jpg
Fichier image/jpeg, 141k
Légende Figure 22: Spatial Distribution of the German University Co-Authors (25km - <475 km) weighted Numbers, 2000 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-18.jpg
Fichier image/jpeg, 140k
Légende Figure 23: Spatial Distribution of the German University Co-Authors (475 km -1.000 km) weighted Numbers, 2000 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-19.jpg
Fichier image/jpeg, 143k
Légende Figure 24: Spatial Distribution of the German University Co-Authors, weighted Numbers, 2000-2009 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-20.jpg
Fichier image/jpeg, 143k
Légende Figure 25: Spatial Distribution of the Enterprises, 2000 and 2009 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-21.jpg
Fichier image/jpeg, 127k
Légende Figure 26: Spatial Distribution of the Universities, 2000 and 2009 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-22.jpg
Fichier image/jpeg, 88k
Légende Figure 27: Spatial Distribution of the Universities, 2000 and 2009 (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/238/img-23.jpg
Fichier image/jpeg, 86k

Lire

Open access

Acheter