Elsevier

Journal of Informetrics

Volume 4, Issue 1, January 2010, Pages 97-106
Journal of Informetrics

The impact of small world on innovation: An empirical study of 16 countries

https://doi.org/10.1016/j.joi.2009.09.003Get rights and content

Abstract

This paper investigates the impact of small world properties and the size of largest component on innovation performance at national level. Our study adds new evidence to the limited literature on this topic with an empirical investigation for the patent collaboration networks of 16 main innovative countries during 1975–2006. We combine small world network theory with statistical models to systematically explore the relationship between network structure and patent productivity. Results fail to support that the size of largest component enhances innovative productivity significantly, which is not consistent with recent concerns regarding positive effects of largest component on patent output. We do find that small-world structure benefits innovation but it is limited to a special range after which the effects inversed and shorter path length always correlates with increased innovation output. Our findings extend the current literature and they can be implicated for policy makers and relevant managers when making decisions for technology, industry and firm location.

Introduction

Innovation has been widely studied as it can improve productivity and enhance economy development. The innovation capability has also come to be an important part of the competitive power of the firms and regions as well as countries. Their innovation capabilities have been measured by their performance on patents, research publications, sales of new products and so on (Guan and Liu, 2005, Guan and Ma, 2004, OECD, 2005, Schilling and Phelps, 2007). The studies have shown significant different performance among innovation players. Why do those differences exist? Researchers have studied from different aspects, such as the infrastructures, skilled labors and their mobility, venture capital for innovative risk-taking, R&D stock and intellectual property protect polices (Davila et al., 2003, Feldman, 2001, Florida, 2004, Guo, 2008, Mowery et al., 2002, Stolpe, 2002). Many studies of innovation referred to social networks as the crux of innovation (Almeida and Kogut, 1999, Singh, 2005). It is believed that social networks, both informal friendship and formal collaborate networks, contribute to innovation by their ability of information, knowledge and technology diffusion. Saxenian's (1994) survey of Silicon Valley showed that its informal networks of friendship and collaboration have important contributions to its success. Singh (2005) reported that there is significant positive effect of patent collaboration, both direct and indirect, on the future information flow. Singh also demonstrates that a large fraction of the geographical citation spillover results from collaboration networks. Fleming, King, and Juda (2007) investigated the collaboration network of USPTO patents and interview some of the inventors. Most interviewees reported some degree of technical interaction after a patent collaboration and even when they no longer worked at the same firm.

Although more researches have connected innovation with social network, we still know little about the relationship between network properties and innovation performance among countries. Several years before, this kind of study had been restricted by the computation ability for complex networks. Now it becomes much more possible with help of the rapid development and application of complex network theories, computer computation ability and improved network software.

Small world networks have received widely notice in diverse fields because of their argued advantage in information diffusion, creativity achievements, trust, learning and collaboration. Small world networks refer to a kind of networks with high local clustering, as measured by clustering coefficient, while the average number of intermediates needed to connect any two actors relatively short, as measured by average path length (Watts & Strogatz, 1998). For the character of local dense, small world networks are believed to enable trust, close collaboration, and further promote accurate and efficient communication and information diffusion, while on the other hand, the short path length can bring fresh and nonredundant information from distant ties, and then inspire new ideas and creativity. So in the past studies there was a general hypothesis that small world networks should enhance innovative creativity (Cowan and Jonard, 2004, Schilling and Phelps, 2007, Uzzi and Spiro, 2005, Watts, 1999). Uzzi & Spiro (2005) examined the existence of small-world structure of the Broadway musicals collaborate network and its effect on artists’ creativity, as measured by both financial profitability and artistic critics’ reviews. The results show a parabolic relationship between a small world and the performance of the actors within it, which means that performance increased up to a threshold, after which point the positive effects reversed. Guimera, Luis, and Amaral (2005) constructed the co-authorship networks for four academic disciplines—economics, social psychology, astronomy, and ecology. They also found that the best performance accompany with a medium level of connectivity and cohesion, which was consistent with Uzzi and Spiro's findings (2005). Another study of the industry alliance networks showed that, firms embedded in networks with dense local clustering and short average path lengths are more likely gain access to knowledge important for innovation, and then have better innovation output, as measured by patent performance (Schilling & Phelps, 2007). Besides these empirical studies, a simulation test of network architecture and the knowledge diffusion performance also found that the performance of the system exhibits clear ‘small world’ properties, in that the steady-state level of average knowledge among agents is maximal whereas the coefficient of variation is minimal when the structure is a small world (Cowan & Jonard, 2004). Fleming et al. (2007) studied the patent inventors’ collaboration network of the U.S. metropolitan statistical areas. Though they failed to find relationship between the small-world structure of these networks and patenting rates in these regions, they did find that short path lengths and the size of the largest component positively correlated with patenting, suggesting that the level of connectivity within the larger subcomponents of a small world does have an effect on performance.

Despite the several empirical studies listed above of small world network and innovation at industrial, discipline or regional level, we still know little about the relationship between network properties and innovation at national level. In this study, we develop and exploit a novel database on patent collaboration for 16 innovative countries to investigate the impacts of small world networks on innovation output. This study extends the existing literature in several ways. First, we draw from a broad sample of patent co-authorship at national level, rather than focusing on industries or areas. We have included most of the innovation countries for a wide time span, from 1975 to 2007. Second, when we calculate the small world indices, the collaboration networks are treated as bipartite networks using methods developed by Newman, Strogatz, and Watts (2001), rather than as unipartite networks. Finally, the networks we build are dynamic through five-year time window, rather than collapse time and look at the entire network as one large cross-section.

The remaining parts of the paper are organized as follows. The next section considers the possible relationship between small world networks and innovation performance, as measured by patent numbers, and develops the hypotheses to be tested. Then we describe our data, methods for network generation and small world measures, and statistics models in Section 3. Section 4 presents our empirical results and explores the relationship between small world networks and innovation. Discussions of the results are reported in Section 5 and conclusions are provided in Section 6.

Section snippets

The relationship between small worlds and innovation

We focus our study at the country level, allowing us to explore the relationship between small world collaboration network and the national innovation output. It is argued that small worlds can improve creativity and innovation. The influences can be organized into three aspects, clustering, path length and their interaction.

The study of Broadway musical collaboration networks has found a parabolic relationship between clustering and musical performance (Uzzi and Spiro, 2005). There are

Network generation

In this study we will construct the patent collaboration networks of the main innovation countries to investigate the impact of their small-world structure on innovation output. The Growth Competitiveness Index used in the Global Competitiveness Report since 2001–2002 separates countries into core innovators and non-core innovators (Porter & Schwab, 2004). Core innovators are those economies “whose growth is largely driven by their capacity to innovate because they are close to the

Results

Table 4 presents our regression analysis of the patent number. At the beginning model 1 looks at the control variables. Then we add the size of the largest component measure in model 2 and small world measures in models 3 and 4. Here we use two specifications of the small world model as previous study (Uzzi & Spiro, 2005). We first separately include the PL ratio and CC ratio along with its square in model 3. Then we investigate their interaction term Q and its square in model 4.

Both R&D

Discussion

The objective of this paper was to study the relationship between small-world structure of the patent collaboration networks and their creativity output. We constructed the networks using USPTO inventor data that included every inventor who worked on patents granted from 1975 to 2006. These networks are unipartite projections of bipartite networks (also called as affiliation networks) and they have different statistical properties from that of unipartite networks, such as their clustering

Conclusions

The literature on small worlds has grown rapidly and its relation with creativity has received more attention in the social science and management literature. However, we still known little of the mechanisms it works in real world. This research contributes to the existing literature by providing an empirical investigation of the relationship between small worlds and innovation with a data set containing patent inventors from the main innovative countries across a long period. We construct

Acknowledgments

This research is funded by National Natural Science Foundation of China (Project No. 70773006) and Shanghai Leading Academic Discipline Project (Project No. B210). The authors are grateful for the valuable comments and suggestions of Prof. Egghe and anonymous reviewers, which significantly improved the article.

References (33)

  • J.C. Guan et al.

    Comparing regional innovative capacities of PR China based on data analysis of the national patents

    International Journal of Technology Management

    (2005)
  • J.C. Guan et al.

    A comparative study of research performance in computer science

    Scientometrics

    (2004)
  • R. Guimera et al.

    Cartography of complex networks: Modules an universal roles

    Journal of Statistical Mechanics, Theory, and Experiments

    (2005)
  • B. Guo

    Technology acquisition channels and industry performance: An industry-level analysis of Chinese large- and medium-size manufacturing enterprises

    Research Policy

    (2008)
  • A. Hargadon

    How breakthroughs happen: The surprising trust about how companies innovate. Chapter3: Bridging Small Worlds

    (2003)
  • J. Hausman et al.

    Econometric models for count data with an application to the patents R&D relationship

    Econometrica

    (1984)
  • Cited by (87)

    View all citing articles on Scopus
    1

    These authors contributed equally to this paper.

    View full text