Technology transfers from universities and innovation

Master in Economics: Empirical Applications and Policies University of the Basque Country UPV/EHU Master Thesis Technology transfers from universities and innovation Mirian Oiz Chueca Supervised by Javier Gardeazabal Matías July 22nd, 2024 Mirian Oiz / Master in Economics: Empirical Applications and Policies - 2 - Table of contents 1. INTRODUCTION …………………………………………………………… 4 2. LITERATURE REVIEW ……………………………………………………. 5 3. DATA DESCRIPTION ……………………………………………………… 6 3.1. Variable definition ……………….…………………...…………….…… 7 3.2. Descriptive analysis ………………………..…………………….……… 9 4. METHODOLOGY …………………………………………...……………... 10 5. RESULTS …………………………………………………………………... 12 5.1. Effect of collaboration with universities ……………….……………… 12 5.2. Comparison with the effect of collaboration with other companies ……. 20 5.3. Comparison with the effect of collaboration with any type of organization ……………………………………………………………. 21 6. CONCLUSIONS …………………………………………….……………... 22 7. REFERENCES ……………………………………………………….…….. 24 8. APPENDIX ..………………………………………………………………. 26 Mirian Oiz / Master in Economics: Empirical Applications and Policies - 3 - Abstract This study uses causal inference methods to assess the impact of university research and development (R&D) transfers on firm innovation, utilizing the Spanish survey (PITEC) spanning the period from 2003 to 2016. In particular, multivariate Nearest Neighbor Matching (NNM) method is employed for the estimation of the Average Treatment Effect on the Treated (ATET). The findings indicate that university collaboration has a mostly positive impact on a range of innovation outcomes, including process innovation, product innovation, and patent registration, but it mainly not significant. These findings lend support to the proposition that university R&D transfers are an important factor in stimulating firm innovation. Finally, the research acknowledges the limitations of its methodology and suggests avenues for future research. Key words: R&D transfers, innovation, universities Mirian Oiz / Master in Economics: Empirical Applications and Policies - 4 - 1. INTRODUCTION In the contemporary business environment, characterized by rapid change and intense competition, the capacity to innovate is of paramount importance for firms seeking to succeed and gain a competitive advantage. Universities, with their substantial intellectual capital and pioneering research, serve as a important conduit for instilling an environment conducive to innovation. In light of this, there has been a notable increase in the number of university-industry collaborations, with firms proactively engaging with universities in order to gain access to their expertise and knowledge resources. This research explores the complex interrelationship between university R&D collaboration with firms and this last ones’ innovation. This study specifically examines the impact of university collaboration on a firm's innovative performance, encompassing process innovation, product innovation, and patent registration. The analysis employs data from the Spanish survey, PITEC (Panel de Innovación Tecnológica, Panel of Technological Innovation), which covers R&D transformation in companies from 2003 to 2016. The survey gathers comprehensive data on a range of innovation-related topics, including university-industry collaboration, innovation outcomes, and firm characteristics, to be used for the research analysis. The methodology employed combines multivariate nearest neighbor matching, with accounts for observed confounding, and difference-in-difference, which accounts for time invariant unobserved confounding. This research is driven by the research question: does the transfer of research and development (R&D) from universities to firms lead to increased innovation in those firms? By analyzing data from the PITEC survey and employing a robust methodology, this study aims to answer this question and contribute to the understanding of the role that this university transfers plays in driving innovation in firms. The analysis will commence with a review of the extant literature, followed by a comprehensive account of the available data and the methodology to be employed. Subsequently, the findings will be presented, and some conclusions will be drawn. Mirian Oiz / Master in Economics: Empirical Applications and Policies - 5 - 2. LITERATURE REVIEW The convergence of academic research and industrial innovation represents a significant driver of economic growth. This review examines the impact of university research and development (R&D) transfers on firm innovation. This study examines the role of universities as catalysts, facilitating the transformation of knowledge into cuttingedge products and processes. Several studies support a positive association between university technology transfer and firm innovation. García-Vega & Vicente-Chirivella (2020) demonstrate a significant rise in innovativeness for firms acquiring R&D from universities. Carboni & Medda (2021) reinforces this link, highlighting the effectiveness of university R&D in promoting product innovation. Furthermore, Li & Tan (2020) find a positive spillover effect from university R&D activities on firm innovation within the same geographic location. The effectiveness of university R&D transfers depends on various factors. Bellucci & Pennacchio (2015) emphasize the importance of a university's entrepreneurial spirit and the quality of its research. Firms with open search strategies and a focus on radical innovation benefit more from university knowledge (Bellucci & Pennacchio, 2015). Additionally, García-Vega & Vicente-Chirivella (2024) suggest that a firm's absorptive capacity, its ability to understand and utilize external knowledge, plays a critical role in leveraging university R&D effectively. University R&D transfers can also generate positive externalities beyond direct knowledge acquisition. García-Vega & Vicente-Chirivella (2020) find that these transfers enhance firms' internal R&D capabilities, potentially creating a virtuous cycle of innovation. Similarly, Acosta et al. (2009) highlight the role of university graduates as a source of knowledge spillovers, contributing to new business formation in hightechnology sectors. Collaboration networks further influence the benefits firms derive from university R&D. Bolívar-Ramos (2017) finds that firms collaborating with universities in national Mirian Oiz / Master in Economics: Empirical Applications and Policies - 6 - or regional networks show a stronger patenting propensity compared to those without such collaborations. However, Becker et al. (2023) suggest that university spillovers can be limited, with some evidence for positive effects on patenting and new-tomarket innovation in specific circumstances. While university R&D transfers offer significant advantages, challenges remain. Kanama & Nishikawa (2015) identify various obstacles in innovation activities, such as financial and technological limitations, that drive firms to seek university knowledge. However, access to university knowledge may not always translate into profitable innovation. In conclusion, a growing body of research supports the positive impact of university R&D transfers on firm innovation. The effectiveness of these transfers depends on factors such as the firm's absorptive capacity, the university's research environment, and the nature of collaboration networks. Further research is needed to explore the mechanisms underlying university R&D spillovers and identify strategies to maximize their benefits for firms. 3. DATA DESCRIPTION This section presents a description of the database used, the variables employed for the analysis, and the new variables that have been generated from the existing ones. It will conclude with a brief descriptive analysis of the sample. In order to ascertain the influence of collaboration with universities on organizational innovation, the present study utilizes an existing database from a Spanish survey, namely PITEC, conducted between 2003 and 2016. The PITEC (Panel de Innovación Tecnológica, Panel of Technological Innovation) project has the objective of representing the situation and evolution of innovative companies in Spain, as well as the detection of opportunities and needs in this field. The project commenced in 2003 with the collaboration of the FECYT (Fundación Española para la Ciencia y la Tecnología, Spanish Foundation for Science and Technology) and Mirian Oiz / Master in Economics: Empirical Applications and Policies - 7 - the Cotec Foundation, with the input of a group of researchers from various universities. However, this ceased to be carried out in 2016. The data from this period offers more than 460 variables derived from approximately 12,000 companies surveyed from 2003 onward, facilitating the construction of time series that can be used to study the evolution and impact of innovation in the business sector, as well as to identify the diverse innovation strategies employed by companies in question. As it is a fixed panel, an annual observation is collected from each firm, which ensures that the data obtained are of high quality and reliability. The panel of companies is selected from the national surveys carried out by the INE (Instituto Nacional de Estadística, National Statistics Institute) in the innovation sector: "Survey on Technological Innovation in Companies" and "Statistics on R&D activities". 3.1. Variable definition As previously stated, the database comprises a multitude of variables measuring diverse qualities, from which we have identified those pertinent to our investigation. The primary variable of interest, which will serve as the basis for inference, is the transfer of research and development (R&D) funds from universities to firms. They appear as 'Purchase of R&D services in Spain/universities' and 'Purchase of R&D services abroad/universities'. They represent the proportion of external R&D expenditure in the firm, along with the purchase of services in other entities, which we will later additionally analyze. From these, the main treatment variable for the study has been created, which takes value of 1 if the firm has received transfers from universities and 0 if it has not. Furthermore, as in García-Vega and Vicente-Chirivella (2020), we have defined a class of firms that have been called "starters". This denomination is accompanied by the year in which the company collaborates with a university for the first time (first time it receives transfers). In this way, as many starter variables have been generated as there are years in the database, with a dummy category and a value of 1 if the company is a starter that year and 0 otherwise. In table 1 the starter count for each year can be seen. The number of starters for each year of the sample is spread over the years, with all years Mirian Oiz / Master in Economics: Empirical Applications and Policies - 8 - having figures in the approximate range of between 400 and 800, except year 2014, with more than 1800, and 2015 with 203 firms. Table 1 Starters (first time receiving transfers) Year Number of starters 2004 591 2005 884 2006 503 2007 631 2008 547 2009 516 2010 511 2011 468 2012 409 2013 426 2014 1826 2015 203 2016 575 Source: Own elaboration with PITEC data The variables employed as treatment are the ‘starters’ ones discussed in the preceding paragraphs, with 2006 being the example year: the pre-treatment period is defined as spanning 2003 to 2005, while the post-treatment period, during which the effects of treatment can be observed, is defined as spanning 2006 to 2016. The treated group is defined as the firms which commenced collaboration with universities in 2006, while the control group is defined as the firms that neither collaborated with universities nor any other of the entities. As for the dependent variables, different outcomes have been taken as indicators of the degree of innovation or innovative capacity of the firms. These are whether or not the company has had a process innovation in the previous two years, whether or not it has had a product innovation in the previous two years, whether or not it has had a patent application and the number of patents registered. The impact of collaboration between companies and universities on the four has been investigated over the course of the following years. Mirian Oiz / Master in Economics: Empirical Applications and Policies - 9 - In closing, several control variables have been selected for use as covariates in the estimation of effects. These include the number of employees of the firm, the amount of physical capital, the amount of external research and development expenditure, the amount of internal research and development expenditure, the presence or absence of exports, and finally, whether the firm belongs to a business group. 3.2. Descriptive analysis Prior to estimating the effects in question, a brief descriptive analysis was conducted to gain a comprehensive understanding of the sample. The data are presented in Table 2, which shows the number of observations, the mean, and the standard deviation for the full sample, as well as for the observations of firms that have received university transfers and those that have not, separately. Table 2 Descriptive statistics Full sample With university transfers Without university transfers Obs. Mean Std. dev. Obs. Mean Std. dev. Obs. Mean Std. dev. Treatment Va r i a b l e Transfers from universities 171.511 0,261 0,439 44.812 1 0 126.699 0 0 Outcome variables Process innovation 135.798 0,467 0,498 9.110 0,704 0,457 126.688 0,451 0,498 Product innovation 135.799 0,464 0,498 9.110 0,744 0,437 126.689 0,445 0,497 Patents 135.797 0,101 0,301 9.109 0,311 0,463 126.688 0,086 0,281 Number of patents 125.717 0,487 5,938 8.325 2,238 14,385 117.392 0,363 4,781 Control variables Employment 135.810 333,624 1.527,846 9.111 393,854 1.337,530 126.699 329,296 1.540,540 Physical investment 135.810 4,507 12,970 9.111 10,807 17,271 126.699 4,055 12,482 Researchers in R&D 135.810 6,765 18,684 9.111 23,904 23,773 126.699 5,533 17,634 Internal R&D 135.810 39,609 43,815 9.111 66,055 26,574 126.699 37,708 44,195 Exports 171.511 0,628 0,483 44.812 0,937 0,243 126.699 0,519 0,500 Group 135.724 0,411 0,492 9.088 0,503 0,500 126.636 0,405 0,491 Source: Own elaboration with PITEC data Mirian Oiz / Master in Economics: Empirical Applications and Policies - 16 - Figure 2 shows the effects in a timeline fashion, and although as noted above most of them are positive, they are not as robust or representative as might be expected at first. Figure 2 Impact of treatment on product innovation Source: Own elaboration with PITEC data A similar pattern emerges with respect to the probability of filing a patent when initiating collaboration with universities. This outcome underscores that, except for a few years in the 2006 cohort and towards the end of 2015 and 2016, the probability of filing a patent is higher when starting to collaborate with universities. The numerical effects are presented in Table 5. It is in this outcome that the effects are most significant, and thus it can be stated that, with regard to innovation, the most notable effect observed in the -.12 -.08 -.04 0 .04 .08 .12 Impact 2011 2012 2013 2014 2015 2016 Date Impact estimate 95% confidence interval Data from 2011-2016 Impact on Product Innovation for 2011 starters Mirian Oiz / Master in Economics: Empirical Applications and Policies - 17 - analysis is the increase in the proportion of those who have registered patents with respect to those who have not registered any. Furthermore, as illustrated in Figure 3, most effects are positive, with values exceeding zero. Table 5 Change in the probability of registering a patent Impact in year 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Starter of 2006 .0011 (.0152) .0243 (.0178) .0209 (.0205) -.0124 (.0206) .0080 (.0216) -.0170 (.0210) -.0163 (.0218) .0070 (.0227) -.0180 (.0241) .0015 (.0238) -.0115 (.0235) 2007 .0322** (.0156) .0373** (.0182) .0269 (.0197) .0247 (.0211) .0244 (.0216) .0251 (.0219) .0359 (.0223) .0242 (.0237) .0478** (.0240) .0308 (.0234) 2008 .0499*** (.0169) .0281 (.0185) .0401* (.0209) .0291 (.0218) .0328 (.0221) .0634*** (.0226) .0361 (0238) .0471* (.0241) .0304 (.0247) 2009 .0082 (.0165) .0283 (.0196) .0204 (.0216) .0438** (.0221) .0589** (.0240) .0467* (.0254) .0319 (.0254) .0299 (.0252) 2010 -.0048 (.0173) -.0181 (.0195) .0084 (.0202) .0254 (.0233) .0202 (.0241) .0210 (.0239) .0301 (.0237) 2011 .0065 (.0183) .0136 (.0205) .0458** (.0227) .0289 (.0237) .0146 (.0246) .0099 (.0261) 2012 .0264 (.0194) .0383* (.0213) .0247 (.0236) .0093 (.0251) -.0073 (.0258) 2013 .0514** (.0199) .0402* (.0212) .0195 (.0228) .0010 (.0249) 2014 .0210 (.0218) .0243 (.0225) .0133 (.0114) 2015 .0171 (.0203) -.0153 (.0219) 2016 -.0224 (.0246) ∗ Significant at 10%; ∗∗ Significant at 5%; ∗∗∗ significant at 1%. Source: Own elaboration with PITEC data Mirian Oiz / Master in Economics: Empirical Applications and Policies - 18 - Figure 3 Impact of treatment on patent registration Source: Own elaboration with PITEC data Finally, the last outcome to be analyzed is the change in the number of patents filed (Table 6). The estimates obtained exhibit a high variability, and there is no clear pattern of positive or negative effects with almost all estimates being insignificantly different from zero, so no firm conclusions can be drawn about these effects. It can also be seen that the standard errors are considerably larger than the others, and all this could be related to the continuous nature of the outcome. Mirian Oiz / Master in Economics: Empirical Applications and Policies - 19 - Table 6 Change in the number of registered patents Impact in year 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Starter of 2006 .0438 (.0920) .1165 (.2004) -.4500 (.6423) -.0622 (.3324) .0164 (.4542) -.2942 (.2816) -.1071 (.4754) .1195 (.4782) .3030 (.5500) -.1630 (.5754) -.1946 (.6685) 2007 .1545 (.1819) -1.163 (1.339) .1819 (.2491) .2150 (.3651) -.0374 (.2029) .0055 (.3950) .0041 (.4108) .1086 (.4862) -.4652 (.8415) -.6558 (.8689) 2008 -.6358 (.8092) .2230 (.2658) .1796 (.3966) -.3360 (.3484) .2267 (.5077) -.1975 (.5495) .3050 (.4998) -.8837 (1.377) -.8229 (1.328) 2009 .9361 (.9608) 1.128 (.9464) .5410 (.9700) 1.040 (1.016) 1.091 (.9966) .6151 (.9276) -.9582 (1.569) -1.029 (1.494) 2010 .0302 (.2256) -.6533* (.3855) .0250 (.4572) -.4365 (.3945) -.2279 (.4886) -.6273 (.5424) -1.141 (.7150) 2011 -.7564 (.5721) -.3828 (.3853) -.3576 (.4088) .2046 (.5254) -1.997 (1.423) -1.629 (1.636) 2012 .3527 (.4991) .4364 (.4661) .5516 (.5408) -1.387 (1.594) -1.483 (1.554) 2013 .2240 (.1956) .5124* (.2846) -1.538 (1.601) -1.046 (1.679) 2014 .6782 (.5710) -.4408 (.9640) -.3039 (.2792) 2015 -1.539 (.9540) -2.17** (1.012) 2016 -.7493 (.5091) ∗ Significant at 10%; ∗∗ Significant at 5%; ∗∗∗ significant at 1%. Source: Own elaboration with PITEC data Graphically, we also find more irregular lines than in the graphs of the other results, highlighting a drop in the number of patents registered in 2008, probably related to the Spanish crisis of that year, which can be analyzed in the graphs of several of the starters. Mirian Oiz / Master in Economics: Empirical Applications and Policies - 20 - Figure 4 Impact of treatment on number of patent registration Source: Own elaboration with PITEC data 5.2. Comparison with the effect of collaboration with other companies (product innovation) With the intention of extending the analysis carried out and finding new characteristics of the determinants of entrepreneurial innovation, the initial variables and the ATET estimates were reformulated, but considering that the treatment was now collaboration with other firms. In this way, the effect on entrepreneurial innovation can be compared with the estimates obtained for collaboration with universities. Mirian Oiz / Master in Economics: Empirical Applications and Policies - 21 - Due to the size of the results tables, one result (in this case product innovation) has been selected and included in Appendix 1. The results for R&D transfers with other enterprises are similar to those for universities. For most starters, the effect on the probability of having a product innovation is positive when they start to purchase research services from other enterprises, with these figures being very significant for starters enterprises in recent years. Figure 5 shows that the effects follow a similar pattern for both types of collaboration and that the range of the estimates and the confidence intervals are the same (collaboration with universities is the blue line and with companies the red line). Figure 5 Comparison for product innovation for starters of 2008 and 2009 Source: Own elaboration with PITEC data 5.3. Comparison with the effect of collaboration with any type of organization (patent registration) To complete this extension of the study, a different treatment group was again selected, consisting of those enterprises that started to collaborate with one of the organizations for which the sample provides data (group enterprises, other enterprises, universities, public administration organizations, private non-profit organizations and Mirian Oiz / Master in Economics: Empirical Applications and Policies - 22 - international organizations). The control group includes all firms that have never collaborated with anyone, and to show the comparison of results, the outcome of patenting has been chosen (the table of estimates can be found in Appendix 2). As expected, the effect found is also predominantly positive. If the individual collaborations with universities and other companies were positive, we would expect an even larger change in the probability of filing a patent. However, the observed effect is not as significant as anticipated and cannot be extended to the entire population (only to some starters in the sample). This comparison can also be seen in graphical form in Figure 6, where collaboration with universities is represented by the blue line and joint collaboration by the green line. Figure 6 Comparison for patent registration for starters of 2009 and 2010 Source: Own elaboration with PITEC data 6. CONCLUSIONS This study provides weak evidence in support of the proposition that university collaborations foster innovation within firms. The findings illustrate that the majority of participating organizations have experienced positive but mostly insignificant impacts on various innovation outcomes, including process and product innovation. These results are Mirian Oiz / Master in Economics: Empirical Applications and Policies - 23 - not quite in accordance with existing research which has identified the important role played by the transfer of university research and development to firms in stimulating innovation. It is important to acknowledge that the findings of this study are subject to certain limitations. The methodology employed differs from that described in the original papers. Furthermore, the study concentrates on the influence of university collaboration, whereas other factors, including the abilities and expertise of employees, business activity, financial resources, and technological infrastructure, also exert a considerable influence on innovation. Further research should investigate these multifaceted factors in greater depth in order to gain a more comprehensive understanding of their interaction with university collaboration and its ultimate impact on firm innovation. By investigating the moderating and mediating effects of these variables, researchers can provide more detailed knowledge of the circumstances under which university collaborations are most effective in driving innovation. Despite these limitations, the findings of this study highlight the significant role of institutional collaboration in stimulating innovation within firms. By leveraging the expertise and resources of universities, firms can enhance their innovative capabilities, leading to the development of new products, processes, and services that drive economic growth and societal well-being. Mirian Oiz / Master in Economics: Empirical Applications and Policies - 24 - 7. REFERENCES Abadie, A., & Imbens, G. W. (2011). Bias-Corrected Matching Estimators for Average Treatment Effects. Journal Of Business & Economic Statistics, 29(1), 1-11. https://doi.org/10.1198/jbes.2009.07333 Abadie, A., Drukker, D., Herr, J. L., & Imbens, G. W. (2004). Implementing Matching Estimators for Average Treatment Effects in Stata. The Stata Journal, 4(3), 290-311. https://doi.org/10.1177/1536867x0400400307 Acosta, M., Coronado, D., & Flores, E. (2009). University spillovers and new business location in high-technology sectors: Spanish evidence. Small Business Economics, 36(3), 365376. https://doi.org/10.1007/s11187-009-9224-4 Becker, B., Roper, S., & Vanino, E. (2023). Assessing innovation spillovers from publicly funded R&D and innovation support: Evidence from the UK. Technovation, 128, 102860. https://doi.org/10.1016/j.technovation.2023.102860 Bellucci, A., & Pennacchio, L. (2015). University knowledge and firm innovation: evidence from European countries. The Journal Of Technology Transfer, 41(4), 730-752. https://doi.org/10.1007/s10961-015-9408-9 Bolívar-Ramos, M. T. (2017). The relation between R&D spending and patents: The moderating effect of collaboration networks. Journal Of Engineering And Technology Management, 46, 26-38. https://doi.org/10.1016/j.jengtecman.2017.11.001 (Bolívar-Ramos, 2017) Carboni, O. A., & Medda, G. (2021). External R&D and product innovation: Is over-outsourcing an issue? Economic Modelling, 103, 105601. https://doi.org/10.1016/j.econmod.2021.105601 Cerulli, G., Gabriele, R., & Potì, B. (2016). The role of firm R&D effort and collaboration as mediating drivers of innovation policy effectiveness. Industry And Innovation, 23(5), 426-447. https://doi.org/10.1080/13662716.2016.1169159 García-Vega, M., & Vicente-Chirivella, Ó. (2020). Do university technology transfers increase firms’ innovation? European Economic Review, 123, 03388. https://doi.org/10.1016/j.euroecorev.2020.103388 García-Vega, M., & Vicente-Chirivella, Ó. (2024). The Role of Public External Knowledge for Firm Innovativeness. International Journal Of Industrial Organization, 103056. https://doi.org/10.1016/j.ijindorg.2024.103056 Kanama, D., & Nishikawa, K. (2015). What type of obstacles in innovation activities make firms access university knowledge? An empirical study of the use of university knowledge on innovation outcomes. The Journal Of Technology Transfer, 42(1), 141-157. https://doi.org/10.1007/s10961-015-9459-y Li, X., & Tan, Y. (2020). University R&D activities and firm innovations. Finance Research Letters, 37, 101364. https://doi.org/10.1016/j.frl.2019.101364 Mirian Oiz / Master in Economics: Empirical Applications and Policies - 25 - Datasource INE, Intituto Nacional de Estadística. (2003-2016). Microdata Panel de Innovación Tecnológica (PITEC) [Conjunto de datos].