Knowledge-based Development in Leading Regions across the Globe: An Exploratory Analysis of the co-Evolution of Resources,Capabilities and Outputs

Abstract

It is widely observed that the global geography of innovation is rapidly evolving. This paper presents evidence concerning the contemporary evolution of the globe’s most productive regions. The paper uncovers the underlying structure and co-evolution of knowledge-based resources, capabilities and outputs across these regions. The analysis identifies two key trends by which the economic evolution and growth patterns of these regions are differentiated—namely, knowledge-based growth and labour market growth. The knowledge-based growth factor represents the underlying commonality found between the growth of economic output, earnings and a range of knowledge-based resources. The labour market growth factor represents the capability of regions to draw on their human capital. Overall, spectacular knowledge-based growth of leading Chinese regions is evident, highlighting a continued shift of knowledge-based resources to Asia. It is concluded that regional growth in knowledge production investment and the capacity to draw on regional human capital reserves are neither necessarily traded-off nor complementary to each other.

Introduction

It is widely observed that the location where innovation occurs is evolving, with the stock of knowledge and other knowledge-based resources constantly shifting, reflecting ever-changing contexts for new and more advanced knowledge requirements (Dicken, 2007). Furthermore, the sources of regional productivity and growth are increasingly based on the role that knowledge plays within and across regional economies (Capello and Nijkamp, 2009). As a result, the concept of the knowledge-based economy has emerged to aid a better understanding of how the effective production, distribution and use of knowledge underpin innovative and competitive modern economies (Huggins and Izushi, 2007). The development of the regional innovation systems literature, in particular, is recognition of the role of knowledge for growth through innovation (Cooke, 2004). Innovation systems theory views an economy as an interlinked systemic network of components facilitating innovation (Freeman, 1987; Lundvall, 1992).

In an evolutionary context, the knowledge-based development of a regional economy involves multiple threads of relationships among its actors and resources at both a firm and spatial level, which interact in a complex manner (Maskell and Malmberg, 2007). For instance, under growing competitiveness pressures in virtually all sectors, firms are increasingly focusing on their core activities and searching external knowledge sources as part of their innovation management strategies. These firm-level strategies may facilitate knowledge-based investment and stimulate the growth of related resources and capabilities within their region, resulting in productivity improvement at both firm and regional level. Yet, there are growing concerns that such knowledge-based development may not necessarily contribute to employment growth at the regional level and may in fact lead to ‘jobless growth’ (Vivarelli and Pianta, 2000; Döpke, 2001). Indeed, the evolution of a regional economy and its innovation capacity may involve multiple independent trajectories, through which different sets of resources and capabilities evolve together. However, there is a dearth of evidence concerning how these trends are occurring across the globe.

In this paper, we seek to present evidence concerning the recent development of the globe’s most productive regions from the viewpoint of their innovation systems and knowledge-based economies. Our aim is to uncover the underlying structure of the changes in knowledge-based resources, capabilities and outputs across regions, and offer an analysis of these regions according to an uncovered set of key trends.

A number of theories, including agglomeration, industrial districts and clusters, as well as innovation systems, attempt to explain the way in which productive resources are combined and productivity is enhanced within a regional economy. In general, these concepts suggest a number of factors, such as industry and industry structure, firm type and the geographical range of external economies, to explain regional economic evolution, with them all highlighting the multiple trajectories that regions may follow in their development (Markusen, 1996; Phelps and Ozawa, 2003; Iammarino and McCann, 2006). Yet, while a large number of extant empirical studies resting on these theories are informative and useful in their own right, they are not always without shortcomings in light of the aim of uncovering an underlying structure of economic evolution and development across regions. For instance, a significant number of studies are designed to account for a specific dependent variable, such as gross domestic product (GDP) as a measure of economic output, or patents as a measure of innovation. This is typically achieved through the identification of a linear combination of other variables which maximises the proportion of the variance of the dependent variable they seek to explain. However, such accounting is different in its purpose and result from approaches which seek to identify a set of commonalities across variables. When variables represent the growth of resources, capabilities and outputs of regional innovation systems and economies, the commonalities identified across these variables may suggest processes of co-evolution across regions.

A further limitation of many extant studies is a lack of a common framework and dataset that is applicable to an analysis of regions across the globe. Empirical findings on innovation systems are typically generated initially on the basis of regions selected from a particular country or continental bloc, rather than a fuller global coverage, resulting in a lack of harmony in terms of the frameworks employed, or the capability to provide regional comparisons that uncover regional trends across the globe (Doloreux and Parto, 2005). This paper seeks to go some way to overcoming this particular shortcoming by examining with a single analytical framework the globe’s most productive regions, the coverage of which is wider than most previous studies.

Our analysis, which is exploratory in nature, aims to summarise in a concise but accurate manner the co-evolution of knowledge-based resources, capabilities and outputs, represented by their change across the globe’s most productive regions in the period 2001/02 to 2004/05. We seek to measure the position of regions in terms of key trends and identify groups of regions that show similar changes, establishing commonalities and differences in the routes underlying contemporary development in leading regions.

The paper is structured as follows: in the following section we review some of the key concepts relating to regional knowledge-based development and introduce the analytical framework adopted in this study. After outlining our data and methodology, we present the key results. We first offer descriptive statistics of each of the indicators and then present the key growth trends identified and discuss groups of regions delineated by these trends. In a concluding section, we provide our interpretations of the results in relation to the globalisation of knowledge-based development and their implications for policy-makers.

Knowledge-based Regional Development

Analysing Knowledge-based Regional Development

Endogenous growth theory has placed knowledge at the centre of economic development (Romer, 1986, 1990). The use of the term ‘endogenous’ is recognition that economic growth is influenced by investment into knowledge production of resources generated by the economy itself, rather than a view of knowledge as a factor derived outside the economy, which is associated with traditional Solow-type growth models (Solow, 1957). Theorists of economic development have increasingly drawn upon models of endogenous growth to reach a better understanding of the factors underpinning such development (Izushi, 2008).

Whilst endogenous growth can be considered the desired outcome of knowledge-based development and innovation, it is the process of endogenous development which underpins the growth trajectories of economies (Vázquez-Barquero, 2007). In particular, regions are increasingly considered to be key territorial units within which endogenous forms of development flourish through their innovative milieu. Cooke (2004), for instance, suggests that regional innovation systems are a vital component for regional economic development, while others have focused on the notion of clusters as the key focus of regional economic theory and policy, with the underlying tenet being that competitiveness is determined by the strength of key concentrations of specific industries (Porter, 1998; Huggins and Izushi, 2011).

While views on the prominence of knowledge for regional economic development remain contested (Lagendijk and Cornford, 2000; MacKinnon et al., 2002), one of the outcomes of both theoretical and policy developments in this area is a view of regions as innovation systems in which various system components, such as universities, government research institutes and venture capitalists, are linked with firms as key sources of the knowledge utilisable in the pursuit of economic growth (Huggins et al., 2008). The ability to ensure that knowledge, generated internally or externally, adds economic value in a region depends on how firms and other organisations in the region are linked through business, academic and social networks (Mahroum et al., 2008). In this sense, the region itself acts as an organisational form of co-ordination facilitating knowledge flow and new knowledge creation (Vázquez-Barquero, 2007).

Based on the view of a region as an interlinked system, innovation systems theory proposes the co-evolution of its components (Freeman, 2002; Borrás, 2004; Heidenreich, 2004). However, there is a dearth of evidence about key dimensions through which a region’s resources, capabilities and outputs evolve together. Doloreux and Parto (2005) find that studies of regional innovation systems are typically designed either to offer ‘snapshots’ of individual regional innovation systems, or to specify desirable factors and mechanisms for promoting innovation through a comparison of regions. Surveying approximately 200 studies of regional innovation systems, Carlsson (2007) finds that slightly more than half are empirically oriented, focusing mostly on a particular region or multiple regions, with more than half of the empirical studies focusing solely on regions within Europe. A significant proportion of early studies are qualitative case studies as they aim to provide a narrative on the intangible dimension of the knowledge circulation and learning (Doloreux and Parto, 2005).

More recently, a growing body of literature takes a quantitative approach, such as a knowledge production function model, to investigate regional innovation systems (Fritsch, 2002; Bilbao-Osorio and Rodríguez-Pose, 2004; Buesa et al., 2006; Crescenzi et al., 2007; Rodríguez-Pose and Crescenzi, 2008; Buesa et al., 2010). The aim of these studies is to account for a specific variable selected a priori, such as patents or GDP, and to examine the effects of variables representing innovation system characteristics. In these studies, variables are often measured only at one point in time, thus failing to show their changes over time. Such analyses are different in their aims and findings from approaches seeking to identify the co-evolution of the key components of regional economic systems. An exception to this is a recent study of four selected knowledge-based sectors across German regions by Buerger et al. (2012), who investigate the co-evolution of patents, R&D and employment, analysing panel data for these three variables with a vector autoregression model. The study finds that past innovation growth, as measured by patents, is associated with the subsequent growth of employment in certain sectors, suggesting that, in these sectors: the employment gains of innovators are greater than the losses of non-innovators; and, the labour-saving effects of process innovations are compensated for by other positive effects resulting from these innovations. However, the study focuses on a relatively small number of variables in selected knowledge-based sectors, leaving their relationships with other less knowledge-intensive sectors unexplored. Furthermore, the study concerns regions in a single nation. This latter point of limited regional coverage is shared by most qualitative case studies examining a single or only a few regions, as well as econometric studies which examine regions in a single nation (Buesa et al., 2006), in Europe (Fritsch, 2002; Bilbao-Osorio and Rodríguez-Pose, 2004; Rodríguez-Pose and Crescenzi, 2008; Buesa et al., 2010) or in Europe and the US (Crescenzi et al., 2007).

Therefore, there is a gap in the literature in terms of identifying the co-evolution of regional economic system components at a global level. In order to achieve this, the analytical framework presented here seeks to combine theories of knowledge production and innovation systems (Fritsch, 2002; Cooke, 2004; Buesa et al., 2006; Buesa et al., 2010) with the wider economic systems underlying the production of goods and services, allowing us to examine interrelationships among these components at a regional level. In particular, the impact of knowledge-based sectors upon the rest of their regional economy is likely to be influenced by how closely the former is tied to more traditional, less knowledge-intensive economic activity within the region through knowledge spillovers, as well as through input–output linkages and the multiplier effects of demand creation (Adams, 1990; Rodríguez-Pose and Crescenzi, 2008). Through the development of an analytical framework, outlined next, the co-evolution of key components of regional knowledge production and the more general economic conditions underlying the production of goods and services across knowledge-intensive and less knowledge-intensive sectors are captured. This facilitates a better understanding of the connections between knowledge-based development and the fundamentals of economic development, such as job creation, at a regional level (Vivarelli and Pianta, 2000; Döpke, 2001).

Analytical Framework Employed

In order to empirically analyse knowledge-based development at a regional level, in this section we develop an analytical framework and explain the rationale underlying its conceptualisation. As shown in Figure 1, the framework consists of two domains, both of which are underpinned by the stock of knowledge or technology. One domain (the top half in the figure) represents the production of goods and services. A standard set of production factors—physical capital, labour and human capital—is combined with technology. The other domain (the bottom half) represents the production of new knowledge, which in turn gives rise to productivity growth. Key activities in this domain are R&D within business firms and research within the science base (i.e. universities and government laboratories). Unlike the production of goods and services, the factors of knowledge production consist of physical capital and human capital (as the hand–eye co-ordination of labour by definition does not count here).

Figure 1.

Variables employed by the framework.

The activities in the two domains are linked to one another through various feedback loops (Kline and Rosenberg, 1986). A growing body of studies, particularly studies of the innovation system, identify various linking mechanisms which influence how individuals and firms relate to each other and learn. Lastly, part of the value created in the production of goods and services is invested in the factors in both domains. Such investment is crucial to the long-term sustainability of productivity.

The framework employs a set of 17 indicators (which are presented in Figure 1). In the domain of new knowledge production, we choose the number of employees in five high-tech sectors as proxies for the human capital devoted to innovation. A number of studies attempt to classify sectors based on criteria related to research and technology intensity (Lee and Has, 1996). We use the Eurostat (the European Commission’s statistical office) scheme of industry classification, which was developed in collaboration with the OECD (Hatzichronoglou, 1997; Laafia, 1999). We classify into five groups those sectors considered by Eurostat as ‘higher-tech manufacturing sectors’, ‘medium–high-tech manufacturing sectors’ and ‘high-tech service sectors’. The five groups are: IT and computer manufacturing; high-tech services; biotechnology and chemicals; instrumentation and electrical machinery; and automotive and mechanical engineering.

Other technology-input measures include R&D expenditures performed by the business and government sectors. Compared with corporate R&D, the impact of public-sector R&D is less direct in its route in terms of both diffusion and timing. Nonetheless, there is evidence that spillovers from public-sector R&D raise an economy’s productivity (Jaffe, 1989; Adams, 1990). Due to issues of data availability, R&D expenditures performed by the government sector are used as an indicator of public-sector R&D.

As for technology-output measures, we use the number of patents granted. The propensity to patent is known to vary widely across industries, with many patents turning out to be worthless, while a few are extremely valuable (Pavitt, 1982). Yet, patent statistics are the most widely available data of research outputs (Griliches, 1990). Furthermore, there is some evidence that suggests a close association between patents and other productivity-based measures at the national and regional levels (Fagerberg, 1996; Acs et al., 2002). With regard to linking mechanisms, we use private equity investment capital as a proxy of the availability of funds for knowledge-based, start-up firms. Private equity funding is often concentrated in small or medium-sized firms, including venture capital and start-up investments, which tend to be in knowledge-based activities.

In the production of goods and services domain, we include the employment rate (defined as one minus the unemployment rate) and the economic activity rate (defined by the ratio of the labour force to the working-age population) in our analysis. While these indicators show an economy’s capacity to draw and develop a greater amount of human capital out of its population, higher rates also tend to suppress average productivity levels, with there being an increasing proportion of low-wage, low-productivity jobs in places with relatively high activity rates. Furthermore, in the face of the global challenge from low-cost producers, those industries where productivity remains low compete by primarily using low-cost and low-skilled labour (Dunford, 2005). This phenomenon is particularly evident in those countries with deregulated labour markets (such as the US), which generally show a greater capacity to use a larger part of the labour pool (Crouch et al., 1999).

We also include the number of managers as a proxy of human capital. Although this is hardly a perfect indicator of human capital, a similar indicator is used in international studies of the labour market (OECD, 1994). The wages of managers are generally higher than those of other occupations, reflecting the greater amount of investment made in education and training.

For indicators of the long-term sustainability of productivity, we include public expenditures on primary and secondary education and higher education. There is a sequential interaction between a region’s education and training system and its stock of high-skilled workers. The rate of enrolment in education is influenced by a region’s employment and career prospects, as well as the socioeconomic background of pupils and the quality of schooling. Enrolment, in turn, determines the region’s workforce skills, productivity and economic performance (Bradley and Taylor, 1996). Public investment in education plays an important role in this sequential cycle, particularly improving the quality of local schooling over time.

Lastly, we adopt labour productivity, gross domestic product (GDP) and mean gross monthly earnings as indicators of regional economic performance.

To remove effects of the size of each region analysed, we take per capita figures for the following variables: GDP, R&D expenditures performed by business sector and government sector, patents granted, private equity investment capital, and public expenditures on primary and secondary education, and higher education. Employment in the five high- or medium-high-tech industries and number of managers are based on a per total regional employment basis.

Data Collection and Analysis

The data utilised in the analysis are drawn from 145 regions across the globe. The regions are sampled on the basis of a relatively high level, or fast growth, of GDP per capita. Of the 145 regions contained in the dataset there are 63 representatives from North America, 54 from Europe and 28 from Asia and Oceania.¹ Of the North American regions, 57 are US and 6 Canadian. Among the Asian and Oceania regions, there are 7 Chinese regions (Hong Kong and Taiwan are also included separately), 9 Japanese regions, 3 Australian regions and 3 Indian regions. For the US, regions are based on the metropolitan statistical areas (MSAs) classification. As defined by the US Census Bureau, MSAs consist of an area with a substantial population centre and adjacent counties having a high degree of economic homogeneity, which is, compared with counties, cities and states, more robust for economic analysis, as they reflect the boundaries of clusters of firms in related industries. The European regions are based on the European Union’s definition of regional units, NUTS-1.² The Canadian regions are based on their defined provincial units. The Asian and Oceanic regions consist of prefectures in Japan and are defined by city or provincial boundaries for most other nations (for example, provinces for China), as well as the inclusion of Taiwan, Singapore and New Zealand as region-states. Under constraints imposed by data availability, these units of analysis provide as high a degree of consistency as possible across the nations under analysis. For each of the 17 indicators discussed earlier and presented in Figure 1, we measure their annual growth rates for the period 2001/02 to 2004/05, which are the most recent periods with coverage across all regions at the time of undertaking the analysis.

The annual growth rates of the 17 indicators may overlap with one another and, to identify such overlapping across the variables, we adopt factor analysis in our methods. Factor analysis reduces an original set of variables into a smaller number of composite variables called ‘factors’. Each factor is a latent dimension underlying the original set of variables, presented as a condensed statement of the relationships between them. Furthermore, the position of a region in each dimension is given as a score called a factor score. The factors identified are orthogonal with one another and accordingly factor scores are uncorrelated across factors. With the use of the Anderson–Rubin method of estimation, factor scores have a mean of zero and a standard deviation of one. As for the extraction of factors and rotation of a factor matrix, we use the maximum likelihood method and the varimax method respectively. The maximum likelihood method provides the best statistical procedure as a goodness-of-fit test of the factor model (Gorsuch, 1983).

Due to a lack of data for some regions across the 17 indicators, Indian regions and a number of other regions are excluded from this analysis. Similarly, because of data gaps for the private equity indicator, we drop the growth rate for this variable after confirming its close association with the growth rates of GDP, labour productivity and business R&D. Furthermore, three variables—the growth rates for labour productivity, public expenditures on primary and secondary education, and employment in automotive and mechanical engineering—are dropped as they are found to conform to Heywood case variables.³ After these variables are removed, a reliable set of factors is obtained using the Schwarz (1978) method to determine the number of factors. A goodness-of-fit test of the factor model obtains the chi-squared value of 171.20 and the significance value of 0.00, showing a highly satisfactory level. Utilising the factor score of each region, cluster analysis is then undertaken to establish a grouping of regions. Cluster analysis is the most commonly used technique for identifying groups of homogeneous objects within the population. This is achieved by maximising the homogeneity of objects within the clusters while also maximising the heterogeneity between the clusters. Similarity between regions is defined by the Euclidian distance using their factors scores as co-ordinates. For a hierarchical clustering algorithm the average linkage procedure is adopted. The number of clusters is decided by examining changes in the agglomerative coefficients (Hair et al., 2010).

Results

Descriptive Statistics

The average growth rates for the 17 indicators by nation or continental bloc—the US, Europe, Canada, Australia, Japan, China and India—are shown in Table 1. In terms of overall economic performance, the growth performance of Chinese regions is spectacular, which is in stark contrast to regions in North America, Europe, Australia and Japan. The top seven fast-growth regions in terms of GDP per capita are all Chinese, followed by Hyderabad and Bangalore in India. The slowest-growing regions are predominantly US, Italian and Canadian. It is clear that new wealth generation is increasingly concentrated in leading Chinese regions, although it must be recognised that they are growing from a relatively small base level.

Table 1.

Average annual growth rates of 17 variables by region 2001/02 to 2004/05 (percentage)

	US	Europe	Canada	Australia	Japan	China	India
Gross domestic product	3.5	3.8	3.2	6.2	4.2	46.0	7.8
Labour productivity	3.0	2.9	4.2	4.8	6.4	44.7	7.8
Mean gross monthly earnings	−0.8	6.4	7.3	1.2	12.5	32.2
Economic activity rate	3.9	0.1	0.0	0.6	6.6	1.5
Employment rate	0.2	−0.4	0.2	0.4	−0.1	0.0	−0.4
Number of managers	−5.8	11.8	0.2	−4.1	−4.0	13.1
Patents registered	7.1	3.9	11.1	106.0	20.5	17.3
R&D expenditures performed by government	6.0	3.8	25.1	−7.1	3.4	19.9
R&D expenditures performed by business	3.9	4.7	19.5	17.0	8.7	48.1
Employment in instrumentation and electricalmachinery	−2.1	−2.9	16.6	20.3	−3.2	−15.9
Employment in automotive and mechanicalengineering	−2.0	−1.5	−9.6	−0.4	−0.4	−8.0
Employment in biotech and chemicals	−4.3	−2.6	1.8	−27.3	−0.2	−14.7
Employment in IT and computermanufacturing	−7.5	−6.1	−31.3	−30.8	−7.2	29.5
Employment in high-tech services	−3.7	0.1	−6.6	2.5	−15.6	−1.7
Public expenditures on primary and secondaryeducation	5.1	5.8	5.6	12.0	1.9	33.3	−9.4
Public expenditures on higher education	−0.7	−3.3	29.2	−1.5	6.5	24.8	6.1
Private equity funding	−11.9	0.8	20.3			143.6

Labour productivity growth has also been most pronounced among the Chinese and Indian regions, with Beijing and Guangdong recording the highest levels of growth. With some notable exceptions—Virginia Beach, Greensboro and Alberta—labour productivity growth in North American regions has generally been below average. In Europe, labour productivity growth has shown the greatest growth in Bratislava and Prague. Similarly, Chinese regions have shown the greatest growth in gross monthly earnings, led by Tianjin, Shandong and Beijing. The majority of regions showing the lowest growth in earnings are those located in the US. Many of the relatively low-growth regions actually recorded a decline in earnings levels.

In terms of labour market characteristics, economic activity rates have grown across the majority of regions, indicating the greater engagement of citizens within regional labour markets. Contrary to a number of other indicators, many US regions increased economic activity, with the greatest growth in economic activity recorded in Riverside, San Diego and Miami. As for employment rates, regions in the US, Australia and Canada made relatively good progress with regard to growth. By contrast, European and Japanese regions have generally witnessed negative growth. The biggest rise in unemployment has mainly concerned German regions. As for the proportion of managers, those regions generally showing the highest growth in this measure are predominantly Chinese and European (particularly Italian). The biggest fall in the proportion of managers has occurred in the US, which may be an impact of the growth in offshoring.

Growth rates in patent registrations show that Australian regions have seen the greatest growth. European regions have generally suffered the biggest declines. With the exception of one US region, all those regions showing a decline in patent registrations are European. In terms of R&D expenditures, those regions in receipt of the biggest increases in R&D expenditures performed by government are regions in Canada and China. Conversely, a number of US and European (especially Italian) regions have seen a dilution of government R&D investment. As for R&D investment by businesses, Chinese regions have seen by far the strongest growth, with the biggest growth occurring in Shandong, Tianjin and Zhejing. The Canadian and Australian regions have also shown relatively good growth, particularly in comparison with US and European regions. Once again, a significant number of US and European (particularly Italian) regions have seen a drop in business R&D investment. It is also worthy to note that, unlike GDP and earnings, which still remain at relatively low levels, Chinese regions already achieve relatively high levels of business R&D activity.

As indicated by Table 1, in a number of cases there is significant volatility in employment change across the so-called knowledge-based sectors, with some of the key changes being

(1) Instrumentation and electrical machinery: the majority of regions have seen a decline rather than growth in these sectors.

(2) Automotive and mechanical engineering: overall, the data strongly suggest that leading regions around the globe are becoming increasingly less dependent on the automotive sector, with regions, on average, seeing a relative decline in the proportion of employment.

(3) Biotechnology and chemicals: growth in biotechnology and chemicals employment as a proportion of total employment is relatively well spread across continental blocs.

(4) IT and computer manufacturing: there appears to be a trade-off relationship between IT/computer manufacturing and automotive/mechanical engineering in their growth performance, with those regions experiencing high growth in IT/computer manufacturing often suffering a loss of employment in automotive and mechanical engineering. Regions in China gained most, with only 21 regions showing positive growth rates, which gives a strong indication of the extent to which employment has become increasingly concentrated in key locations.

(5) High-tech services: the general pattern across all regions of low or negative growth is, perhaps, somewhat surprising, as one might expect stronger growth resulting from the increased deindust- rialisation of advanced regional economies. However, the impact of both the dotcom crash and growth in the offshoring of high-technology services in many advanced regions appears to have dampened employment growth.

In terms of growth in public expenditures on primary and secondary education, the majority of the analysed regions recorded an increase, led by the Chinese, UK and Australian regions. By contrast, regions in the US, Europe and Australia have seen a marginal fall in the expenditures on higher education, with those regions experiencing the biggest falls in expenditure being predominantly European, and, more particularly, German.

Finally, although private equity data are not available for all regions, for those regions where data exist it is clear that the biggest growth has occurred in the Chinese regions. The Canadian and Dutch regions have also shown relatively high growth rates. In the US, the traditional heartland of venture capital, most regions have seen a fall in levels of such capital investment.

Factor Analysis

Utilising the results of the factor analysis, the original variables are reduced to two factors, each of which represents a unique combination of the original variables. Table 2 shows the factor loadings obtained from the analysis. Given the fact that more than 100 regions are included in the analysis, we focus on those variables with a loading greater than ±0.40 (Gorsuch, 1983, p. 209).

Table 2.

Factor loadings of the variables

	1 (Knowledge-basedgrowth)	2 (Labour marketgrowth)
Gross domestic product	0.82	0.00
Mean gross monthly earnings	0.83	−0.28
Economic activity rate	−0.14	0.37
Employment rate	−0.01	0.82
Number of managers	0.15	−0.10
Patents registered	0.15	0.33
R&D expenditures performed by government	0.41	0.04
R&D expenditures performed by business	0.84	0.16
Employment in instrumentation and electricalmachinery	−0.13	0.19
Employment in biotechnology and chemicals	−0.27	−0.14
Employment in IT and computer manufacturing	0.29	−0.11
Employment in high-tech services	−0.05	−0.11
Public expenditures on higher education	0.64	0.37

Note: Factor loadings greater than 0.40 are shown in bold.

Factor 1 indicates a close association between the growth of GDP, gross monthly earnings and business R&D; growth in public expenditures on higher education and government R&D are also significantly loaded on this factor. We label this factor ‘knowledge-based growth’ as it indicates a close association between overall regional economic growth and growth in key knowledge-based investments, and to some extent, confirms the relationships we propose in our overall analytical model. Of course, our analysis cannot be said to infer causality, but at an exploratory level it does suggest that across this cohort of leading regions in both established and emerging economies the trajectory of economic growth generally moves in the same direction as growth in investment in R&D and higher education, chiming with the propositions of innovation systems theory.

The second factor mainly represents growth in employment rates, with growth in economic activity rates and higher education expenditures also marginally loaded (although slightly below the 0.40 threshold) and we label this factor ‘labour market growth’. This is particularly relevant in the case of leading regions, which despite their relative success often possess dual economy characteristics, whereby significant population pockets remain economically inactive and outside the regional labour market. Interestingly, it is important to note that employment growth in any of the five knowledge-based sectors is not significantly loaded on either the knowledge-based or labour market growth factor. This is likely to suggest that the regions tend to specialise sectorally and none of the sectors plays a central role when examining the contemporary dynamics of regional development at a global level. In a similar vein, the growth of patents and managers, as proxies of new knowledge and human capital respectively, are not loaded strongly on either of the two factors, showing that changes in these indicators are to a great extent independent of the trends represented by the factors.

Cluster Analysis

Following the factor analysis, cluster analysis was undertaken to identify groups of regions. Regions within a group are close to one another according to the distance defined by their factor scores. Figure 2 shows the identified 13 groups of regions, with a clear indication that Chinese regions are set apart from the rest of the regions in terms of their extremely high values for the knowledge-based growth factor. Although there are a few hot-spots of knowledge-based growth in North America (such as Alberta and Saskatchewan), the great majority of fast knowledge-based growth regions are located in the emerging part of the East Asian economy. However, forming four groups on their own, the Chinese regions also show significant and distinct differences among themselves.

Figure 2.

Relative knowledge-based and labour market growth by region, 2001/02 to 2004/05.

The differences across Chinese regions tends to confirm evidence that the modes of growth across these regions vary as a result of differing development processes, especially in terms of the key pre-existing regional resources and institutions (Fleisher et al., 2010). In Shanghai, for instance, growth has evolved through a model incorporating both indigenous and external knowledge sources in the form of foreign investment (Wu, 2007). In Beijing, both state and local governments have allocated large human and financial resources to education and technology, as well as the setting-up of spin-off enterprises to commercialise technology and the establishment of high-technology development zones within the region (Zhao and Tong, 2001). By contrast, in Guangdong, which has a knowledge-based growth factor score considerably lower than its Chinese counterparts, growth has been particularly associated with FDI-driven characteristics, especially overseas Chinese-owned FDI (Huang et al., 2012). Yet, on the whole there has been a rapid increase in R&D units in firms across all regions in China.

Unlike the Chinese regions, knowledge-based growth in leading regions in the US has not gone hand-in-hand with labour market growth. The US regions fall into six groups, indicating significant diversity across the nation. At the centre of Figure 2 is a group consisting of nearly 40 regions in the US, Europe and Japan; and, on the right side of the figure, San Jose, (the home of Silicon Valley), and a few other regions in the US, along with British Columbia and regions in Australia, constitute one of the best-performing groups for the labour market growth factor, as well as being near the mean of knowledge-based growth factor. Between these two groups is a cluster representing one-third of the US regions covered in the study (including New York) for which performance on the knowledge-based growth is near the bottom of the sampled regions. These three groups form a U-shaped arch disturbed by another cluster including Charlotte, Richmond, Virginia Beach and Washington DC, for which knowledge-based growth is above the mean. Hartford is an outlier belonging to an all-European group on the left, and another group of Riverside and Miami is situated at the right bottom corner.

Other research concerning growth across US regions suggests that growth rates in per capita income are associated with knowledge-based inputs whereas labour market growth rates are mainly associated with industry structure characteristics, which can be further related to economic history and industrial legacy (Barkley and Dudensing, 2011). Furthermore, for the period 2000–06, Barkley and Dudensing (2011) find only a limited role for knowledge-based economic development policies with respect to enhancing job growth across many US regions. These findings confirm the results of the cluster analysis, whereby for the majority of leading US regions labour market growth is not necessarily associated with knowledge-based growth.

In the rest of North America, the Canadian regions analysed in this study show above-average factor scores for knowledge-based and labour market growth, with Alberta and Saskatchewan forming the best-performing group outside East Asia. This is in line with a wider analysis of regional growth in Canada, which suggests that productivity gains emerge from knowledge spillovers (Baldwin et al., 2008), coupled with strong employment growth over the surveyed period.

Falling into six groups, the European regions covered in this study also exhibit considerable unevenness in their growth trajectories. A group consisting of Île de France and 20 other European regions, and two single-region groups of Berlin and Brussels, represent the most sluggish in terms of the labour market growth factor, being on their own except for Hartford in the US. By contrast, the rest of the European regions analysed are mixed with regions in the US, Canada and Japan, showing near- or above-average-factors scores for labour market growth. In terms of knowledge-based growth, a number of regions in Sweden, the Netherlands and the UK score above-average, whereas the rest lag behind with below-average scores. A study of regional growth across European regions finds that only within certain northern nations of Europe is there a significant relationship between regional growth and the intensity of R&D and higher education, suggesting a fragmented pattern of knowledge-based growth across regions (Sterlacchini, 2008). Such fragmentation and growth disparities across European regions have also been found by other empirical studies (Rodríguez-Pose and Crescenzi, 2008; Rodríguez-Pose and Tselios, 2010).

In Oceania and East Asia—with the exception of China—the Australian regions covered in this study form a group, along with a small number of regions in North America, that shows high factor scores for labour market growth accompanied by average or above-average factor scores for knowledge-based growth. Stimson et al.’s (2009) finding that regional growth in Australia is a complex interaction between population growth, human capital and economic structure, suggests that knowledge and labour markets are closely entwined. Although split into two groups, the Japanese regions analysed in this study show a relatively homogeneous pattern of growth—above-average knowledge-based growth and near-average or above-average labour market growth. Essletzbichler and Kadokawa (2010) find a pattern of rising overseas investment coupled with the increased import penetration of less technology-intensive products in Japan since the 1990s, which is most likely to stimulate knowledge-based growth across the nation’s leading regions.

Discussion and Conclusion

Our analysis of leading regions across the globe identifies two key trends by which their economic evolution and growth patterns are differentiated—namely, knowledge-based growth and labour market growth. The knowledge-based growth factor identified in the analysis shows commonality across the growth of GDP, gross monthly earnings and business R&D and, to a lesser extent, growth in public expenditure on higher education and government R&D. Given the aim and analytical method of this study, this finding does not infer causality between any pair of the variables, but the association found between key investments in knowledge production and an economy’s output is in agreement with the tenet of innovation systems theory, as well as more general endogenous growth theory. While some studies (for example, Sterlacchini, 2008) report the lack of a significant relationship between accumulation of knowledge-based resources and economic growth among regions within a continental bloc, this association clearly emerges across the leading regions covered by this study.

In terms of the knowledge-based growth factor, we observe a spread of knowledge-based resources in the form of investments in knowledge production and human capital formation in regional economies within China. Compared with the spectacular growth of leading Chinese regions, knowledge-based growth in the other regions is more modest. However, there are still significant variations among more established regions. Whereas a sluggish performance is shown across many regions in Europe and the US (with the exception of a few hot-spots such as San Jose, Greensboro, Richmond, Virginia Beach, Washington DC), leading regions in Australia, Japan and Canada made relatively strong progress. Due to their geographical proximity, regions in the Pacific Rim are more likely to enjoy an opportunity for the offshoring of labour-intensive operations to China, as well as robust demand from within the Chinese economy, facilitating a shift to more knowledge-intensive and innovation-driven activities. Overall, we observe a continued shift of knowledge-based resources and capabilities to the East, or what can be termed the globalisation of knowledge-based development.

The other key trend identified is the labour market growth factor, representing a region’s capability to draw on a greater part of its human capital. The cross-regional independence of the labour market growth factor from the variables constituting the knowledge-based growth factor indicates the probability of more widely varying, complex interactions between innovation and employment across regions than those suggested by emerging studies focusing on the co-evolution of regional economic components (for example, Buerger et al., 2012).

Only through a broadening of the frameworks employed to analyse these associations can both the commonality and differentiations in the evolutionary trajectories of regions become more visible. This broadening is required to occur at three levels: the scope of regional coverage; the scope of analysed variables; and, the scope of the evolutionary time frame. With reference to the third point, it is clear that, in addition to the complex interactions within knowledge-intensive sectors, there may be a relatively long time lag before a potential relationship between changes in investment in knowledge-intensive sectors and employment in their less knowledge-intensive counterparts within the regional economy becomes manifest. Therefore, we acknowledge the possibility that the time frame adopted in our analysis is too short to fully capture associations between them, suggesting the need for future studies to build upon the underlying analytical framework presented in this paper but also to seek to encompass a longer time frame. However, within the time frame, regional coverage and variables adopted by this study, it can be concluded that regional growth in knowledge production investment and the capacity to draw on regional human capital reserves are neither necessarily traded-off nor complementary to each other.

From a policy perspective, in regions where firms commonly pursue a strategy of positioning themselves in a market for high value-added, knowledge-intensive goods and services, these firms are more likely to either outsource lower value-added, labour-intensive tasks to other providers—within or outside their region—or move altogether to another market for new, higher value-added products. By contrast, in regions where firms seek to serve the market for low-cost products, these firms may attempt to improve their competitive position by combining ICT-enabled operations with less-skilled, low-wage workers. Whether firms go down the former or latter route is a strategic choice mediated by the policy environment in which they operate, including labour market regulations and tax incentives for R&D. Indeed, regions in the US, which tend to operate in an environment of relatively deregulated labour markets, generally show higher factor scores for labour market growth than regions in a more regulated European environment. Yet, within both the US and Europe there are significant regional variations in labour market growth factor scores, indicating the role played by the strategic choices made at a firm level and overall industry mix at the regional level. In conclusion, the findings suggest that policies designed to facilitate knowledge-based development should pay close care and attention to the region-specific conditions that influence the interactions between knowledge- and innovation-based development and wider labour-market-related economic development. Without such attention, there is a possibility of disconnection between innovation-led policies and the economic outcomes that regional policy-makers hope to realise.

Footnotes

Notes

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

References

Acs

Z. J.

Anselin

Varga

(2002) Patents and innovation counts as measures of regional production of new knowledge, Research Policy, 31(7), pp. 1069–1085.

Adams

J. D.

(1990) Fundamental stocks of knowledge and productivity growth, Journal of Political Economy, 98(4), pp. 673–702.

Baldwin

J. R.

Beckstead

Brown

W. M.

Rigby

D. L.

(2008) Agglomeration and the geography of localization economies in Canada, Regional Studies, 42(1), pp. 117–132.

Barkley

D. L.

Dudensing

R. M.

(2011) Industrial legacy matters: implications for the development and use of indices of regional competitiveness, Economic Development Quarterly, 25(2), pp. 130–142.

Bartholomew

D. J.

Knott

(1999) Latent Variable Models and Factor Analysis, 2nd edn. London: Arnold.

Bilbao-Osorio

Rodríguez-Pose

(2004) From R&D to innovation and economic growth in the EU, Growth and Change, 35(4), pp. 434–455.

Borrás

(2004) System of innovation theory and the European Union, Science and Public Policy, 31(6), pp. 425–433.

Bradley

Taylor

(1996) Human capital formation and local economic performance, Regional Studies, 30(1), pp. 1–14.

Buerger

Broekel

Coad

(2012) Regional dynamics of innovation: investigating the co-evolution of patents, research and development (R&D) and employment, Regional Studies, 46(5), pp. 565–582.

10.

Buesa

Heijs

Baumert

(2010) The determinants of regional innovation in Europe: a combined factorial and regression knowledge production function approach, Research Policy, 39(6), pp. 722–735.

11.

Buesa

Heijs

Pellitero

M. M.

Baumert

(2006) Regional systems of innovation and the knowledge production function: the Spanish case, Technovation, 26(4), pp. 463–472.

12.

Capello

Nijkamp

(Eds) (2009) Handbook of Regional Growth and Development Theories. Cheltenham: Edward Elgar.

13.

Carlsson

(2007) Innovation systems: a survey of the literature from a Schumpeterian perspective, in: Hanusch

Pyka

(Eds) Elgar Companion to Neo-Schumpeterian Economics, pp., 857–871. Cheltenham: Edward Elgar.

14.

Cooke

(2004) Regional innovation systems: an evolutionary approach, in: Cooke

Heidenreich

Braczyk

H.-J.

(Eds) Regional Innovation Systems, 2nd edn. pp. 1–18. London: Routledge.

15.

Crescenzi

Rodríguez-Pose

Storper

(2007) The territorial dynamics of innovation: a Europe–United States comparative analysis, Journal of Economic Geography, 7(6), pp. 673–709.

16.

Crouch

Finegold

Sako

(1999) Are Skills the Answer? Oxford: Oxford University Press.

17.

Dicken

(2007) Global Shift, 5th edn. New York: Guilford Press.

18.

Doloreux

Parto

(2005) Regional innovation systems: current discourse and unresolved issues, Technology in Society, 27(2), pp. 133–153.

19.

Döpke

(2001) The ‘employment intensity’ of growth in Europe. Working Paper No. 1021, Kiel Institute of World Economics, Kiel, Germany.

20.

Dunford

(2005) Old Europe, new Europe and the USA: comparative economic performance, inequality and market-led models of development, European Urban and Regional Studies, 12(2), pp. 149–176.

21.

Essletzbichler

Kadokawa

(2010) The evolution of regional labour productivities in Japanese manufacturing, 1968–2004, Regional Studies, 44(9), pp. 1189–1205.

22.

Fagerberg

(1996) Technology and competitiveness, Oxford Review of Economic Policy, 12(3), pp. 39–51.

23.

Fleisher

Zhao

M. Q.

(2010) Human capital, economic growth and regional inequality in China, Journal of Development Economics, 92(2), pp. 215–231.

24.

Freeman

(1987) Technology, Policy, and Economic Performance. London: Pinter Publishers.

25.

Freeman

(2002) Continental, national and sub-national innovation systems: complementarity and economic growth, Research Policy, 31(2), pp. 191–211.

26.

Fritsch

(2002) Measuring the quality of regional innovation systems: a knowledge production function approach, International Regional Science Review, 25(1), pp. 86–101.

27.

Gorsuch

R. L.

(1983) Factor Analysis, 2nd edn. Hills Dale, NJ: Lawrence Erlbaum Associates.

28.

Griliches

(1990) Patent statistics as economic indicators: a survey, Journal of Economic Literature, 28(4), pp. 1661–1707.

29.

Hair

J. F.

Jr Black

W. C.

Babin

B. J.

Anderson

R. E.

(2010) Multivariate Data Analysis, 7th edn. Upper Saddle River, NJ: Pearson.

30.

Hatzichronoglou

(1997) Revision of the high-technology sector and product classification. Working Paper No. 1997/2, OECD Directorate for Science, Technology and Industry, Paris.

31.

Heidenreich

(2004) The dilemmas of regional innovation systems, in: Cooke

Heidenreich

Braczyk

H.-J.

(Eds) Regional Innovation Systems, 2nd edn, pp. 363–389. London: Routledge.

32.

Huang

Liu

(2012) Regional innovation and spillover effects of foreign direct investment in China: a threshold approach, Regional Studies, 46(5), pp. 583–596.

33.

Huggins

Izushi

(2007) Competing for Knowledge: Creating, Connecting, and Growing. London: Routledge.

34.

Huggins

Izushi

(Eds) (2011) Competition, Competitive Advantage, and Clusters: The Ideas of Michael Porter. Oxford: Oxford University Press.

35.

Huggins

Johnston

Steffenson

(2008) Universities, knowledge networks and regional policy, Cambridge Journal of Regions, Economy and Society, 1(2), pp. 321–340.

36.

Iammarino

McCann

(2006) The structure and evolution of industrial clusters: transactions, technology and knowledge spillovers, Research Policy, 35(7), pp. 1018–1036.

37.

Izushi

(2008) What does endogenous growth theory tell about regional economies? Empirics of R&D worker-based productivity growth, Regional Studies, 42(7), pp. 947–960.

38.

Jaffe

A. B.

(1989) Real effect of academic research, American Economic Review, 79(5), pp. 957–970.

39.

Kline

S. J.

Rosenberg

(1986) An overview of innovation, in: Landau

Rosenberg

(Eds) The Positive Sum Strategy, pp. 275–305. Washington, DC: National Academy Press.

40.

Laafia

(1999) Regional employment in high technology. Research and Development Theme No. 9(1), Statistics in Focus, Eurostat, Luxembourg.

41.

Lagendijk

Cornford

(2000) Regional institutions and knowledge: tracking new forms of regional development policy, Geoforum, 31(2), pp. 209–218.

42.

Lee

F. C.

Has

(1996) A quantitative assessment of high-knowledge industries versus low-knowledge industries, in: Howitt

(Ed.) The Implications of Knowledge-based Growth for Micro-economic Policies, pp. 39–78. Calgary, AB: University of Calgary Press.

43.

Lundvall

B.-Å

. (Ed.) (1992) National Systems of Innovation. London: Pinter Publishers.

44.

MacKinnon

Cumbers

Chapman

(2002) Learning, innovation and regional development: a critical appraisal of recent debates, Progress in Human Geography, 26(3), pp. 293–311.

45.

Mahroum

Huggins

Clayton

Pain

Taylor

(2008) Innovation by adoption: measuring and mapping absorptive capacity in UK nations and regions. Nesta, London.

46.

Markusen

(1996) Sticky places in slippery spaces: a typology of industrial districts, Economic Geography, 72(3), pp. 293–313.

47.

Maskell

Malmberg

(2007) Myopia, knowledge development and cluster evolution, Journal of Economic Geography, 7(5), pp. 603–618.

48.

OECD (Organisation for Economic Co-operation and Development) (1994) Jobs Study: Evidence and Explanations. Paris: OECD.

49.

Pavitt

(1982) R&D, patenting and innovative activities: a statistical exploration, Research Policy, 11(1), pp. 33–51.

50.

Phelps

N. A.

Ozawa

(2003) Contrasts in agglomeration: proto-industrial, industrial and post-industrial forms compared, Progress in Human Geography, 27(5), pp. 583–604.

51.

Porter

M. E.

(1998) On Competition. Boston, MA: Harvard Business School Press.

52.

Rodríguez-Pose

Crescenzi

(2008) Research and development, spillovers, innovation systems, and the genesis of regional growth in Europe, Regional Studies, 42(1), pp. 51–67.

53.

Rodríguez-Pose

Tselios

(2010) Inequalities in income and education and regional economic growth in western Europe, Annals of Regional Science, 44(2), pp. 349–375.

54.

Romer

P. M.

(1986) Increasing returns and long-run growth, Journal of Political Economy, 94(5), pp. 1002–1037.

55.

Romer

P. M.

(1990) Endogenous technological change, Journal of Political Economy, 98(5), pp. S71–S102.

56.

Schwarz

(1978) Estimating the dimension of a model, Annals of Statistics, 6(2), pp. 461–464.

57.

Solow

R. M.

(1957) Technical change and the aggregate production function, Review of Economics and Statistics, 39, pp. 312–320.

58.

Sterlacchini

(2008) R&D, higher education and regional growth: uneven linkages among European regions, Research Policy, 37(6/7), pp. 1096–1107.

59.

Stimson

R. J.

Robson

Shyy

T.-K.

(2009) Modeling regional endogenous growth: an application to the non-metropolitan regions of Australia, Annals of Regional Science, 43(2), pp. 379–398.

60.

Tether

B. S.

Tajar

(2008) Beyond industry–university links: sourcing knowledge for innovation from consultants, private research organisations and the public science-base, Research Policy, 37(6/7), pp. 1079–1095.

61.

Vázquez-Barquero

(2007) Endogenous development: analytical and policy issues, in: Scott

A. J.

Garofoli

(Eds) Development on the Ground: Clusters, Networks and Regions in Emerging Economies, pp. 23–43. London: Routledge.

62.

Vivarelli

Pianta

(Eds) (2000) The Employment Impact of Innovation: Evidence and Policy. London: Routledge.

63.

(2007) Cultivating research universities and industrial linkages: the case of Shanghai, China, World Development, 35(6), pp. 1075–1093.

64.

Zhao

S. X.

Tong

S. P.

(2001) Unequal economic development in China: spatial disparity and regional policy: 1985–95, Regional Studies, 34(6), pp. 549–561.