redes empresariales para la innovación

32
i

Upload: others

Post on 26-Mar-2022

0 views

Category:

Documents


0 download

TRANSCRIPT

i

1

Redes empresariales para la innovación en PYMEs de un país en desarrollo

Business Networks and Innovation in SMEs of a Developing Country

Clemente Forero-Pineda*Sandra Corredor Waldron**Nohora Forero Ramírez***

Abstract

The impact of business networking on innovation has been analyzed from different perspectives [Granovetter et al. (2000), Callon (2001), Cowan (2004), Powell and Grodal (2005), Cantner and Graf (2006)]. One central issue is whether informal and formal innovation-related links have differential effects on the innovation performance of a firm. This relationship is of particular interest in the case of SMEs, since it can be argued that external links may compensate the limitations imposed by their size and lack of resources. We address the issue empirically, using data for 4,003 SMEs in Colombia. Poisson regression models are used for explaining innovation counts, and Zero-Inflated Poisson models are best suited for firm-level data of SMEs in developing countries, where a large proportion of firms report zero innovations. Results indicate that both formal and informal links are important determinants of the innovation performance of SMEs. Internal capabilities of the firm moderate the rela-tionship between informal network links and innovation outcomes. Given the extent of imitation activities in firms of a developing country, a distinction is made between the analysis of product innovations based on invention and on imitation. In the case of inventions, informal links have a relatively higher importance than formal links, and mediate the relationship between internal capabilities and invention-based innovation counts. Finally, a comparison is made between innovation networking in SMEs and in large firms. Results show divergence on the relative importance of formal and informal links. The results are valuable for

innovation policy design.

Resumen

El impacto de las redes empresariales en la innovación ha sido analizado desde diferentes perspectivas [Granovetter, et al. (2000), Callon (2001), Cowan (2004), Powell and Grodal (2005), Cantner and Graf (2006)]. Un aspecto central de la discusión es la diferenciación entre las redes formales e informales y sus efectos en el desempeño innovador de la firma. Esta relación es de particular interés en el caso de las PYMEs, ya que se podría argumentar que los vínculos externos tienden a compensar las limitaciones que impo-nen el tamaño y la falta de recursos. Este artículo aborda el dilema empíricamente, utilizando la información de 4003 PYMEs en Colombia. En la literatura, se emplean modelos Poisson para explicar los conteos de innovaciones; sin em-bargo, dada la alta proporción de PYMEs que reportan cero innovaciones en un país en desarrollo como Colombia, se opta por el modelo Zero-Inflated Poisson. Los resultados indican que los vínculos tanto formales como informales son importantes determinantes del desempeño innovador de las PYMEs. Se observa igualmente que las capacidades internas de coordinación de una firma moderan la relación entre los vínculos informales y los resultados de innovación. Dada la importancia de la imitación en las firmas de un país en desarrollo, hacemos una distinción entre el análisis de las innovaciones de producto basadas en la invención y basadas en imitación. En el caso de las invenciones, los vínculos informales tienen una importancia relativa mayor que los vínculos formales, y median la relación que existe entre las capacidades internas y el número de innovaciones basadas en invención. Finalmente, se hace una comparación entre las redes empresariales para la innovación en PYMEs y grandes empresas. Los resultados muestran una divergencia en la importancia relativa de los vínculos formales e informales. Estos resultados son valiosos para el diseño de la política de innovación.

PaLabRaS CLavE: Innovación, redes formales, redes infor-males, capacidades internas, imitación, PYMEs.

* Professor, Universidad de los Andes, School of Management** Graduate Assistant, Universidad de los Andes, School of Management*** Research Assistant, Universidad de los Andes, School of Management

The authors thank Universidad de los Andes School of Management (UASM) for generous financial support. Dane-Colombia provided us with the database of the 2004-2005 Technological Development Survey. R. Winkelmann and M. Ramírez-Gómez gave important methodological suggestions. The students of the PhD Seminar on Innovation in Developing Countries at UASM made initial cluster analysis explorations and valuable comments.

2 Business Networks and Innovation in SMEs of a Developing Country

Introduction

Recent surveys on innovation networks show a wide consensus around the hypothesis that fertile ideas can-not take place in isolated environments (see for instance Powell and Grodal, 2005). based on this idea, some authors have focused on the impact of different type of links on innovation outcomes [Lööf and Heshmati (2002), belderbos et al. (2004a), Nieto and Santamaría (2007)]. The distinction between formal and informal network links, nonetheless, has been the object of few empirical explorations.

In this article we explore whether this distinction is useful, and what are the consequences of using it to classify the network links of small and medium enter-prises (SMEs) when analyzing the determinants of its innovation performance. Specifically, we ask whether some firms depend more than others on one or the other class of network links to obtain innovation re-sults. We explore the relationship between formal and informal business networks and innovation results, in the case of SMEs in a developing country (Colombia), and make a comparison of the effects of these links in SMEs and large firms. We also analyze the interac-tions between both types of external networks and the internal capacities associated with the ability to mobilize firm resources around innovation. Specifically, we explore the hypothesis that SMEs compensate the restrictions related to their size, or the volume of the resources they allocate to R&D, by seeking to build external ties with different types of organizations.

When in 1912 and in 1942, Schumpeter presented his two alternative views of the dynamics of innovation1 he attributed an implicit but crucial role to the size of the firms involved in those dynamics. In 1912, he centered his attention on the dynamics of relatively small firms that innovate to face market competition. In 1942, he moved to emphasize a dynamic where larger firms exhibit advantages to innovate, in view of their larger investment capabilities.

This dual theory of innovation has been the object of both theoretical explanations and empirical explora-

tions. The view that smaller firms, engaged in a compe-titive regime, may be more innovative than large firms, dominating the market through monopoly advantages, has been interpreted as the result of what became known as “arrow’s substitution effect”2, by which large firms enjoying dominant positions prefer to abstain from innovating because the new products might compete with their own marketed products. as far as in 1965, Scherer analyzed this issue empirically, and cast doubts on the existence of a positive relationship between size and innovation outcomes. These results were interpreted from different perspectives [Pavitt, Robson and Townsend (1987), acs and audretsch (1987)]. In a study for Great britain, Geroski (1994) rejects the hypothesis that larger firms are more inno-vative and explains that major innovations threaten the market structures that already exist. In his view, small firms, and other agents who are not well-positioned in that market, try to improve their competitive position with more aggressive strategies, while large firms enjoy the benefits of other activities and have incentives to slow down innovation. The consequence would be that large firms –and even monopolies–, while disposing of resources to innovate, would diminish their profits if they bring their innovations into the market. Cohen and Keppler’s (1996) reinforce this result when observing that small firms account for a disproportionately large share of innovation sales. analyzing the case of Taiwan, Tsai and Wang (2005) find that size is an advantage reflecting on the value-added of the firm.

Many research efforts have been dedicated to analyze the relationship between networking and innovation performance while controlling for size. Their results are contradictory (Tsai, 2009: 765); some studies find a negative –or not significant– relationship of these ties with innovation, while others conclude that the relationship is positive. We argue that these conflicting views are in part the result of not distinguishing bet-ween formal and informal ties, a distinction that is not made by these studies. Following the conceptualization of the distinction between formal ties and informal ties

1 These two dynamics are nowadays known as Schumpeter I and Schumpeter II (Langlois, 2003).2 This effect was later called “replacement effect” by Tirole (1988).

3

Network diversity, internal capacities and innovative activity

proposed by Powell and Grodal (2005), we carry out our empirical exploration.

While most of the literature on the relationship between networks and innovation is based on hypotheses ex-plored for firms in industrialized countries, the analysis of this relationship in the case of a developing country deserves specific attention. This exploration demands the use of certain variables that in the case of develo-ped countries are of no relevance. When studying the impact of networks on the innovative performance of SMEs in a developing country, using patents or expen-diture in Research and Development, as a proxy for the ability to innovate, would not be appropriate. The number of patents developed by SMEs in a developing country and their expenditure in activities related to Research and Development are very low, and would not differentiate among firms that perform innovation efforts, perhaps marginal but nonetheless with eco-nomic consequences. This difference in the nature of innovations of developing countries demands another measure of innovation performance. It should also be

noted that the Research and Development to Gross Domestic Product ratio is 0.53% for Colombia (CaF, 2001), while it is 1.9 % in the European Union, and 2.59 % in the United States (OECD, 2005). There are also differences in the contribution of SMEs to the in-novation volume of a country. In the United Kingdom, the most innovative enterprises are the smallest and the largest ones, while the medium-sized firms are the less innovative (Pavitt, Robson and Townsend, 1987). according to the survey we use in this research, 75% of industrial innovations of all types– not necessarily major innovations – come from SMEs.

Section 1 reviews the literature on the relationship of innovation with networks and size. Section 2 discusses the main hypotheses and the methodology. Section 3 presents the main results, and analyzes the effects of different types of networks on the performance of SMEs. Section 4 shows a comparative analysis of the relationship between networks and innovation in SMEs and large firms. Finally, we conclude and extract some policy implications of our quantitative results.

The hypothesis that collaboration with different partners should enhance innovation is widely accepted in the literature (Powell and Grodal, 2006). The diversity of the sources of knowledge is also attributed an influence on the ability of a firm to create new combinations of te-chnologies and knowledge. Metcalfe (1994) concludes that the variety of network links provides opportunities for the firm to choose among different technological paths. Powell and Owen-Smith (1999) found that the diversity of networks ties had a positive influence on patenting rates; and Powell and Grodal (2005) argue that ties with multiple or multifacetical organizations place the firm in a central position within its network, and centrally positioned organizations are more capable of receiving and transferring knowledge.

The ties of firms with universities, research centers, competitors, suppliers, customers and other agents can enhance the innovative performance of the firm, since they may allow the firm accessing assets, information and knowledge that otherwise it would not receive. In their analysis of Swedish manufacturing firms, Lööf

and Heshmati (2002) find that internal sources of innovation, and links with domestic universities and with competitors, have a positive effect on innovation output. They also find that relationships with customers have a negative association. In belderbos, et al. (2004a) the relationship with universities and research centers has strong positive effects on innovation, but coope-ration with competing firms has a weaker effect. Nieto and Santamaría (2007) obtain three important results: (a) internal R&D has a positive impact on innovation output, either with low or high ‘degree of novelty’; (b) belonging to certain sectors may have positive conse-quences, and (c) links with suppliers have the highest effect on “product innovation”. Taken together, the articles reviewed in this paragraph point out to some important potential links that should matter for deter-mining innovation results. Diverse networks provide access to sources of information and capabilities that are not internally available to the firm, and these links increase the levels of innovation (Powell and Grodal, 2005, p. 68).

4 Business Networks and Innovation in SMEs of a Developing Country

1.1. Formal and Informal Networks

Powell and Grodal (2005) define formal ties as those that are supported by contractual relationships (p. 60). They understand informal ties as those relationships emerging from shared experience (p. 63). as they stress, “This collaborative process generates and refines the intangible raw material of technical change–ideas.” (p. 71). Contractual relationships may be observed, as suggested by Callon (1991), in money-mediated exchanges.

Formal ties are of different types. Ties with public institutions have been analyzed by Dahlstrand and Cetindamar (2000), and by North, Smallbone and vic-kers (2002). The first of these studies found that “seed capital” for innovation in Swedish firms is provided by public funding. North, Smallbone and vickers (2002) conclude that, for british SMEs, funding coming from public institutions has only a partial effect on innovative results. Intuitively, public or private funding can help firms in their innovative processes both because financial resources are expected to have a positive relationship with innovation, and because the evaluation by agencies or banks implies a flow of (mainly explicit) knowledge. In this research, we explore the relative importance of formal ties with public and with private financial insti-tutions in determining innovation results.

Rouvinen (2002) establishes that product and process innovation in Finnish firms are enhanced with internal knowledge generation, along with formal cooperation with clients, suppliers and other non-academic part-ners. among the firms of this country, only product innovation benefits from universities and research organization.

Supporting the importance of formal networks, Man-cinelli and Mazzanti (2008) – in a study for a small province in northern Italy where SMEs are 90% of firms– demonstrated that cooperation with clients, su-ppliers, competitors and universities are complements of innovation effects on productivity. They conclude that networking alone is not sufficient to increasing productivity. They highlight the role of investments in R&D and training.

In emerging and developing economies, a number of researchers have explored the relationship between formal networking and business innovation.

In Chile, while few firms have cooperation agreements with research institutions, a positive relationship bet-ween innovative firms and links with public research institutions is suggested by descriptive statistics (OECD, 2005).

Wang, Ong & Lee (2005) find that internal innovation resources and employee innovative ideas are positively related to Singapore SMEs probability to innovate.They do not find size to have a significant effect on this probability.

Classic innovation determinants – such as size, edu-cation, age, market structure, industry and owners-hip – are explored by Lee & Ging (2007) in SMEs of Malaysia, and by Oluwajoba et al. (2007) in a sample survey of SMEs in Nigeria. The study on Malaysia finds that small firms with more employees are more likely to innovate. In Nigerian SMEs, education is strongly related to innovativeness.

Recent work by Tsai (2009) for the Taiwanese manu-facturing sector separately studies the effects of links with suppliers, clients, universities and R&D expen-diture per capita on innovative sales productivity. His findings indicate that collaboration with suppliers has a positive impact on this indicator. The effect of links with suppliers is especially important for SMEs, inas-much as these firms “face great competitive pressure” (Tsai, 2009). In firms with a high R&D expenditure per capita, links with customers are positively associated with innovation; if R&D expenditure per capita is low, this association is negative.

Previous contributions focus on either formal or infor-mal links, and they do not compare the effects of both types of ties. However it has been suggested that these ties have the potential to make significant contributions to innovation (Powel and Grodal, 2005: 70).

The relevance of the distinction between formal and informal links has yet to be explored. Our hypotheses are oriented to answer whether the effects of formal and informal links have differential effects on SMEs innovation.

1.2. Internal Capacities and Innovation

Internal capacities, as defined by Teece, Pisano and Shuen (1997) are presumably related to innovation performance. vega, Gutierrez and Fernandez (2009)

5

explore, for instance, the effects of external sources on innovation of different types and do not find synergies between internal technological capacities and external sources. aalbers, Koppius and Dolfsma (2006) focus on internal networks, and make a distinction between formal and informal communication networks within the firm, and prove that both types of internal networks contribute to knowledge transfer. Firm-specific assets can be sources of core competences by combining the appropriate complementary resources on distinctive ways of coordinating (Teece, Pisano and Shuen, 1997, p 516). In this paper we argue that the capacity of coor-dinating internal networks around innovation activities is a determinant of innovation output and may interact with formal or informal external networks.

1.3. Size and innovative activity

The relationship between size and innovation perfor-mance is matter of frequent disagreement. In the em-pirical literature, acs and audretsch (1987, 1990) find that small firms have a higher number of innovations per employee than larger firms in certain sectors of the USa, while Pavitt, Robson and Townsend (1987) observe that the most innovative firms in the UK are small or large, but not medium-size or micro.

In contrast, Syrneonidis (1996) surveys the literature on the subject and advances three reasons why innova-tive performance should grow more than proportionally with size: (a) larger firms can undertake several projects simultaneously and by doing this, they can diversify their risk in R&D; (b) the R&D projects involve high

fixed costs that a large firm can assume easier than a small firm; (c) access to funding is easier for the large firm. These arguments have been confronted by di-fferent authors. Wedig (1990) argues that innovation activities increase appreciably the systematic risk of a firm of any size, generating a reduction in their value, and this effect could over-ride the impact of size. In a study on australia, Rogers (2002) found that small manufacturing firms exhibit a stronger relationship bet-ween networking and innovation than large firms; he does not make a distinction between types of networks.. Dickson & Hadjimanolis (2000) conclude that small enterprises, as part of a management strategy, overcome their resources barriers by networking. Their study is based on 25 cases, and the issue is whether these results can be verified in a larger scale. Hall (1992) finds that both small and large firms are financially constrained for innovation. These contradictory views indicate that the relationship between size and innovation activities and results deserves further treatment, differentiating SMEs and large firms’ results and controlling by size within each segment.

External collaboration provides small firms a means to complement their own in-house efforts (ESRC, 2000). External links could provide SMEs a channel for commercializing innovations, access complementary marketing and distribution assets, and even deal with complex legal and regulatory issues. Our analysis will address empirically whether there are complementary effects between internal capacities and external networ-king, as suggested by the ESRC (2000) report.

Concepts, Hypotheses and Methodology

Our main purpose is to analyze the effects of networ-king on innovation performance of SMEs. To measure innovation performance, we consider three dependent count variables: (i) total innovations (product, process, organizational and marketing innovations); (ii) inven-tion-based product innovations, and (iii) imitation-based product innovations. The distinction between invention- and imitation based product innovations is particularly interesting, since imitation is of great importance in the industry of developing countries.

Network links of a firm are classified in two large categories: formal and informal. Following Callon (1991), in this paper we associate a formal link with a money-mediated contractual relationship. Conversely, when the firm simply recognizes one link as a source of ideas for innovation, without necessarily reporting a monetary flow, it is taken as an informal link. We are interested in the diversity of links rather than the absolute number of links, and for this reason we look at different types of links, depending upon the nature of the related organization.

6 Business Networks and Innovation in SMEs of a Developing Country

Important contributions in the literature have analyzed the impact of different types of external collaborators on innovation outputs. Many of these studies include size simply as a control variable (Tsai, 2009; belderbos et al., 2004b). In our exploration, size plays a double role: first, we focus on the behavior of SMEs; second, we consider size to be a moderator variable, in the sense that for different values of size, the coefficient relating external links to innovation performance takes different values, of a different sign.

The main research question we ask is:

Are there differences in the influence of formal ties on inno-vation, as compared to the influence of informal ties?

Derived from this main question, we analyze in more depth the relationship between networks and innovation, proposing answers to the fo-llowing subsidiary issues:

1) Do network links affect differently invention-based and imitation-based innovations in SMEs?

a product innovation may be based on an inven-tion (new to the world) or on an imitation (new to the firm or the country, but not to the world). This distinction is of particular interest to SMEs in developing countries.

2) Are the effects on innovation of internal capacities dependent on informal links outside the firm? Is this true for all types of innovations?

both internal and external sources of ideas are generating value within the innovation process. The issue is how they relate each other. Interac-tions between internal and external networks may have an impact on innovation outcomes. Further, the sign and significance of these inte-ractions could differ between invention-based innovations and imitation-based innovations.

3) Have the disaggregated formal and informal network links differential impacts on innovation performance?

Disaggregated links must be studied and their impacts compared among the three processes.

4) Do SMEs compensate certain limitations of their size, like investment capability or fixed costs, by develo-ping certain kinds of links more than larger firms?

The comparison between SMEs and large firms is important in determining if innovation dyna-mics are indeed different among firm sizes. The analysis includes the subsequent questions:

a) Is the sum of types of formal and informal links more important in SMEs than in larger firms, as a determinant of innovation output?

b) Is the sum of types of formal and informal links more relevant in SMEs than in larger firms, as a determinant of the probability that a firm be innovative (has at least one innova-tion)?

These hypotheses are operationalized in the following sections.

2.1. The Data

To explore our hypotheses, we use the 1995 Second Innovation Survey on Development and Technologi-cal Innovation in Colombian Industry (EDIT II). The survey gathers firm-level data from 6,670 formally registered businesses, including 5,252 SMEs3. The operationalization of dependent and independent variables follows.

Dependent Variables: measures of innovation per-formance

Many empirical studies of innovation use patent counts as the dependent variable, as reported and questioned by Powell and Grodal (2005: 69), or even R&D ex-penditure, as criticized by Nelson and Winter (1973: 46). In the case of a developing country, where patents granted are scarce, and accounting practices seldom differentiate between research and development and other expenditures, criticisms to these measures are reinforced.

Innovation counts of different types seem to be more appropriate for this situation, both theoretically and empirically. The ‘Innovation Results’ section of the

4 From 5,252 SMEs only 4,003 hold complete information on all the dependent and independent variables

7

Colombian survey, asks industrial firms for innova-tion results in four main categories: product, process, marketing methods, and organizational innovations. In this research, we use these counts, which are direct measures of innovation output, in the reference period of two years (2003 – 2004).

We build three indicators of innovation, which are relevant to the issues we intend to address:

(a) Total number of innovations obtained by the firm in the period of observation. It includes product innovations, whether based on invention or imi-tation activities, and process, organizational and marketing-method innovations.

(b) Product innovations based on invention. These innovations are declared as new for internatio-nal markets, and therefore they are presumed to result from invention activities.

(c) Product innovations based on imitation. These innovations are new for the firm or for the coun-try, but not for the world. Imitation activities are very important in innovation processes of indus-try in developing countries.

Independent variables

We group independent variables in five classes. The first four classes refer to external and internal networ-king and to their interaction. The last class groups classical innovation determinants, which are taken as control variables in this study.

While there is an important body of literature on social networks and innovation, in this paper we concentrate on innovation-related networks of businesses. We make a distinction between two types of innovation-related networking: (i) innovation-related links mediated by money flows, which imply a formal contractual relationship; (ii) innovation related links recognized as sources of innovative ideas, and not necessarily for-malized by a contractual money-mediated relationship. Social links are not considered in this study, since the innovation survey does not provide information on this type of networks.

Money-mediated links related to innovation activities include six types of links: Existence of financial links with government agencies, Existence of financial links with private banks, Existence of financial links with foreign orga-nization (foreign banks and multilateral organizations), Existence of financial link with clients, suppliers or partner companies, Existence of financial links with Universities or research centers, and Existence of financial links with firms belonging to the same economic group. Each of these types refers to different organizations that hold a money-mediated innovation relationship with the firm. In the detailed regressions, the six disaggregated formal links are studied as dummy variables. The dummies account for the existence and non existence of these six types of money-mediated links. Rather than a simple count of links, we count the number of types of links. For instance, if a firm has a money-mediated link with different governmental agencies, or with several private banks, the count is one in each case. This stresses the importance of variety of links, rather than their total number.

We are interested not only in the disaggregated effect of each type, but also in a measure of the variety of these formal ties. To build an indicator of variety of these links, the six items are added, and the norma-lized sum of these dummies indicating the existence of each type of money-mediated links is taken as an indicator of variety.

The five types of ties that the firm recognizes as sources of ideas for innovation, but which are not necessarily mediated by money or a formal contractual relationship are: Sum of types of informal links with clients, suppliers and competitors (normalized) recognized as sources of innovation; Sum of types of informal links with universities and research centers (normalized); Sum of types of infor-mal links with firms belonging to the same economic group (normalized); Sum of links with producers associations (normalized); Sum of tacit knowledge links (normalized). a Tacit-knowledge link refers to attendance to technology fairs and events, seminars and lectures. Each disaggre-gated variable was constructed by adding the existence (dummy) of different sources of ideas that belong to each type of informal link4. Then we normalize that sum for making all variables comparable on the re-gression analysis.

4 The Sum of types of informal links with clients, suppliers and competitors is equal to three when the firm holds ties with clients, suppliers and competitors; but it takes a value of one when the firm reports several ties of the same type (suppliers, for instance).

8 Business Networks and Innovation in SMEs of a Developing Country

We also define a variable that aggregates informal links named Sum of informal types of links (normalized). This variable is calculated by normalizing the sum of five dummies of existence of the five types of informal links. This indicator accounts for the variety of types of informal links.

among different internal capacities that the firm may have, the ability to mobilize and coordinate different firm resources around innovative activities is singled out as a critical internal capacity related to innovation. These resources may be: firm divisions, which are re-cognized as sources of ideas for innovation (R&D, pro-duction, sales, or marketing); interdisciplinary groups involved in innovation; the participation of executives or employees in innovation initiatives.

an indicator is built reflecting these aspects of the capacity of management to mobilize various resources for innovation. It aggregates the number of firm divi-sions, the existence of interdisciplinary groups and the involvement of executives and workers in innovation. This indicator is interpreted as the ability of the mana-gement of firms to coordinate their internal resources for innovation.

The interaction term refers to the moderator effect that informal links could have over the internal coordination capacities. In order to explore interactions between in-ternal capacities and external networking, two variables are built as the product of the normalized indicator of internal coordination capacities by the normalized sum of formal or informal types of links.

The last group of independent variables includes clas-sic innovation determinants that have been identified in the literature; (a) The Normalized sum of codified-knowledge recognized as sources of ideas: scientific or technical periodicals, web sites, scientific data bases, patent data bases, and intellectual property data bases; (b) a dummy variable that captures the Existence of Patent and License acquisition Contracts. It takes the value of one if the firm allocates funds to the acquisition of patent or license contracts, and zero otherwise; (c) the fraction of total employees holding a professional degree; (d) the fraction of total employees holding a technical degree; (e) the fraction of total employees in the R&D and design departments; (f) Sectors were clas-sified in three categories: capital goods, intermediate goods and consumption goods; the intermediate-good

sector is taken as the reference sector, and two dummy variables indicate the other sectors. (g) the last variable in this group is the size of the firm, measured by its number of employees, which is the size index available in the survey.

2.2. The Model

In the choice of a model, we faced two constraints. First, all dependent variables in the study are non-negative integer count variables. The validity of the assumption that they are Poisson distributed was checked. OLS regressions would yield biased estimators and, for this reason, we propose using Poisson regression analysis.

We also observe that, in a developing country, many firms exhibit zero innovations (more than 80% of the firms in the Colombian EDT II survey). This high pro-portion advises the use of the Zero Inflated Poisson (ZIP) models, rather than the regular Poisson Regres-sion Models (PRM), that are commonly used when the dependent variable is a count. The Zero Inflated Poisson model simultaneously runs a logit regression of the probability of a systemic or “strategic” zero outcome (Winkelmann, 2008). Its advantage is that the logit regression estimates the determinants of what can be interpreted as the probability that a firm has the potential to innovate.

For long, Poisson regressions have been employed in patent studies where the dependent variable is a count variable. Probit models are sometimes used to explain the probability of a firm being innovative according to some definition, but what the ZIP model does is different. It does not measure the probability of a firm having at least one innovation, but makes a distinction between two types of firms with zero in-novations: those having the potential to innovate as the count of innovations is always non-negative, as there are generally few innovation outputs, and many of the firms surveyed have zero innovation outcomes, the ZIP model minimizes biases in the estimators of the dependent variable, and simultaneously runs a regression on the probability of a firm being potentially innovative. a ZIP model estimates two simultaneous equations: one is a Poisson regression of the count va-lue of the dependent variable, and the other is a logit regression of the probability that a firm in the sample has a systemic zero count for that variable (Cameron and Trivedi, 1998).

9

as observed earlier, there is a large proportion of SMEs in the survey exhibiting a zero count of innovations. The Zero-Inflated Poisson regression [ZIP] makes it easier to identify corner solutions. The analysis of the inflate segment of the ZIP model reflects that corner solutions that are generated through processes are distinct from the processes leading to strictly positive results. From an economic point of view, this more complex procedure allows to simultaneously estimate the marginal effects of different independent variables on both the potential to innovate and the level of in-novation results.

When a standard Poisson regression model is assumed, it takes for granted that all firms are potential innova-tors, or that they have a positive probability to report a certain number of innovations. Notwithstanding, there are firms that do not have the potential to innovate or have strategically decided not to innovate. The ZIP model allows interpreting two different types of zeroes: the first type of firms does not have the potential to innovate; the second type have this potential, but for circumstancial reasons did not innovate during the period of observation.

Analysis and Results

all non-dummy variables are standardized. Correla-tions between all independent variables is presented in Table 3.1 When testing for moderator effects (baron and Kenny, 1986), we calculate interaction terms by

multiplying two standardized variables. Interaction terms used in this paper are constructed with products of the indicators of size, internal capacities and formal and informal networks.

Table 3.1. Correlations

10 Business Networks and Innovation in SMEs of a Developing Country

We estimate ZIP models of the following dependent variables: total number of innovations, invention-based product innovations, and imitation-based product in-novations. In all cases, we check for multicollinearity and the vIF is below 10. The adjustment measure used is McFadden’s R2. The maximum value of R2 obtained was 0.18, which is common in this type of studies5. Other variables not included in this study because they are not available in the survey could explain a large portion of the goodness of fit: lagged R&D investments of the firm, innovative history, market share and age, among others.

3.1. Innovation Networks and the SMEs

The first model, presented in Table 3.2, confirms the theoretical hypothesis that networking has a positive influence on innovation. While both the number of ty-pes of money-mediated (formal) links and the number of types of informal links positively and significantly affect the total number of innovation outcomes, their influence is not symmetric, and this difference shows that the distinction between formal and informal links may be important, a result which will be confirmed in the remaining of this paper. In this first exploration it is shown that informal links have a larger incidence on innovation results than formal links.

Table 3.2. Determinants of Total Innovation Outcomes in SMEs

5 In general, R2s in this type of studies are not high. Tsai (2009) reports between 10% and 40%, and Rogers (2004), 16% and 23%.

11

The inflated segment of the model shows that formal links have a negative effect on the probability of a firm systematically having zero innovations6. This is equi-valent to saying that the probability of an SME having the potential to innovate increases when the number of types of formal links is larger. This probability does not change significantly with the number of types of informal links. Formal links are therefore significant for raising the potential to innovate, but once a firm starts being innovative, these links have less importance than informal links in determining the number of total innovations the firm reports.

The direct effect of internal capabilities positively im-pacts innovation results and the probability of having the potential to innovate. Specification (b) of the model reported in Table 3.2 further demonstrates that, for SMEs, there is a negative moderator effect of the indica-tor of internal capabilities (reflecting the abilities of the firm to mobilize for innovation different departments, interdisciplinary groups, management and employees) on the relationship between external informal links and innovation output. This means that when the internal capacity indicator is larger, the impact of external in-formal networks on total innovation output is lower, indicating that a firm having developed internal coordi-nation capacities is less dependent of external informal links to improve its innovation performance.

The moderator effect is also significant in the inflate segment of the regression. This means that when there is low internal coordination capacity or few resources to coordinate involved in innovation, the impact of informal external links on the potential to innovate is greater. Informal links are therefore important when the level of internal capacities is low.

Control variables, such as patent acquisition, professio-nals’ ratio and relative firm size, augment the expected number of innovation outcomes, confirming what pre-vious studies have found, but they are not significant determinants of the probability of a firm having the potential to innovate.

3.2. Invention and Imitation in SMEs

Two types of product innovation are of particular in-terest for the analysis: invention-based and imitation-based product innovations. Does the development of an entirely new product (new for the world, as the survey defines it), require a radically different structure of internal resources and links than imitation-based products (new for the firm or the country, but not for the world)? Or, as suggested by Mansfield, Schwartz and Wagner (1981), the cost and perhaps the struc-ture of capacities and links is similar in invention and imitation?

6 The inflate segment of the regression refers to the probability of a firm having a systemic zero result. Therefore, the effect of an inde-pendent variable on the probability of a firm having the potential to innovate but not presenting innovations in the period observed has the opposite sign shown in the regression.

Table 3.3. Product Innovations Based on Invention in SMEs

continúa en la siguiente página

12 Business Networks and Innovation in SMEs of a Developing Country

Results in Table 3.3 (specifications a, b and c show the network and capacities structure for product in-novation based on invention. There is no significant moderator effect of internal coordination capacities on the relationship between external informal links and invention-based new products. The aggregate measure of internal capabilities is not significant in the Poisson segment of the model -specifications (a) and (b) in Table 3.3- for the count of inventions.

a deeper examination can be made in model (b) of Ta-ble 3.3. It shows that when informal links are excluded from the regression, internal coordination capacities are significant at a 1% level. Specifications (a) and (c) give us the clue that the relationship between internal capabilities and invention counts could be mediated by informal external ties.

To follow this clue, we explored whether this mediation effect exists. according to baron and Kenny (1986) we have a perfect mediation effect when the mediated variable –internal capacities, in this case– is a signifi-cant determinant of the mediator variable –informal external links–. We then show that β in the following regression is significant:

Mediator=α+α Mediated Variable+ α x+Std Errors.

Table 3.4 shows these regression coefficients. The normalized sum of internal coordination capacities is significant when explaining the normalized number of informal ties, and the relationship is positive.

Viene de la página anterior

Table 3.3. Product Innovations Based on Invention in SMEs

13

Table 3.4. Number of Informal Ties as a function of Internal Capacities in SMEs

We conclude that the relationship between the indicator of internal coordination capacities and invention-based new products is mediated by the effect of informal ex-ternal links. The capabilities of mobilizing ideas from managers, workers, different divisions and interdisci-plinary groups do not affect the expected number of invention-based new products directly; however, they do so through informal links. This result is very im-portant because it shows, together with the moderator effect found on total innovation outcomes, different manners in which internal and external knowledge providers interact, and subsequently improve innova-

tion and inventions outcomes. These interactions and mediations between internal coordination capacities and external informal links have not been previously considered in the literature7.

The impact of formal and informal links on imitation-based product innovations is shown in Table 3.5. Though some authors have emphasized the similarity between imitation and invention activities, we find that there are no moderator or mediator effects of informal links in the relationship between internal capabilities and imitation-based product outcomes.

7 Vega–Jurado et al. (2008) analyze the effects of R&D expenditure, human resources, R&D personnel and external networks, but do not analyze the ability to mobilize internal resources for innovation we are referring to. They evaluate interactions between R&D expendi-tures relative to the sector and R&D expenditures relative to other expenses within the firm.

14 Business Networks and Innovation in SMEs of a Developing Country

Table 3.5. Product Innovations Based on Imitation in SMEs

3.3. Disaggregated Analysis of Types of Links

Disaggregating the indicators of formal and informal links enriches the analysis of the relationship between networks and innovation performance. Results presen-

ted in Table 3.6 suggest that each type of the link has a different relationship with innovation outcomes.

15

Table 3.6. Innovation Outcomes in SMEs–Disaggregated Analysis

16 Business Networks and Innovation in SMEs of a Developing Country

We observe that SMEs that obtain a higher number of innovation outcomes are relatively larger than other SMEs. They have access to private bank funding; they combine the inputs from several firm divisions in their innovation processes; and they share innovation funding and ideas with other members of their value chain –suppliers, clients, partners and competitors–.

Companies linked through property links do not have any significant impact on the innovation outcomes, or on the probability of being innovative. Consistent with the control argument of the financial literature, this finding might reflect the business group structure, where there may exist centralized innovation efforts: a different firm in the group would carry out the R&D activities, providing innovations to other firms in the group.

The potential of being innovative is determined and positively influenced by money-mediated (formal) ties with Government agencies and with private banks. It is also positively related to the internal capacity of mana-gement to involve workers and managers in innovation, and to tacit knowledge links.

The larger the fraction of employees holding a pro-fessional degree, the higher the number of total in-novations. It is interesting to note that, though the fraction of employees holding a technical degree is not a significant determinant of the number of innovations, it contributes to explain the potential to innovate of a SME. The existence of a formal R&D department does not anticipate a solid innovative process; it is not a significant variable explaining the number of innova-tions. acquiring patents or licensing technologies helps the innovative firm to increase its innovation output. Nonetheless, it is not a determinant of the potential of

the firm to initiate innovation activities.

The main finding of this section may have policy implications. a SME significantly increases the proba-bility and potential to become innovative when it has a money-mediated link with a Government agency. actually, it is the third factor in importance that deter-mines the probability of a firm acquiring the potential to innovate. In contrast, once a firm has started to in-novate, Government funds do not make the difference for increasing its innovation outputs. a reallocation of Government funds towards firms seeking to develop their potential to innovate is recommended.

Comparing the regressions describing the impacts of disaggregated types of links on invention-based and imitation-based innovations also yields interesting re-sults (Table 3.7). The most important money-mediated links are those with private banks, suppliers, clients, competitors and partners. Informal innovation links with partners in the value chain (i.e. suppliers and customers), and sources of codified information, con-tribute to a higher number of invention-based new products.

Patent and license acquisition determine the poten-tial of innovation among non-innovative firms. Tacit knowledge links, such as those established in fairs and seminars, raise this potential, presumably because in those events they can benchmark with other firms. The fraction of professionals and the relative size of the SME positively affect invention-based product innovations.

Patent and license acquisition, internal coordination capacities and the fraction of technical employees are important determinants of imitation counts.

17

Table 3.7. Determinants of Invention-Based and Imitation-Based New Products– Formal and Informal Links

18 Business Networks and Innovation in SMEs of a Developing Country

The comparison between the determinants of imitation-based and invention-based product innovations yields important results. The most important formal link for invention-based and imitation-based new products, for inventive and imitative firms, is private bank’s fun-ding. The money-mediated links with universities are not a significant determinant of product innovation of either type, a finding coinciding with that of Lööf y Heshmati (2002) for Sweden. Foreign funding is not significant in the innovative or inventive processes in the SMEs; nonetheless, it reduces the probability of a firm becoming an imitator, a result perhaps related to the policies of the funding source. It is important to highlight that for the three dependent variables – total innovations, invention-based and imitation-based new products – government funding for innovation favors the potential to innovate.

Informal links with value chain partners are not signi-ficant determinants of imitation-based new products. It is important to notice that a non-innovative SME

could build its potential to innovate by acquiring pa-tents and licenses, as well as benchmarking through tacit and codified knowledge. Internal capacities of coordination for innovation significantly determine total innovations, and invention-based and imitation-based new products.

3.4. The contrast between SMEs and large firms

according to acs and audretsch (1987), the determi-nants of innovation may vary between small and large firms. This analysis is relevant in order to revise the fourth subsidiary issue, whether SMEs compensate certain limitations related to their size – like investment capability or fixed costs – by developing certain kinds of links more than large firms.

To explore this issue, we compare the average number of types of links for SMEs and large firms and find that large firms have a higher average. (see Table 3.8).

SMEs exhibit a higher number of total innovations, and of product innovations based on invention or imitation per employee (see Table 3.9). The second part of Table 9 shows that SMEs have more formal and informal links per employee.

Table 3.8 Average Values of Variables: SMEs and Large Firms

Table 3.9 Innovation Results and Links per employee in SMEs and Large Firms

19

Comparing the regressions for SMEs (Table 3.2) and for large firms (Table 3.10), we find that formal (mo-ney-mediated) links and internal capacities boost up innovation outcomes for both categories of firms, but informal links are significant only for SMEs. also, the

interaction between internal coordination capacities and external informal links is not significant for large firms, while it is for SMEs. Informal innovation links are thus important determinants of SMEs innovation, while they are not significant for large firms.

Table 3.10. Total Innovation Outcomes in Large Firms

The results for invention-based and imitation-based new products for large firms are presented in Table 3.11. Only 21.2% of SMEs and 36.3% of large firms

report product innovations new to the world, which can be considered invention-based.

20 Business Networks and Innovation in SMEs of a Developing Country

Table 3.11. Determinants of Product Innovations based on Invention and on Imitation in Large Firms

21

The sum of types of formal (money-mediated) links affects positively innovation, invention and imitation outcomes. aggregated informal links are not signifi-cant when explaining total innovation output of large firms but, for SMEs, they are significant as explanatory variables of innovation output and of the potential to be innovative.

SMEs and large firms show a close relationship bet-ween internal coordination capacity and the counts of innovation and potential to innovate.

Control variables show the expected results. acquiring patents is a significant determinant of the number of total innovations and of imitation-based new products among SMEs. SMEs that report innovations have a larger ratio of workers with a professional degree. The bigger this ratio, the larger the expected number of total innovations, and of invention- and imitation-based new products. The probabilities of having the potential to invent and to innovate increase with this ratio. In contrast, the ratio of technical personnel to total number of workers has no influence on inventive behavior in SMEs, although it has a positive effect on the probabilities of having any kind of innovation and imitations. an effort to reorient technical education towards supporting innovation processes is the ensuing policy recommendation.

Consistently, a SME belonging to the capital-goods sector is more likely to have a higher number of in-novations and of invention- and imitation-based new products than an SME in the intermediate-goods sector. In contrast, a larger number of inventions are expected in large firms of the consumer-goods sector, as com-pared to the intermediate goods sector. The expected number of innovations, invention- and imitation-based new products is also larger when the firm belongs to the capital-goods sector.

Few firms indicate they have workers dedicated to formal R&D activities. Nonetheless, the existence of formal R&D activities increases the expected number of invention-based new products among SMEs. Only 25.8 % of SMEs report R&D personnel, compared to 43.68% of large firms reporting at least one worker in the R&D department.

The disaggregated analysis in SMEs and large firms (see Tables 3.6, 3.7 and 3.11) shows that, for innovative firms, funding by private banks and internal coordi-nation capacities are factors that boost the expected number of innovation outputs. a larger number of the variables explored are significant for SMEs. Go-vernment agency financial links, money-mediated and informal links with suppliers, clients, competitors and partners, and formal links with research organizations are significant in SMEs regression but are not in large firms’ innovation counts.

The existence of links with public agencies is not significant in the large firms regression, but it is for enhancing the potential to innovate of SMEs. This result is in contrast with the fact that Government agencies dedicate funds 9 times larger to large than to small and medium enterprises. Government funding is 11.8 % of total funding of innovation in SMEs, and 15.5% in large companies.

Lins with private banks is the most important formal link, both for SMEs and large companies. It is positi-vely related to total innovation, invention-based new products and imitation-based new products in both categories of firms. It also raises the probability of a firm having the potential to innovate and to invent in both categories. For large firms, links with private bank funding is the only formal link that impacts their innovative results; for SMEs, it enhances the number of expected innovation outputs, as well as the probability of having the potential to innovate.

22 Business Networks and Innovation in SMEs of a Developing Country

Table 3.12. Relative Importance of Different Types of Links in Large Firms

23

Links involving money flows with suppliers, clients and partners have a positive effect on innovation and imitation-based new products in SMEs, but have no significant effect on the large firm innovative perfor-mance. Money-mediated collaboration with research organizations increases the probability of a SME ha-ving the potential to innovate, but has no effect on the expected innovation, invention or imitation outputs. additionally, funding from ownership-related com-panies is not a significant determinant of innovation output or potential to innovate.

Informal links, such as collaboration with suppliers, clients and competitors, are not significant determi-nants of innovation output in large companies. In contrast, informal links with suppliers, clients and competitors are significant and positively related to

innovation and invention-based new product perfor-mance of SMEs. Informal relations with research orga-nizations have no significant influence on innovation output or innovative probability of any kind, neither in large firms nor in SMEs. Property ties appear to matter for the probability of a large firm becoming innovative. Connections with producer’s associations and chambers of commerce do not explain innovation performance, despite the existence of some innovation programs of these organizations in the larger cities.

Tacit knowledge links, such as attendance to seminars and fairs, have significant and positive effects on the innovation counts and probabilities of SMEs but not of large firms. Likewise, codified information only affects innovation counts among the SMEs. These results have immediate policy implications.

Conclusions

The literature on innovation networks generally agrees that firms developing links with other organizations improve their innovative performance. In the extreme, as some authors state it, there are no isolated innova-tors, and both SMEs and larger industrial firms seem to benefit from networking.

The agreement is weaker, though, when the discussion moves to the type of links that are more effective in promoting innovation outcomes. Whether the more effective ties are “strong” or “weak”; formal or infor-mal; financial, among competitors or with suppliers or clients, are complex matters that can only be defi-ned in specific circumstances and through empirical observation. In this study, we have focused on the innovation networks of small and medium enterprises (SMEs) in Colombian industry, and have compared this relationship with that observed in large industrial firms of the same country.

Our results confirm the importance of informal links for innovation in SMEs; the existence of nonlinear effects of informal links on their innovation performance, and the critical role of diversity in external links. We de-monstrated the similarity between the effects of formal links on innovation, in SMEs and large firms, and the

significance of public funding for the probability that a SME gets involved in innovation. also, we found no evidence supporting the existence of a minimum cri-tical size required to engage in R&D cooperation. We explain these results in the following paragraphs.

Informal links are a significant determinant of inno-vation results in SMEs; they increase the number of expected innovations and raise the probability of ha-ving the potential to be innovative. In contrast, among large firms, informal external links are not significantly related to innovation performance. This result can be explained by the outstanding importance of internal coordination capacity in the large firms’ regression, relative to all other independent variables explored.

Nonlinear effects are evident in the role of informal links as a mediator variable in the relationship between internal coordination capacities and invention-based new products. also, informal links moderate the effect of internal coordination capacities on the sum of all types of innovations. The interaction between external links and the capabilities to mobilize internal resources around innovation activities is central to our results. Rather than concluding that these internal capabilities are more or less important than external

24 Business Networks and Innovation in SMEs of a Developing Country

sources of knowledge or that they complement each other, as in a traditional debate, we have found that the interaction between internal and external sources of innovation have a significant effect on the innovation performance of SMEs. The negative moderator effect indicates that the firm faces a strategic choice between internal capacities of the organization to gather the efforts of different departments and personnel around innovation, and the use informal links for innovation. This finding implies that studies of innovation should take into account the interaction between internal and external sources of knowledge and internal and external networking.

The number of types of informal ties also mediates the relationship between internal coordination capacities and SMEs’ invention-based new products. This means that internal coordination capacities do not have a direct impact on invention-based innovation counts in the case of SMEs. They are important only when external informal links exist.

We found evidence confirming the importance of network diversity. When links are of diverse types, they have a larger impact on innovation. There is a general positive statistical relationship between a higher number of types of links and innovation performan-ce. This result supports the hypothesis that firms are seeking to complete diverse types of resources, skills and knowledge capabilities through network links, in order to use them in the innovation process.

Taken together, the number of formal (money-mediated) links has a positive and significant impact on innovation counts of SMEs and of large firms. an important part of these formal links are those with financial sources, and managing a portfolio of diverse sources of innovation funding seems to be an efficient strategy. For SMEs, external formal links raise the probability of having a potential to innovate and also increases the expected number of innovations. On the other hand, formal links do not enhance the innovative potential of large firms. One possible explanation is the larger magnitude of investment funds that large firms can allocate to R&D without depending on external links.

a disaggregated analysis of formal and informal links shows that networks affect in a different way SMEs and large firms. Different types of network associates serve

different purposes in the process of innovation. Formal links with universities and other research organizations help the small and medium firm to initiate innovation based on imitation, and are not significant in large firms. Informal relations with suppliers, clients and competitors increase the number of total innovations and of new products based on invention. Collaboration of SMEs with value chain partners is strongly asso-ciated with both, the development of new products based on invention, and with the overall innovative performance of the firm. Funding by private banks is the most important financial source associated with innovation, both in large enterprises and in SMEs. It has a positive influence on the number of inventions and on the firm’s probability of having the potential to put in the market products that are new to the world (invention-based). among SMEs, the probability of a firm having the potential to innovate increases signifi-cantly with private funding.

Financial resources are not always allocated where they can be more productive. The largest portion of public funding is directed to large firms, a size segment where the effect of Government funds is not significantly rela-ted to innovation output. In contrast, public funding is significantly related to the performance of SMEs. Public funding remarkably determines the probability that a SME gets involved in major innovation processes such as those based on inventions. Since inventions are par-ticularly difficult to evaluate in financial terms, private funding sources may be avoiding risky investments in early stages of the innovation history of an SME. This is where public funding evidently plays a major role. Government institutions should evaluate their finance allocation policies in consequence, allocating a larger share of their credit resources to firms seeking to de-velop their potential to innovate.

Finally, in the specific circumstances of this developing country, we found no evidence supporting the existence of a minimum critical size required to engage in R&D cooperation, as belderbos et al. (2004b: 1256) argue, based on the experience of a different type of country. The external links of SMEs in this developing country are relatively more significant and play an important role of compensating for internal deficiencies in innova-tion resources than they do in the case of large firms.

25

References

acs, Z. and audretsch, D. (1987). Innovation, market structure and firm size. Review of Economics and Statis-tics, vol. 69, pp. 567-575.

ancona, D. and Caldwell, D. (1992). Demography and design: predictors of new product team performance. Organization Science, pp. 321–341.

arrow, K. (1962). “Economic welfare and the allocation of resources for inventions”, in: The Rate and Direction of Inventive activity, edited by Nelson, R. Princeton University Press, New Jersey, pp, 609-625.

barney, J.b. (1991). Firms resources and sustained competitive advantage. Journal of Management, vol. 17, pp. 99–120.

baron, R.M. & Kenny, D.a. (1986). “The Moderator-Mediator variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Consi-derations”. Journal of Personality and Social Psychology, vol. 51, pp. 1173-1182

baruch, Y. (1997). High technology organization: what it is, what it isn’t’. International Journal of Technology Management, vol. 13, No. 2, pp. 179–195.

belderbos, R., Carree, and Lokshin, b. (2004a). Coo-perative R&D and firm performance. Research Policy, vol. 33, pp. 1477– 1492

belderbos, R., Carree, M., Diederen, b., Lokshin, b. and veugelers, R. (2004b). Heterogeneity in R&D cooperation strategies. International Journal of Industrial Organization, vol. 22, pp. 1237– 1263.

CaF - Proyecto andino de Competitividad. Sánchez, F.and acosta, P. (2001) Proyecto Indicadores de Com-petitividad: Colombia. Documento de Trabajo CaF, Colombia.

Callon, M. (1991). “Techno-economic networks and irreversibility” in: a Sociology of Monsters: Essays on Power, Technology and Domination, edited by John Law. London: Routledge, pp. 132-165.

Cameron a. C. and P. K. Trivedi. (1998). Regression analysis of Count Data. Cambridge University Press, Cambridge, UK, 441 p.

Cohen, W.M. and Klepper, S. (1996). a reprise of size and R&D. Economic Journal, vol. 106, pp. 925–952.

Cohen,W.M., Levin,R.C. and Mowery,D.C. (1987). Firm size and R&D intensity: a re-examination. Journal of Industrial Economics, vol. 35, pp. 543–563.

Corredor, S., Forero, N. and Forero, C. (2009). “¿Cómo se innova en Colombia?” in: Tendencias de la adminis-tración en Colombia: presente y futuro. Universidad de los andes, School of Management. Ediciones Uniandes. Currently under review.

Dahlstrand, a. and Cetindamar, D. (2000). The dyna-mics of innovation financing in Sweden. Venture Capital, vol. 2, No. 3, pp. 203 – 221.

Dickson, K. and Hadjimanolis, a. (2000). Innovation Strategies of SMEs in Cyprus, a Small Developing Country. International Small Business Journal, vol. 18, No. 4, pp. 62-79

Drejer, I., Jorgensen, b. (2005). The dynamic creation of knowledge: analysing public–private collaborations. Technovation, vol. 25, pp. 83-94.

ESRC Economic Social Research Council (2000). Eu-ropean Research, Technology and Development: Issues for a Competitive Future. Key research findings from the ESRC Research Programme: The European Context of UK Science Policy.

Friedrich, R.J. (1982). In defense of multiplicative terms in multiple regression equations. American Jour-nal of Political Science, vol. 26, No. 4, pp. 797–833.

Geroski, P. (1994). Market Structure, Corporate Per-formance and Innovative activity. Oxford University Press, New York.

Hall, b. (1992). Investment and research and develo-pment at the firm level: Does the source of financing matter? NbER Working Paper No. 4096

26 Business Networks and Innovation in SMEs of a Developing Country

Kogut, b. (1988). Joint ventures: Theoretical and Em-pirical Perspectives. Strategic Management Journal, vol. 9, pp. 312-332.

Langlois, R. (2003). Schumpeter and the obsolescence of entrepreneurship and the entrepreneur. Advances in Austrian Economics, vol. 6, pp. 287–302.

Lee, C. and Ging, L. (2007) SME Innovation in the Malaysian Manufacturing Sector. Economics Bulletin, vol. 12, No. 30 pp. 1-12

Lööf, H. and Heshmati, a. (2002). Knowledge capital and performance heterogeneity: a firm-level innovation study. Production Economics, vol. 76, pp. 61–85

Mansfield, E.; Schwartz, M. and Wagner, S. (1981). Imitation Costs and Patents: an Empirical Study. The Economic Journal, vol. 91, pp. 907-918

Metcalfe, J. (1994). The economics of evolution and the economics of technology policy. Economic Journal, vol. 104, pp. 931–944.

Nelson, R. and Winter, S. (1977). In research of a useful theory of innovation. Research policy, vol. 19, pp. 193-214

Nieto, M. de J. and Santamaría, L. (2007). The im-portance of diverse collaborative Networks for the novelty of product innovation. Technovation, vol. 27, pp. 367-377.

North, D., Smallbone, D. and vickers, I. (2001). Public Sector Support for Innovating SMEs. Small Business Economics, vol. 16, pp. 303–317.

OECD (2005). OECD Economic Surveys: Chile – En-couraging innovation. OECD Publishing / Ministry of Finance, Chile.

Pavitt, K., Robson, M., and Townsend, J. (1987). The size distribution of innovating firms in the UK: I945- I983. Journal of Industrial Economics, vol. 35, pp. 297-316.

Powell, W. and Owen-Smith, J. (1999). “Network position and Firm performance”, in S. andrews and D. Knoke (eds.), Research in the Sociology of Oganzia-tions. Greenwich, Conn. JaI Press, 16: 129-59

Powell, W. and Grodal (2005). “Networks of Innova-tors” in: Oxford Handbook of Innovation, edited by: Fagerberg, J, David C. Mowery and Nelson, R. Oxford: Oxford University Press. Chapter 3.

Rogers, M. (2004). Networks, Firm Size and Innova-tion. Small Business Economics, vol. 22, pp. 141-153.

Rouvinen, P. (2002). Characteristics of Product and Process Innovators: Some Evidence from the Finnish Innovation Survey. Applied Economics Letters, vol. 9, pp. 575-580.

Scherer, F. (1992). Schumpeter and plausible capitalism. Journal of Economic Literature, 30, pp. 1416-1433.

Scherer, F.M. (1965). Firm size, market structure, op-portunity and the output of patented inventions. Ame-rican Economic Review, vol. 55, no. 4, 1097–1125.

Schumpeter, J. (1912). Teoría del desenvolvimiento económico. Traducción española, Fondo de Cultura Económica,. 1944. México.

Schumpeter, J. (1942). Capitalism, Socialism, and Democracy. New York: Harper and brothers.

Syrneonidis, G. (1996). “Innovation, Firm Size and Market Structure: Schumpeterian Hypotheses and some new themes”. OECD Economic Studies No. 27.

Teece, D.; Pisano, G. and Shuen, a. (1997) Dynamic Capabilities and Strategic Management. Strategic Ma-nagement Journal, vol. 87, pp. 509-533

Tirole, J. (1988). The Theory of Industrial Organiza-tion. MIT Press, Cambridge, Ma

Tsai, K. and Wang, J. (2005). Does R&D performance decline with firm size?—a re-examination in terms of elasticity. Research Policy, vol. 34, 966–976

Tsai, K. (2009). Collaborative networks and product innovation performance: toward a contingency pers-pective. Research policy, vol. 38, 765-778.

vega-Jurado, J.; Gutiérrez-García, a.; Fernández-de-Lucio, I. and Manjarrés-Henríquez, L. (2008) The effect of external and internal factors on firms’ product innovation. Research Policy, vol. 37, pp. 616–632

27

vestergaard, J. (2005). Innovation and university interaction with industry in Colombia: policies, expe-riences and future challenges. Working Paper 40727 – World bank.

von Hippel, E. and von Krogh, G. (2006). Free Revea-ling and the Private-Collective Model for Innovation In-centives. R&D Management, vol. 36, No. 3, 295-306.

Wedig, G. 1990. How risky is R&D? a financial approach. Review of Economics and Statistics, vol. 72, pp. 296-303

Winkelmann, R. (2008). Econometric analysis of Count Data. berlin: Springer. 5th edition.

28 Business Networks and Innovation in SMEs of a Developing Country

30 Business Networks and Innovation in SMEs of a Developing Country