Innovation and Development: The Evidence From Innovation Surveys

ABSTRACT

In this article we investigate the existing evidence on innovation produced by innovation surveys in developing and emerging countries in Europe, Asia, Africa, and Latin America. We review the relevant literature, discuss methodological issues, and analyze the results for the countries with the most comparable surveys, considering the well-established findings of innovation surveys for Europe as a benchmark. From the evidence we considered, regional patterns are identified and some stylized facts on innovation and development are proposed, pointing out the specificity of innovation processes in economies engaged in industrialization and catching up.

RESUMEN

En este artículo investigamos evidencias encontradas sobre las innovaciones, obtenidas mediante encuestas sobre innovación realizadas en países en desarrollo y emergentes situados en Europa, Asia, África y América Latina. Examinamos la literatura pertinente, discutimos cuestiones metodológicas y analizamos los resultados relativos a los países que demostraron tener resultados similares, considerando como referencial los resultados ya reconocidos para las encuestas sobre innovación elaboradas para Europa. A partir de la evidencia analizada, identificamos patrones regionales y sugerimos algunos factores estilizados sobre la innovación y el desarrollo, indicando la especificidad de los procesos de innovación en las economías dedicadas a la industrialización y la actualización.

RESUMO

No presente artigo investigamos as evidências existentes sobre as inovações produzidas por levantamento de inovação em países em desenvolvimento e emergentes na Europa, Ásia, África e América Latina. Examinamos a literatura relevante, discutimos questões metodológicas e analisamos os resultados para os países com os levantamentos mais semelhantes, considerando os achados bem definidos de levantamentos de inovação para a Europa como um benchmark. A partir das evidências contempladas foram identificados padrões regionais e foram propostos alguns fatos estilizados sobre inovação e desenvolvimento, indicando a especificidade dos processos de inovação em economias engajadas na industrialização e compensação.

KEYWORDS:

• emerging countries
• innovation surveys
• patterns of innovation

ACKNOWLEDGMENTS

We thank Franco Malerba, Maurizio Franzini, Fulvio Mulatero, Pierre Mohnen, and various participants to seminars in Milan KIBS, a workshop at Roma Tre University and seminars at Universidad EAFIT (Medellin), IPTS (Seville), Urbino. A special thanks to the managing editor and two anonymous referees for their careful reading and the many suggestions to improve a previous version of the paper.

Notes

Note. M = Manufacture; S = services; KIBS refers to knowledge intensive business services. In column (2), the unit is the share of total firms; in columns (3)–(5) the share of innovative firms; and in the final column the share of total turnover.

*includes turnover due to imitation.Sources: see Appendix.

Note. M = Manufacture; S = services; KIBS refers to knowledge intensive business services. The unit is the share of total innovative expenditure.

Source: see Appendix.

Note. M = Manufacture; S = services; KIBS refers to knowledge intensive business services. Data are expressed as shares of innovative firms.

Source: see Appendix.

Note. M = Manufacture; S = services; KIBS refers to knowledge intensive business services. Data are expressed as shares of innovative firms.

Source: see Appendix.

Note. M = Manufacture; S = services; KIBS refers to knowledge intensive business services. Data are expressed as shares of innovative firms.

Source: see Appendix.a. The Argentinean questionnaire asks about high-risk and not high-cost of innovation.b. Brazil does not distinguish between internal and external source of financing, so the total value is reported in both columns.c. Chilean data do not consider the lack of markets, and we have replaced it with information on “long period of return”; there is no distinction between the two types of information problems, and the total value is reported in both columns.

Note. Eigenvectors associated to the first components. Eigenvector normalization: sum of squares equal to one. Presence of demand is the rescaling of the lack of demand variable: “Presence of demand” is equal to one hundred minus the value of the variable “Lack of demand.” All the other variables are in shares (scale 0–100). Eleven observations: Brazil (M and S), China (M), Colombia (M), EU-15 (M), EU-NAC (M), Russia (M), Turkey (M and S), South Africa (M and S).

Note. Unit of measure: in column (2), (4), (5), and (6) it is the share of GDP, in column (3) it is the share of Labor Force (expressed in thousands), and in column (7) the share of population aged 25–64 years.

Sources:

	Column two, three and four: Lederman and Saenz (2005); European Statistics New Cronos Database; World Bank, World Development Indicators database; South Africa Human Science Research Council (2009); Association of Southeast Asian Nations (2010), S&T Indicators/Technology Competitiveness Indicators.
	Column five: World Bank, 54 indicators; Taiwan Council for Economic Planning and Development (2009), Taiwan Statistical Data Book.
	Column six: both FDI and GDP are expressed in current US dollars; for countries outside Europe the average covers 2002–2005; World Bank, 54 indicators; Taiwan Council for Economic Planning and Development (2009), Taiwan Statistical Data Book.
	Column seven: ILO, LaborSta; OECD Labour Force Statistics, Education at Glance; Taiwan Statistical Data Book.

.

Some of the European and Asian countries we will consider are developed economies. However, our evidence includes the 1990s, when industrialization and international integration accelerated, and is illustrative of the role of innovation in successful development strategies. For the purpose of this article, the “development” ranking is the one based on real Gross Domestic Product per capita (in PPP), as standard in the literature.

The positive role that innovation surveys can play in a better design of policies is stressed by Mairesse and Mohnen (2008) and Sutz (2007), especially with reference to the existence of policy complementarities. Fagerberg and Srholec (2009) carried out a comprehensive analysis of innovative patterns (using data from various sources that do not include innovation surveys), showing that policies tailored to the attraction of high-tech activities can be an appropriate strategy only when coupled with a strengthening of the quality of the environment (capacity to mobilize the proper factors, reliability of the social and institutional structure, capability to move from idea to innovation). Cimoli, Dosi, and Stiglitz (2009) linked policies for technological change, accumulation of capabilities, and development.

A growing debate is addressing the global governance of technology flows; although some of the reforms of the 1990s have been recognized as productivity enhancing (Figueiredo, 2008; López, 2008; Dijiofack-Zebaze & Keck, 2009), there is a widespread concern that the actual regime of Intellectual Property Rights Protection is too strong (Chang, 2001; Falvey, Folster, & Memedovic, 2006, Bogliacino & Naranjo, 2008; Stiglitz, 2008); new avenues could be opened in terms of South-South cooperation (Pérez, 1994), while new conflicts may emerge with the rise of new actors such as China (Gu et al. 2008).

A private survey following the Oslo Manual has also been conducted in Sri Lanka, mainly addressing the issue of entrepreneurship; preliminary evidence can be found in De Mel, McKenzie, and Woodruff (2009).

We thank a referee for pointing out this issue.

New Member States of the European Union (EU-NMS) include the following countries of Central and Eastern Europe: Bulgaria, Cyprus, Czech Republic, Estonia, Latvia, Lithuania, Hungary, Malta, Poland, Romania, Slovenia, and Slovakia.

As shown by a large literature, the strong differences in innovative patterns across industries and the frequent concentration of technological activities in a limited number of industries present in developing countries suggest that an industry breakdown would greatly improve our understanding of the position of each country. Limitation in access to data is the reason more-detailed data were not available.

When considering data for EU-15—a benchmark for advanced countries—we should keep in mind a few caveats. EU-15 data are unweighted averages of values for the 15 countries that include some economies with limited innovative activities. European countries are characterized by slow growth, industrial decline, consolidated markets, and strong international integration, and this reflects on their innovative performances.

Hampering factors tend to correlate positively with the resources invested in innovation, clear-cut evidence that innovation is a matter of capabilities, i.e., of seeing the opportunities and the related difficulties. When we move toward countries closer to the technological frontier, we may find that data are more representative of the problems encountered by the universe of firms and not just by the more innovative ones. Moreover, in a subjective survey firms may point out the lack of funds as an issue that sums up all problems. In the case of Colombia, where we have investigated the microdata of the second innovation survey (2003–2004), when we run a regression of innovative expenditures on the hampering factors we have always positive (and sometimes significant) coefficients for all factors except financing, which is negative and significant; technically, we run Tobit-II and hurdles models, so the estimates are robust to the sample selection induced by the survey (non innovators do not fill in this part of the questionnaire).

Due to the selection problem in the data, we do not investigate the effect of innovation on performance measures, and we prefer to focus on the identification of key patterns of technological efforts. However, they play a major role in the growth of emerging countries and there is broad consensus that without innovation and technology, the development process cannot be explained.

As suggested by one referee, this result is consistent with the previous literature that suggests that R&D is a critical input mostly for product innovation and that product innovation requires also stronger interaction with clients. However, the clients variable in the Innovation Surveys is usually very correlated with the suppliers one, due to the evolution of value chain but also to the presence in many cases of specialized suppliers very much integrated with (and dependent for sources from) both clients and suppliers. Also in this database, the correlation between the two variables is 88.58%. If we re-run the principal component analysis with the “clients” variables instead of the “demand” variable, we get essentially the same results (the scores for the retained components in this new version correlate respectively at 98.9% and 78.9% with the ones discussed previously). However, the behavior of clients is pretty much indistinguishable from the suppliers one (in the associated eigenvectors they have the same coefficients, positive with the technology adoption and negative with the technology generation). We thank the referee for suggesting this robustness check.

A similar picture could emerge from plotting the scores of the principal components, but we prefer to identify clearly the combined effect of the dominant variables shaping technology adoption and technology generation in emerging countries. This is coherent with the statistical evidence that the two components are clearly associated to different sets of variables.

In studies of the determinants of innovation, Langebaek and Vasquez (2007) found this result for Colombia, Turriago (2003) for Argentina, Colombia and Venezuela, and Benavente (2005a, 2005b, 2006) for Chile. Chudnovsky, López, and Pupato (2006) for Argentina, Johnson (2002) for Brazil, and finally Gonçalves, Lemos, and Negri (2008) for Argentina and Brazil. Partial confirmation can be found in Marotta, Mark, Blom, and Thorn (2007), whose coefficient for large firms is significant for Colombia but not for Chile.

The supporting evidence can be found in Alvarez (2001) for Chile, Gonçalves and colleagues (2008) for Argentina and Brazil, Marotta and associates (2007) for Chile, Correa (2005) for a sample of a metropolitan area in Colombia.