Journal of Informetrics

Size: px
Start display at page:

Download "Journal of Informetrics"

Transcription

1 Journal of Informetrics 4 (2010) Contents lists available at ScienceDirect Journal of Informetrics journal homepage: The Chinese innovation system during economic transition: A scale-independent view Xia Gao a,, Xiaochuan Guo a, Katz J. Sylvan b,c, Jiancheng Guan d a School of Economics and Management, Inner Mongolia University, Da Xue West Road 235, Hohhot, PR China b SPRU, The Freeman Centre, University of Sussex, Brighton, East Sussex BN1 9QE, UK c Department of Mathematics and Statistics, University of Saskatchewan, Saskatoon, Sask., Canada S7N SE6 d School of Management, Fudan University, Shanghai, PR China article info abstract Article history: Received 15 March 2010 Received in revised form 4 July 2010 Accepted 14 July 2010 Keywords: Scale-independent indicators Chinese national and regional innovation systems Economic transition This paper uses scale-independent indicators to explore the Chinese national and regional innovation systems during economic transition. Our perception of an innovation system is frequently informed by conventional indicators based on linear assumptions while actually innovation systems may behave differently. Scale-independent indicators characterize non-linear properties of an innovation system. They can give decision makers deeper insight into the dynamics of innovation systems, and they may lead to more practical public policies [Katz, J. S. (2006). Indicators for complex innovation systems. Research Policy, 35, ]. As reported for the European and Canadian innovation systems the Chinese systems exhibited scaling correlations between GERD (Gross Expenditure on Domestic R&D) and GDP (Gross Domestic Product) over time and at points in time. The scaling factors of the correlations tell us that between 1995 and 2005 the Chinese GERD exhibited a strong nonlinear tendency to increase with GDP. Furthermore they show that the GERD of the Western region is growing much slower than its GDP as compared with Eastern and Central regions. This observation has policy implications suggesting further improvements need to be made to the research infrastructure and funding of the Western region. The GDP POP (Population) scaling factor shows that the wealth intensity or GDP per capita is increasing much faster than the exponential growth of the Chinese population. In contrast the systemic GDP POP scaling factor shows that regional development is non-linear. Finally, the paper GDP and patent GDP scaling factors tell us that outputs of science and technology for China are growing faster than economic growth. The systemic paper GDP and patent GDP scaling factors show that the growth rates are uneven across the provinces Elsevier Ltd. All rights reserved. 1. Introduction A system of innovation can be defined as the set of institutional and business organizations which, within a specific geographical area, 1 interact with the aim of allotting resources to performing activities geared to generating and spreading knowledge which supports the innovations which are the basis of economic development (Buesa, 2002). Broadly speaking, Corresponding author. address: gaoxia1976@sina.com (X. Gao). 1 The authors see no reason that a system of innovation need be limited to a geographical area. Such a limitation implies the innovation system is a regional system /$ see front matter 2010 Elsevier Ltd. All rights reserved. doi: /j.joi

2 X. Gao et al. / Journal of Informetrics 4 (2010) an innovation system is composed of individuals and organizations that directly and indirectly invest time, energy and resources in the production of scientific and technical knowledge (Katz, 2006). Initially, the study of innovation systems made reference to the national environment (Edquist, 1993; Freeman, 1987; Lundvall, 1992; Nelson, 1993) but soon researchers applied the concepts also at the regional level (Buesa, Heijs, Pellitero, & Baumert, 2006; Cooke, 2000, 2001; Cooke, Gómez Uranga, & Etxebarría, 1997; Guan & Liu, 2005; Tuzi, 2005). A regional innovation system is a geographical subset of a larger (often but not necessarily, national) system whose main identifying characteristics are similar at both levels of observation. A regional innovation system uses networks between public and private agents in a specific territory to develop and use their infrastructure to adapt, generate and extend knowledge (Buesa et al., 2006). The impact of technology and industrial policies on the economic growth of a nation cannot be understood unless their impact on the geographical regions is understood (Hilpert, 1991). As one of the largest transitional economies, China has experienced a dramatic change in its innovation system. It has evolved over two distinct time periods; namely, under central planning ( ) and after 30 years of reform (1978- present). During the planned economy period, neither market competition nor other operational efficiency-based criteria for organizational performance existed (Guan, Yam, Tang, & Lau, 2009). By the late 1970s, China had come to recognize the inefficiencies and low level of effectiveness of such a centrally planned economy in practice (Liu & White, 2001). The government therefore initiated a series of reforms to China s economy, and one objective of the reforms is to increase efficiency by replacing the central planning system of resource allocation with a free market mechanism (Xu, 2002). Significant achievements have been made in the last three decades. Central government has moved from a policy of control through direct administration of the economy to one based on macroeconomic management. Organizational boundaries have changed dramatically, and primary actors are more autonomous and functionally diversified (Liu & White, 2001). The characteristics of China s innovation system continue to emerge from the interactions between its regional innovation systems and other innovation systems. At the regional or provincial level, Chinese regional innovation systems are also in transition from a centrally planned regime to a market-driven system. Under the central planning regime, provincial and municipal governments are strictly under the control of the central governments. Since the open-door reform, provincial governments have gained autonomy for formulating economic and social development policies (Gu & Lundvall, 2006; Liu & White, 2001). Chinese regional innovation systems share several prominent features that are typically found in developing or transitional countries. A rapid increase in R&D and innovation activity in China is witnessed, particularly in the late 1990s and the early 21st century; many organizations are involved in innovation and R&D performance in regions undergoing transition (Li, 2009). The pace of reform and growth is uneven across regions, and the coastal region grew more rapidly than the mountain areas in the hinterland. Although the structure, dynamics and performance of China s innovation system have been dramatically altered over the last 30 years the regional markets still are not functioning well. These difficulties have been highlighted in many recent studies on the Chinese national and regional innovation systems (Bao, Chang, Sachs, & Woo, 2002; Brun, Combes, & Renard, 2002; Guan & Liu, 2005; Liu & White, 2001; Xu, 2002). Some observers of innovation systems consider them to be complex adaptive systems. Such systems are capable of changing themselves to adapt to a changing environment and changing the environment to suit themselves. Complex systems have a propensity to exhibit power law or scaling correlations between primary measures used to characterize them. These scaling correlations can be used to construct scale-independent indicators that are properly normalized for the sizes of the members of the system (Katz, 2000, 2006). Scale-independent indicators can provide insights into innovation system characteristics that are not observable using conventional measures. They can accurately inform decision makers how much groups of different sizes contribute to an innovation system. Katz (2006) used the European and Canadian innovation systems as examples to explore scaling correlations between GERD & GDP and GDP & population. The scale-independent indicators derived from these relationships were used to examine characteristics of the two systems. There are significant differences between the Chinese innovation system and these two systems. The European innovation system is a collection of 15 national systems that has been evolving into a supranational system (Katz, 2006; Schuch, 1998). The Canadian innovation system is composed of 10 provincial and two territorial systems that has evolved for nearly 135 years into a federal innovation system (Bührer & Ludewig, 2004; Katz, 2006). The Chinese innovation system is composed of 26 provinces and 5 municipalities. The Chinese innovation system is in transition from a planned economy to a market economy. The institutional context in which the Chinese innovation system has and is evolving is quit different than the institutional context in the Western countries. The World Bank (1994) demonstrates that, relative to other regions such as the US and European Union, market integration within China is far from realizing the benefits of its potentially large internal market (Xu, 2002). The Chinese innovation system is immature and less fully integrated than the European and Canadian innovation systems. Thus, one may wonder whether scale-independent indicators which are originally used to compare the equivalent mature and stable economies between EU and Canada, are suitable and applicable for such an emerging innovation system in China. In fact, this question has been partially answered by Gao and Guan (2009). Based on time series data, they use scale-independent indicators to explore the performance of the Chinese innovation system from an economic and from a science and technology point of view and compare it with 21 other nations. To go further, this paper argues that China s regional and national innovation systems exhibit scaling behaviors that emerge with time and exist at points in time. What are the similarity and difference between the Chinese, European and Canadian innovation systems when they are characterized using scale-independent indicators? What do these indicators tell us about

3 620 X. Gao et al. / Journal of Informetrics 4 (2010) the Chinese innovation system that is not available using conventional measures? Solving these issues is the main purpose of the empirical analyses in this paper. The remainder of the paper is organized as follows. Section 2 describes the data source and methodology. The empirical results are presented in Section 3. The conclusions and policy implications are summarized in the final section. 2. Data source and methodology Innovation systems are commonly characterized by primary measures such as GDP, GERD, papers, citations, population and patents. A variety of ratios derived from these measures such as national wealth (GDP per capita), R&D intensity (GERD/GDP) and scientific impact (citations/paper) are commonly used to rank members of a system. The rankings are used to inform decision makers about the performance of the members. However, scaling correlations can exist between the numerators and denominators of the ratios derived from a collection of primary measures used to describe a system (e.g. a collection of publication and citation measures for groups of researchers, institutions or nations). Unless the exponent of the scaling correlation is 1.0 (i.e. a perfectly linear correlation) the indicators need to be adjusted to account for the non-linear correlation that exists across the measures in the collection. Failure to properly normalize indicators results in rankings that do not accurately reflect member contributions to a system given their sizes (Katz, 2006). We will explore the Chinese regional and national innovation systems using scale-independent indicators. Before 1997 and not including Hong Kong, Taiwan, and Macao, China had 30 provinces and municipalities. Chongqing became the fourth municipality in Here, we focus on 30 provinces and municipalities excluding Xizang, 2 Hong Kong, Taiwan, and Macao. The primary measures we will use are GDP, GERD, population, scientific papers and patents. We collected data on the above measures between and 2005 for the provinces and municipalities from China Statistical Yearbook and China Statistical Yearbook on Science and Technology published by the National Bureau of Statistics of China (NBS). The economic data for the Chinese system are in a common currency (i.e. Yuan Renminbi). Here scientific papers refer to publications covered by the Science Citation Index (SCI). Frequently SCI-papers are regarded as a proxy measure of scientific output because it has reasonably comprehensive coverage of the significant contribution to science. National patents capture a sense of the degree to which a regional economy is developing and commercializing new-to-the-nation technologies (Guan & Liu, 2005). National patents include three subtypes: invention patents, utility patents, and design patents. Scaling relationships that exist between primary measures such as GDP & GERD, GDP & population and citations & papers (Katz, 1999, 2000, 2006) can be used to construct scale-independent indicators. Scale-independent indicators are derived from power law distributions and correlations that are normalized to account for scaling effects; they can be used to compare groups over a wide range of sizes (Katz, 2006). This paper focuses on only two types of scale-independent indicators: scaling factor indicators and relative magnitude indicators. A scaling factor indicator is derived from the exponents of a power law correlation or distribution. A relative magnitude indicator is derived from the ratio of observed values to values predicted from a power law correlation or distribution. 3. Results and discussion Unlike some physical processes, social activities are never completely deterministic nor are they completely random. Innovation processes and the systems in which they are embedded are complex and adaptive. Such systems are expected to exhibit scaling properties (Katz, 2006). In this section it will be shown that China s regional and national innovation systems exhibited scaling behaviors that emerge with time and exist at points in time Scaling correlations between GERD and GDP Fig. 1 plots the scaling correlation between GERD and GDP for Chinese innovation system from 1995 to 2005 on a log-log scale. The measured value of GERD GDP scaling factor is 1.67 ± 0.26 (R 2 = 0.84). The scaling correlation over time between GERD and GDP for the 30 provinces is quite good and clearly shows that the Chinese GERD is growing much faster than its GDP. During this same period the Chinese GERD tended to grow 3.51 ( ) times every time the GDP doubled (2 1.0 ). The Chinese GERD was growing very non-linearly with increasing GDP. Chinese provinces can be roughly divided into three geography regions, namely, Eastern region, Central region and Western region. Generally speaking, the most economically dynamic provinces are located in Eastern region, the moderate in Central region, and the poorest in Western region. Table 1 gives the GERD GDP scaling factors for the provinces in the Chinese innovation systems over the time period. The standard errors and the R 2 values indicate that the GERD GDP scaling factors look reasonable for most of the provinces. 23 of the 30 provinces had R 2 values greater than Further analysis indicates that the provinces with poor correlation are not as economically developed as those with higher R 2 values. On the other hand, the GERDs of all provinces in Eastern and Central regions grew faster than their respective GDPs while the opposite is true for most of the provinces in Western region. From a regional perspective the GERD GDP scaling 2 Xizang, i.e. Tibet, is excluded in this research due to its incomplete data. 3 The exceptions are the data of Chongqing and Sichuan, starting from the year of 1997.

4 X. Gao et al. / Journal of Informetrics 4 (2010) Fig. 1. Scaling correlation between GERD and GDP for Chinese innovation system from 1995 to factors for the Eastern and Central regions were greater than 1.0 and similar in magnitude while the values for Western region are much less than 1.0. The GERD in Western region is growing much slower than its GDP which may be one of reason why it is less developed than Eastern and Central regions. Fig. 2 is a log log plot of the year 2000 GERD and GDP values for 30 Chinese provinces. The year 2000 is halfway through the time interval under consideration. This figure shows that a scaling correlation existed between GERD and GDP across members of the Chinese system at a point in time. The systemic GERD GDP scaling factor is 1.21 telling us that there was a systemic tendency for the GERD to increase 2.31 times ( ) with a doubling in the size of the provincial economy. It is important to note that the systemic scaling factor is not determined by any individual member in the system. It is an emergent property of an innovation system that evolves from the complex interaction between its members and members of other systems (Katz, 2006). Table 1 Chinese GERD GDP scaling factors ( ). Province SE a R 2 Eastern region Fujian Shandong Zhejiang Guangxi Jiangsu Liaoning Guangdong Hebei Tianjin Shanghai Hainan Beijing Central region Anhui Hubei Heilongjiang Jiangxi Jilin Hunan Shanxi Henan Neimenggu b Western region Sichuan Guizhou Qinghai Shaanxi Ningxia Yunnan Gansu Xinjiang Chongqing a SE is the standard error for. b Also known as inner Mongolia.

5 622 X. Gao et al. / Journal of Informetrics 4 (2010) Fig. 2. Systemic GERD GDP scaling correlation in Fig. 3. For Chinese innovation system. (a) Value of the GERD GDP systemic scaling factor over time; (b) variance the GERD GDP systemic scaling factor over time. Fig. 3a and b contain plots of the values and the variances of the systemic GERD GDP scaling factors over the time period. 4 It increased rapidly over the decade and the variance in the scaling trend across the provinces decreased. In summary, the Chinese innovation system, like the European and Canadian innovation systems, exhibits scaling relationships between GERD and GDP over time and at points in time. 4 Provincial GERD data for the year 1997 are questionable bringing into question the 1997 scaling factor reported in Fig. 3b and variance reported in Fig. 4. The reported 1997 GERD values were 30 70% lower than values reported for 1996 and % less than the values report for 1998 leading the authors to question the official records. Since there is no way to verify these data using secondary sources the reported values were used in this analysis.

6 X. Gao et al. / Journal of Informetrics 4 (2010) Table 2 Comparison of Chinese R&D intensity (2000). Province GERD/GDP (%) Province RGERD a Beijing 6.28 Beijing 9.27 Shaanxi 2.98 Shaanxi 4.78 Shanghai 1.62 Tianjin 2.43 Tianjin 1.51 Shanghai 2.11 Sichuan 1.12 Sichuan 1.49 Guangdong 1.11 Ningxia 1.47 Liaoning 0.89 Gansu 1.32 Jiangsu 0.85 Guangdong 1.23 Hubei 0.81 Qinghai 1.16 Gansu 0.74 Jilin 1.16 Jilin 0.73 Liaoning 1.15 Anhui 0.66 Hubei 1.07 Chongqing 0.64 Chongqing 1.03 Ningxia 0.62 Shanxi 0.97 Shandong 0.61 Jiangsu 0.97 Shanxi 0.60 Anhui 0.93 Zhejiang 0.55 Guizhou 0.75 Fujian 0.54 Fujian 0.72 Hunan 0.52 Hunan 0.71 Hebei 0.52 Shandong 0.69 Qinghai 0.49 Zhejiang 0.68 Henan 0.48 Hebei 0.65 Heilongjiang 0.46 Heilongjiang 0.64 Guizhou 0.42 Jiangxi 0.63 Jiangxi 0.41 Guangxi 0.63 Guangxi 0.41 Henan 0.61 Yunnan 0.35 Yunnan 0.54 Neimenggu 0.24 Neimenggu 0.40 Xinjiang 0.24 Xinjiang 0.40 Hainan 0.16 Hainan 0.33 a RGERD is short for relative GERD; it is normalized for size Scale adjusted and conventional R&D intensity Table 2 gives the conventional R&D intensity indicator and the relative GERD indicator for the members of the Chinese innovation system in the year The relative GERD is the ratio between the observed GERD and the GERD predicted by the systemic scaling relationship between GERD and GDP. Provinces are listed in decreasing rank by each indicator. The conventional R&D intensity and the relative GERD indicators for provinces in the Chinese innovation system ranged from 0.16 to 6.28 and 0.33 to 9.27, respectively. The relative GERD indicator is adjusted for the systemic scaling correlation between provincial sizes measured with GDP and provincial expenditures on R&D. We see that 19 of 30 provinces have a different rank than the one given using the conventional indicator. For example, the less-developed province of Ningxia is ranked 14th by the R&D intensity indicator and 6th by the relative GERD indicator. The variance from the population mean of the relative GERDs for the Chinese innovation system was calculated for each year and the values are plotted in Fig. 4. The variance for provinces in the Chinese innovation system decreased smoothly from above 2.60 to about 1.0 except for the high value of about 5.0 in The variances of the relative GERD in the Chinese innovation system are much larger when compared to the European and Canadian innovation systems. This may be due to the time span over which the innovation systems have been evolving and the differences in their governance structure. Larger variance from the systemic scaling trend could indicate that the members of the Chinese innovation system are less tightly integrated than the European and Canadian innovation systems. This agrees with one of Xu s (2002) findings that although economic integration of the Chinese provinces has progressed under reform, it is by no means complete Scaling correlations between GDP & population Fig. 5 is a log log plot of GDP versus population (POP) for China. The measured GDP POP scaling factor is ± 2.90 (R 2 = 0.80). A doubling of the population would be expected to produce an increase in GDP by nearly 184,339 ( ) times. This number may seem large but considering how long it would take for the Chinese population to double the magnitude is quite reasonable. A scaling correlation between GDP and population guarantees there is a scaling correlation between GDP per capita and population (Katz, 2006). The Chinese National Wealth would be expected to increase 92,169 ( ) times every time the population doubles. Fig. 6 is a plot of the values of the systemic GDP POP scaling factors for the Chinese innovation system from 1995 to The magnitude of the scaling factor for the Chinese system is just less than 1.0 while the values for Europe were greater than or close to 1.0 and the values for Canada were always well above 1.0. There was a systemic tendency for Chinese provincial

7 624 X. Gao et al. / Journal of Informetrics 4 (2010) Fig. 4. Variance of the relative GERD for the Chinese innovation systems. Fig. 5. Scaling correlation between GDP and population for China from 1995 to GDP to grow slower than the provincial population and over the time the magnitude of GDP POP scaling factor decreased about 8%. Table 3 presents the GDP POP scaling factor for the power law correlation between GDP and population for the 30 provinces of China from 1995 to The standard errors and the R 2 values indicate that the GDP POP scaling factors look reasonable for most of the provinces. The scaling factors ranged from ± 2.93 for Shaanxi to 3.94 ± 0.71 for Guangdong with an average magnitude of Fig. 6. GDP population systemic scaling factor for China from 1995 to 2005.

8 X. Gao et al. / Journal of Informetrics 4 (2010) Table 3 Chinese GDP population scaling factors ( ). Province SE a R 2 Province SE a R 2 Shaanxi Tianjin Neimenggu Sichuan Liaoning Zhejiang Gansu Fujian Chongqing Qinghai Jiangxi Ningxia Shandong Anhui Heilongjiang Yunnan Hebei Hainan Hubei Guizhou Jilin Guangxi Jiangsu Beijing Hunan Xinjiang Henan Shanghai Shanxi Guangdong a SE is the standard error for Scaling correlations between GDP and outputs of science and technology Papers and patents are one of the main outputs of science and technology for an innovation system. Funding can influence not only the quantity of a research group but its quality or impact. A power law correlation can also exist between GDP and outputs of science and technology. Fig. 7a and b are log log plots of Chinese papers versus GDP and patents versus GDP, respectively. The standard errors and the R 2 values indicate that the power law correlations seen in the two figures are statistically significance. The paper GDP and patent GDP scaling factors are 1.83 and 1.48, respectively, both greater than 1.0. Over the 11-year period the Chinese papers and national patents grew faster than GDP. Papers and national patents tended to grow 3.55 (21.83) and 2.80 (21.48) Fig. 7. Scaling correlations between GDP and (a) paper; (b) patent for China from 1995 to 2005.

9 626 X. Gao et al. / Journal of Informetrics 4 (2010) Fig. 8. Paper GDP systemic scaling factor for China from 1995 to times, respectively, every time the GDP doubled. Chinese papers and national patents grew non-linearly with GDP and the growth rate of the Chinese papers was higher than that of national patents. Intensity measures used to compare nations are determined with respect to a variety of primary measures including population and GDP. For example, wealth intensity is measured by GDP per capita and national citation intensity is measured by citations per GDP (King, 2004). We define papers per GDP and patents per GDP as paper intensity and patent intensity, respectively. Using the laws of exponents and the scaling correlations given in the preceding paragraph we know that paper intensity and patent intensity scaling correlation between paper intensity and GDP and between patent intensity and GDP have scaling factors less than 1.0. Paper and patent intensities were expected to increase 1.77 ( ) and 1.40 ( ) times, respectively, each time GDP doubled. Figs. 8 and 9 are plots of the values of the systemic paper GDP and patent GDP scaling factors for the Chinese innovation system over the time period. Fig. 8 shows that the values of paper GDP scaling factors generally fluctuated around 1.40 between 1996 and The reason for the higher values at the beginning and near the end of the period may only become apparent within the context of a longer time series. Fig. 9 shows that the systemic patent GDP scaling factor is increasing from around 1.0 in 1995 to over 1.40 in 2004 for China. The mean value is lower than that for the paper GDP scaling factors shown in Fig. 8. Furthermore the scaling correlation between papers and patents was 1.23 ± 0.09 indicating that published papers are growing much faster than registered patents. For every doubling in patent registration papers are growing 2.35 times. Tables 4 and 5 give the paper GDP and patent GDP scaling factor for the 30 provinces in China between 1995 and 2005, respectively. Table 4 shows that the 29 provinces have R 2 values greater than 0.7 except for Ningxia. The paper GDP scaling factors in China rang from 3.09 ± 0.24 for Heilongjiang to Ningxia 0.33 ± 0.62 with an average magnitude of Similarly, the standard errors and the R 2 values in Table 5 indicate that the patent GDP scaling factors look reasonable for most provinces. The patent GDP scaling factors vary from a high of 2.31 ± 0.18 for Chongqing to a low of 0.36 ± 0.13 for Gansu. The average is 1.16 which is lower than that of the paper GDP scaling factor. Tables 4 and 5 also show that the reliability of the paper GDP and patent GDP scaling factors for some provinces are questionable. For example, the paper GDP scaling factor for Ningxia has a large standard error and a low R 2 value. This Fig. 9. Patent GDP systemic scaling factor for China from 1995 to 2005.

10 X. Gao et al. / Journal of Informetrics 4 (2010) Table 4 Chinese paper GDP scaling factors ( ). Province SE a R 2 Province SE a R 2 Heilongjiang Zhejiang Guangxi Anhui Guizhou Qinghai Hubei Shanxi Hainan Henan Hunan Guangdong Yunnan Shanghai Liaoning Tianjin Jiangxi Jilin Fujian Jiangsu Sichuan Shaanxi Xinjiang Beijing Shandong Neimenggu Chongqing Gansu Hebei Ningxia a SE is the standard error for. Table 5 Chinese patent GDP scaling factors ( ). Province SE a R 2 Province SE a R 2 Chongqing Henan Shanghai Jiangxi Hubei Heilongjiang Jiangsu Yunnan Zhejiang Xinjiang Tianjin Beijing Guangdong Ningxia Sichuan Hebei Fujian Guangxi Liaoning Hainan Shandong Shanxi Anhui Shaanxi Hunan Neimenggu Guizhou Qinghai Jilin Gansu a SE is the standard error for. occurred because while the provincial GDP exhibited exponential growth the papers had both positive and negative growth period. 4. Conclusions and implications An innovation system is an open complex system composed of many actors and the interactions between them. They have a propensity to exhibit scaling properties such as the Matthew effect. Using these properties scale-independent indicators are constructed which account for such things as the non-linear effect of size. Scale-independent indicators provide insights into innovation systems and their activities that are not available using conventional measures. This paper used scaleindependent indicators to explore the Chinese innovation system during the transition process at the national and regional levels. Similar to the European and Canadian innovation system, the Chinese innovation system exhibits scaling relationships between GERD and GDP over time and at points in time. The GERD GDP scaling factor tells us that over the observed period the Chinese GERD exhibited non-linear tendency to increase with GDP. It suggests that the target to increase R&D expenditure to above 2.5% of GDP by 2020 may be realized if China continues to sustain faster economic growth. On the other hand, the systemic GERD GDP scaling factor indicates that the GERD of the Western region is growing far slower than its GDP compared with Eastern and Central regions. The Western region needs investment in research infrastructure and funding if it wishes to catch up with developments in the other regions. There is a strong non-linear relationship between GDP and population. The population of China has been growing for years and one might expect the wealth intensity or GDP per capita to be increasing rapidly too. From the wide range of GDP POP scaling factors for the 30 provinces we know there are large differences in the growth of provincial wealth with respect to the growth in population. Regional development in China is quite inequitable just as it is for European countries and Canadian provinces. We demonstrated that a power law correlation exists between the papers & GDP and patents & GDP. In both cases the scaling factor was greater than 1.0 showing Chinese SCI-papers and national patents grew non-linearly with increasing GDP

11 628 X. Gao et al. / Journal of Informetrics 4 (2010) indicating that these outputs of science and technology are increasing faster than economic growth. Also, SCI-papers are growing faster and patents slower than the growth of GERD relative to the growth in GDP. On the other hand, the paper GDP and patent GDP scaling factors for the 30 provinces indicate that GDP has a much different non-linear effect on the outputs of science and technology across the provinces. Therefore, to evaluate the productivity of science and technology between the provinces, we should not only focus on the quantity of papers (patents), but also attach importance to these non-linear effects on outputs by GDP. What policy implications can be drawn about the national and regional innovation systems of China? First, China is an emerging innovation system, which functions at a different scale than Europe and Canada. Scale-independent indicators facilitate comparisons across a wide spectrum of scales. We suggest they are more reliable for this sort of comparison than many conventional measures. Second, policy makers should take account of scaling properties between economic indicators and other measures when planning regional programs for science and technology development. In particular, for less favored regions, scaling properties such as the Matthew effect indicates that they have great potential for catching up with advanced regions. Acknowledgements This research is funded by the National Natural Science Foundation of China (Project No ) and the Program of Higher-level talents of Inner Mongolia University of China (Project No. Z ). The authors are grateful for the valuable comments and anonymous reviewers and the editors, which significantly improved the article. The authors, naturally, bear responsibility for any remaining shortcomings. References Bao, S. M., Chang, G. H., Sachs, J. D., & Woo, W. T. (2002). Geographic factors and China s regional development under market reforms, China Economic Review, 13, Brun, J. F., Combes, J. L., & Renard, M. F. (2002). Are there spillover effects between coastal and noncoastal regions in China? China Economic Review, 13, Buesa, M. (2002). El sistema regional de innovación de la Comunidad de Madrid. Documento de trabajo no. 30. Madrid: Instituto de Análisis Industrial y Financiero, Universidad Complutense. Buesa, M., Heijs, J., Pellitero, M. M., & Baumert, T. (2006). Regional systems of innovation and the knowledge production function: The Spanish case. Technovation, 26, Bührer, S., & Ludewig, N. (2004). A comparative guide to multi actors and multi measures programmes (MAPS) in RTDI policy. Wien StarMAP project. Vienna: Technologie Impulse Gesellschaft. Cooke, P. (2000). Business processes in regional innovation systems in the European union. In Z. Acs (pp ). Cooke, P. (2001). Regional innovation systems, clusters and knowledge economy. Industrial and Corporate Change, 10(4), Cooke, P., Gómez Uranga, M., & Etxebarría, G. (1997). Regional systems of innovation: Institutional and organisational dimensions. Research Policy, 26, Edquist, C. (1993). Systems of innovation A conceptual discussion and a research agenda. In Paper presented at workshop no. 3. Globalisation versus National or local systems of innovation Strassbourg, March 1993, Freeman, C. H. (1987). Technology policy and economic performance: Lessons from Japan. London: Pinter. Gao, X., & Guan, J. C. (2009). A scale-independent analysis of the performance of the Chinese innovation system. Journal of Informetrics, 3, Gu, S., & Lundvall, B. A. (2006). China s innovation system and the move toward harmonious growth and endogenous innovation. Innovation: Management, Policy & Practice, 8(1/2), Guan, J. C., & Liu, S. Z. (2005). Comparing regional innovative capacities of PR China-based on data analysis of the national patents. International Journal of Technology Management, 32, Guan, J. C., Yam, R. C. M., Tang, E. P. Y., & Lau, A. K. W. (2009). Innovation strategy and performance during economic transition: Evidences in Beijing, China. Research Policy, 38, Hilpert, U. (1991). Regional innovation and decentralization: High-tech industry and government policy. London and New York: Routledge. Katz, J. S. (1999). The self-similar science system. Research Policy, 28, Katz, J. S. (2000). Scale independent indicators and research evaluation. Science and Public Policy, 27, Katz, J. S. (2006). Indicators for complex innovation systems. Research Policy, 35, King, D. A. (2004). The scientific impact of nations. Nature, 430, Li, X. B. (2009). China s regional innovation capacity in transition: An empirical approach. Research Policy, 38, Liu, X. L., & White, S. (2001). Comparing innovation system: A framework and application to China s transitional context. Research Policy, 30, Lundvall, B. A. (1992). National systems of innovation: Towards a theory of innovation and interactive learning. London: Pinter. Nelson, R. R. (Ed.). (1993). National innovation systems: A comparative study. New York: Oxford University Press. Schuch, K. (1998). The emergence of the European innovation system and its impact on the Austrian S&T system. In Proceedings from the 38th Congress of the European Regional Science Association 28 August 1 September, Vienna. Tuzi, F. (2005). The scientific specialization of the Italian regions. Scientometrics, 62(1), World Bank. (1994). China: Internal market development and regulation. Washington, DC: World Bank. Xu, X. P. (2002). Have the Chinese provinces become integrated under reform? China Economic Review, 13,