Empirical Studies on National Innovation Systems (NIS)

NIS are key drivers of economic growth and competitiveness. The NIS perspective originally developed during the 1990s to understand the broad set of factors shaping the innovation and imitation ability of countries, and how these factors could contribute to explain cross-country differences in economic growth and competitiveness (Lundvall 1992; Edquist 1997). The study of NIS focuses on the main components of the system, such as private firms and public organizations, and investigates their mutual interactions as well as their relationships with the social and institutional framework in which the system is embedded (Freeman 1995; Lundvall 2007). Empirical studies in this tradition initially focused mostly on advanced economies in the OECD area (Nelson 1993), although the NIS literature has recently shifted the focus toward the study of innovation systems within the context of developing and less developed economies (Lundvall et al. 2009).¹

NIS empirical literature has traditionally devoted substantial attention to the process of international knowledge diffusion and investigated the set of factors that affect the extent to which a national system is able to grow and catch up with the technological frontier by means of international learning and imitation activities. This approach was originally inspired by the work of authors such as List, Landes, Gerschenkron, and Abramovitz, which, by focusing on historical case studies of the technological catch-up process, pointed out that international knowledge diffusion is a complex and demanding process, and investigated the set of factors that are necessary for imitation-based technological development, so-called absorptive capacity of a country (Abramovitz 1986).

More recently, a strand of empirical research made an effort to measure some of the relevant dimensions composing a national system through aggregate data and statistical indicators, and carried out applied econometric investigations of the relationships between these dimensions, on the one hand, and income per capita dynamics, on the other. However, a well-known challenge for applied research in this field is how to operationalize the innovation system view in empirical studies and, relatedly, how to measure the complex and multifaceted concept of NIS and its relationship to countries’ economic performance. Quantitative applied studies of NIS and development have so far made use of two different (albeit complementary) approaches.

The first approach is rooted in the traditional literature on technology and convergence (Abramovitz 1986; Verspagen 1991; Fagerberg 1994). Following a technology-gap Schumpeterian approach, recent econometric studies have focused on a few key variables that explain (or summarize) cross-country differences in the innovation ability of countries as well as their different capabilities to imitate foreign advanced knowledge, and analyzed the empirical relationship between these innovation and imitation factors and cross-country differences in GDP per capita growth (Fagerberg and Verspagen 2002; Castellacci 2004; Castellacci 2008; Castellacci 2011; Fagerberg et al. 2007). Since one main motivation of this type of studies is to analyze the dynamics and evolution of national systems in a long-run perspective, they typically consider a relatively long time span (e.g., from the 1970s or 1980s onward), but must for this reason focus on a more restricted sample of countries (e.g., between 70 and 90 countries). Due to the lack of statistical data for a sufficiently long period of time, therefore, a great number of developing economies and the vast majority of less developed countries are often neglected by these cross-country studies.

The second approach is based on the construction and descriptive analysis of composite indicators. In a nutshell, this approach recognizes the complex and multidimensional nature of national systems of innovation and tries to measure some of their most important characteristics by considering a large set of variables representing distinct dimensions of national systems and capabilities, and then combining them together into a single composite indicator – which may be interpreted as a rough summary measure of a country’s relative position vis-à-vis other national systems. Desai et al. (2002) and Archibugi and Coco (2004) have firstly proposed composite indicators based on a simple aggregation (simple or weighted averages) of a number of technology variables. Godinho, Mendonça, and Pereira (2005), Castellacci and Archibugi (2008), and Fagerberg and Srholec (2008) have then considered a larger number of innovation system dimensions and analyzed them by means of factor and cluster analysis techniques. As compared to the first approach, the composite indicator approach has a more explicit focus on the comparison across a larger number of countries. Consequently, due to the lack of data availability on less developed countries for a sufficiently long period of time, these studies typically focus on a relatively short time span, giving a cross-section description of the sample in one point in time (e.g., the 1990s and/or the 2000s).

Most of this empirical research, however, has so far focused on the cross-country comparative aspect (“why growth rates differ”) and mostly neglected the time series dimension and the analysis of the dynamics of the technological catch-up and economic growth process over time. Further, an exceptional amount of research has been devoted to the study of the determinants of GDP and income per capita growth. By contrast, only a limited number of studies have empirically investigated the dynamics of innovative capability and absorptive capacity over time and the main factors that may explain their long-run evolution (Filippetti and Peyrache 2011; Castellacci and Natera 2013). This is a crucial task for future research in this field, and the present chapter intends to take a step in this direction.

Conceptual Framework and Dimensions

Although the concept of national system was originally intended to comprise a broad and multidimensional set of capabilities, modeling and empirical exercises have often provided a stylized and simplified operationalization of it. As recently argued by Archibugi and Coco (2004), Godinho et al. (2005), and Fagerberg and Srholec (2008), it is indeed important to adopt a multifaceted description and measurement of the various factors that contribute to shape the innovative capability and absorptive capacity of nations.

Given that the concept of national systems is complex, multifaceted, and comprises a great number of relevant factors interacting with each other, our empirical analysis adopts a broad and multidimensional operationalization of it. Our stylized view, broadly in line with the previous literature, is presented in Figure 1.1.² The diagram provides a stylized representation of the main aggregate dimensions that define the dynamics of a national system. Specifically, we represent NIS as composed of six main dimensions: (1) Innovation and technological capabilities; (2) Openness; (3) Infrastructures; (4) Education and human capital; (5) Political institutions; (6) Social cohesion.

The first three dimensions are broadly related to a host of technological and economic factors, which classical neo-Schumpeterian work by Chris Freeman (1987) summarized under the heading techno-economic system. The next three dimensions refer instead to a heterogeneous set of social and institutional factors that may be combined together under the label socio-institutional system (see also Freeman and Louça 2001; Castellacci 2004). The underlying idea motivating this framework is that the dynamics and complex interactions between these six dimensions represent the driving force of national systems’ social and economic development, and it is therefore crucial for empirical analyses in this field to have statistical information available for a large number of indicators and country-year observations, and to study how each of these dimensions evolve in the long run.³

National Specificities and Cross-Country Heterogeneity

This stylized framework intends to provide a general background for our empirical analysis of the global evolution of national innovation systems. Nevertheless, it is important to emphasize that, although these six conceptual dimensions are arguably important for all countries, the way in which they combine and interact with each other greatly differs among national economies. In particular, the specific trajectory followed by distinct national systems of innovation and the specific set of key factors driving the dynamics of the system may differ across country groups characterized by different levels of development. To illustrate this, consider for instance some of the main differences between advanced (high-income), middle-income, and less developed economies, as often depicted in the literature.

Advanced countries are usually rich in terms of both absorptive capacity (infrastructures, openness, and human capital) and innovative capability. In particular, private R&D capabilities become a more central engine of growth as compared to less advanced national systems, and they are also more closely related to the development of public scientific activities. This is one of the central points made by historically oriented studies on the emergence of new world leaders such as Imperial Germany (Veblen), the United States (Abramovitz), and Japan (Freeman), as well as the seminal studies on national systems of innovation in advanced economies (Nelson 1993). Therefore, this group of countries is typically characterized by an intense and complex web of relationships among the key variables driving the system, and in particular a two-way process of coevolution between innovative capability and absorptive capacity.

Compared to the advanced group, middle-income countries often lack the innovative capability that is required to become a world technological leader. However, they have in general a well-developed absorptive capacity and may then exploit the international diffusion of advanced technologies. For instance, previous studies of East Asian innovation systems have pointed out their rapidly increasing levels of human capital, ICT infrastructures, and export specialization and competitiveness (Edquist and Hommen 2008). Compared to the East Asian group, Latin American economies are instead characterized by lower levels of human capital and technical training, as well as a more traditional (resource-based) export specialization profile (Nelson 1993; Alcorta and Peres 1998). Finally, former Soviet countries in Eurasia, in spite of their high levels of human capital, have in general lower private R&D capabilities and weak scientific and technical interactions, which can be explained as the historical heritage these countries have received from the Soviet innovation system and its heavy reliance on publicly funded defense R&D (Freeman 1987; Nelson 1993). All in all, within the broad (and heterogeneous) group of middle-income economies, we normally find weaker interactions among the innovative capability variables, because of the less central role of R&D in the national system and, for the same reason, weaker interactions between innovative capability and absorptive capacity.

Finally, in less developed countries, national systems do not only lack strong innovative and R&D capabilities, but are also characterized by a more limited absorptive capacity, for example, in terms of lower levels of human capital and technical training, infrastructural investments, and ability to reap benefits from international trade activities (Fagerberg et al. 2007; Castellacci 2008; Lundvall et al. 2009). As a consequence, for this group of countries most of the dynamic interactions and coevolution patterns among the six conceptual dimensions noted above are likely to be weaker or non-existent. Below a threshold level of development, innovative capabilities and absorptive capacity are low, and their interaction is therefore not expected to be a key driver of the economic system.

Dimension 1: Innovation and Technological Capabilities

• R&D: R&D expenditures as a percentage of GDP (source: UNESCO, OECD).
• Scientific articles: Number of scientific articles per million people (source: NSF).
• High-tech products: Export of high-tech products as a share of manufacturing exports (source: World Bank).
• Patents: Number of patents registered at the US Patent and Trademark Office per million people (source: USPTO).

Dimension 2: Openness

• Inward FDI: Inward FDI flow, share of GDP (source: World Bank).
• Import: Imports of goods and services as a share of GDP (source: UNCTAD).
• Export: Export of goods and services as a share of GDP (source: UNCTAD).
• Openness: (Export + Import) / GDP (source: UNCTAD).

Dimension 3: Infrastructures

• Electricity: Kilowatts of electricity consumed per hour per capita (source: World Bank).
• Telephony: Sum of telephone mainlines and mobile phones per 1000 people (source: ITU, UNDP, World Bank).
• Carrier departures: Domestic takeoffs and takeoffs abroad of air carriers registered in the country, per 1000 inhabitants (source: World Bank).
• Paved roads: Paved roads as a percentage of the whole road length of the country (source: World Bank).

Dimension 4: Education and Human Capital

• Primary enrollment ratio: Share of primary students (source: UNDP, World Bank).
• Secondary enrollment ratio: Share of secondary students (source: UNDP, World Bank).
• Mean years of schooling: Average number of years of school completed in the population over 14 (Barro and Lee 2001; UNDP, World Bank).
• Tertiary enrollment ratio: Share of tertiary students in science and engineering in the population of that age group (source: UNDP, World Bank).

Dimension 5: Political Institutions⁵

• Political rights: People’s free participation in the political process. It ranges from –7 (low freedom) to –1 (total freedom) (source: Freedom House).
• Civil liberties: People’s basic freedoms without interference from the state. It ranges from –7 (low freedom) to –1 (total freedom) (source: Freedom House).
• Freedom of press: It reflects the degree of freedom that journalists and news organizations enjoy in each country, and the efforts made by the authorities to respect and ensure respect for this freedom. Index from –115 (no freedom) to 0 (high freedom) (source: Reporters Without Borders).
• Freedom of speech: Extent to which freedoms of speech and press are affected by government censorship, including ownership of media outlets. Index from 0 (Government censorship) to 2 (No Government Censorship) (source: Cingranelli and Richards 2008).

Dimension 6: Social Cohesion

• Corruption: Corruption Perception Index, ranging from 0 (High Corruption) to 10 (Low Corruption) (source: Transparency International).
• Inequalities: Gini index (within-country income inequalities).
• Trust: Percentage of respondents who agree with the statement “Most people can be trusted” (source: World Value Survey).
• Happiness: Index ranging from 3 (very happy) to 0 (not happy) (source: World Value Survey).⁶

As noted in the previous section, these dimensions and indicators of national systems have arguably different relevance and patterns for countries characterized by different levels of economic, institutional, and social development. It is therefore important that our empirical analysis of these indicators will not only consider the overall global patterns across all countries in the world, but also take into account the existence of cross-country heterogeneity and the specificities of different groups, or clubs, of economies. We do this by clustering countries into different groups, and then analyzing convergence and divergence patterns within each of these clubs too.

We have chosen to cluster countries in a simple hierarchical two-step manner. First, we identify three major country clubs, which are broadly in line with the existing literature: advanced economies, catching-up countries, and least developed economies (Howitt and Mayer-Foulkes 2005; Castellacci and Archibugi 2008). Second, in order to achieve a finer characterization of the widely different nature of economies within these three heterogeneous clubs, we further divide them into a few subgroups. We make this based on an exogenous and intuitive criterion: we follow broad geographical areas, which on the whole group together countries that are similar with respect to both the initial GDP per capita level (the usual clustering variable in this literature) and the overall institutional context and capitalist mode of development. All in all, we end up with a total of seven subgroups, as defined in Table 1.1.

Although it is clear that no clustering exercise is faultless, the advantage of the intuitive clustering method described here is twofold: it is broadly in line with the three-club specification adopted by most theoretical models and empirical exercises in this field, and at the same time, by working with seven fairly homogeneous subgroups, it deals in a satisfactory manner with the cross-country heterogeneity issue.

In order to illustrate that heterogeneity at a greater detail, we propose a comparative exercise for selected countries from different groups. We evaluate countries’ deviation from the world average for each dimension, in order to see how changes are taking place around the globe in two points of time, 1980 and 2008. If a developing country exhibits a higher level in 2008 than in 1980, it means that its relative position has improved or, equivalently, that it has experienced a process of catching up. By contrast, if an advanced country shows a lower level in 2008 than in 1980, it means that its leadership position has weakened in relative terms.

Figure 1.2 presents some descriptive evidence on some of the indicators noted above, for a selection of some major large economies for these seven groups. Each panel in Figure 1.2 reports a radar graph, which contains the composite indicators of each national economy on the six NIS dimensions highlighted in this paper and shows the evolution of the NIS between 1980 and 2008 relatively to the world average.

The radar graphs are informative. More advanced countries have on average a much greater surface than the catching-up BRICS and other less developed economies, indicating an overall greater level of the set of relevant technological, social, and economic capabilities both at the beginning and at the end of the period. Regarding the time evolution of these national systems, South Korea is a good illustration of a country that has improved rapidly its relative position during the last three decades, particularly in terms of innovation capabilities and infrastructures. Within the group of BRICS countries, the catching-up process between the beginning and the end of the period has also been striking for China, Brazil, and South Africa, and less so for Russia and India.

Methods

Our empirical analysis of the global evolution of national systems intends to provide a comprehensive description of cross-country differences in terms of the indicators outlined in the previous section, and how these differences have evolved over the period 1980–2008. We make use of a simple and commonly used method of applied economic research: convergence analysis. Cross-country convergence is one of most popular areas of research in applied growth theory and, despite some important limitations, it still provides an appealing method to summarize cross-country patterns and dynamics.

Most of the convergence literature has however focused on the dynamics of GDP per capita and the related determinants. But, while using the same convergence analysis methodology, our paper has a different objective. We intend to investigate the dynamics of each of the six NIS dimensions outlined in the previous sections, and how these have evolved during the last three decades. Since there exists no prior theory or model indicating how exactly these six dimensions evolve over time, we are not able to set up a structural model of conditional convergence, and follow instead a simpler empirical strategy. We carry out a standard analysis of (unconditional) convergence for each of these six factors separately, and study how their statistical distributions have evolved in the period 1980–2008. For each dimension, a pattern of convergence would indicate that a process of catching up is in place, meaning that less developed economies have experienced a more rapid dynamics than industrialized countries and hence been able to partly close their initial gap. By contrast, a finding of divergence would indicate the presence of a cumulative mechanism that is leading to increasing disparities between rich and poor countries.

More specifically, the analysis proceeds by considering the two most standard notions of (unconditional) convergence: β- and σ-convergence. For the β-convergence, we have estimated a simple regression model for each indicator A, where the dependent variable is the (average annual) growth of that indicator over the period 1980–2008, while the level of the same indicator at the beginning of the period is the only regressor (along with a constant). The cross-country regression model is:

(1)

where ΔA_i/A_i is the growth of each indicator for country i over the period, and A_i,0 is the log of its level at the beginning of the period. The parameter of interest in these regressions is β_A, which measures the speed of convergence, or divergence, for that indicator.

By contrast, the idea of σ-convergence is to study whether the dispersion of a variable has increased or decreased over time, thus providing a synthetic measure of the dynamics of the variability of its distribution. For each indicator, σ-convergence is computed as follows. First, we have calculated the standard deviation of the cross-country distribution at each time t; then we have regressed the standard deviation’s time series on a constant and a time trend:

(2)

If the estimated coefficient of the time trend σ turns out to have negative (positive) coefficient, this means that the indicator in question has decreased (increased) its variability over time, hence it shows σ-convergence (σ-divergence).

It is important to emphasize that the two notions of convergence, albeit related, are quite distinct from each other. β-analysis tells us whether, on average, a cross-country distribution has experienced a process of catching up – where initially poorer economies have developed more rapidly than richer countries (β-convergence) – or instead a cumulative process according to which initially richer economies have been able to reinforce their leadership position over time (β-divergence). On the other hand, σ-analysis provides information on the dynamics of the cross-country variability of an indicator: σ-convergence indicates decreasing cross-country dispersion, whereas σ-divergence points to increasing polarization between rich and less developed economies.

Further, in order to take into account the existence of substantial differences across country groups, we have undertaken our convergence analysis not only for the whole world distribution, but we have also repeated it for each of the seven geographically defined country groups in Table 1.1.

Results

We will now present the results of the convergence analysis for each dimension separately, and will then conclude by summarizing the main general results. Table 1.2 reports a summary of convergence analysis results,⁷ and Figures 1.3–1.8 show geographical maps and kernel densities to illustrate visually the cross-country distribution of selected indicators and their evolution over time.⁸