Innovation and growth in the Nordic economies (IGNOREd)

Size: px
Start display at page:

Download "Innovation and growth in the Nordic economies (IGNOREd)"

Transcription

1 Innovation and growth in the Nordic economies (IGNOREd) Study of the links between innovation and firm performance in the Nordic region Innovation input results in positiv productivity gains Public support, cooperation and foreign markets strongly influence innovation activity May 2008 Authors: Mark Knell, et al.

2

3 IGNOREd Innovation and Growth in the Nordic Economies

4 IV

5 Participants Norway, project leader NIFU STEP, Oslo Mark Knell and Svein Olav Nås Denmark CFA, Aarhus Carter Bloch and Ebbe Graversen Sweden KTH, Stockholm Hans Lööf Finland VTT, Helskinki Olavi Lehtoranta and Mariagrazia Squicciarini Iceland Rannis, Reykjavik Thorvald Finnbjørnsson Estonia University of Tartu Priit Vahter and Jaan Masso Reference group Peter Hanel University of Sherbrooke Canada Pierre Mohnen UNU-Merit, University of Maastricht Netherlands V

6 VI

7 Project page Title: Innovation and growth in the Nordic economies Nordic Innovation Centre project number: Author(s): Mark Knell, et al. Institution(s): NIFU STEP Norwegian institute for Studies in Innovation, Research and Education Abstract: The main objective of the project was to study the relationship between innovation and productivity growth in the Nordic region using firm level data and to show how this analysis could be used for targeting policies that promote better economic performance. It contained four components: (1) Identification and presentation of indicators through graphical and descriptive methods; (2) Design and application of an econometric model for policy analysis; (3) Evaluation of the relationship between innovative activities and productivity; and (4) Interpretation and use of the results to advise policymaking at the national and international levels. Using data from fourth community innovation survey, a Nordic econometric model was developed and estimated across all countries. The model confirmed that innovation output is positively and significantly related to innovation input and innovation input results in positive productivity gains. Innovation activity is also strongly influenced by public financial support, cooperation with other firms and institutions and exposure to foreign markets Topic/NICe Focus Area: Innovation theory and policy ISSN: Language: English Pages: 44 Key words: Innovation, R&D, productivity, firm performance, econometric models Distributed by: Nordic Innovation Centre Stensberggata 25 NO-0170 Oslo Norway Contact person: Mark Knell NIFU STEP Wergelandsveien 7 N-0167 Oslo Norway Tel VII

8 VIII

9 Table of contents Executive summary X 1. Introduction The idea The literature The Nordic region 4 2. The extended CDM model Some preliminaries The Nordic common model Some econometric issues Some alternative models Conclusions and policy recommendations Appendix: The econometric model Bibliography Background papers for the project 30 IX

10 Executive summary The main objective of this project is to study the link between innovation and productivity growth in the Nordic region using firm level data and to show how to use these analyses in targeting policies that promote better economic performance. Its analytical core was to design an econometric model that links innovation with productivity at the firm level. The analytical framework used draws on earlier work by Pakes and Griliches (1984) and Crepon, Duguet and Mairesse (1998). This framework conceives innovation as a process where inputs are transformed into economically useful knowledge, which is thereafter utilized in production. The approach followed in this project relies on innovation survey data, with a knowledge production function relating innovation expenditures to innovation output (innovative sales) and the use of innovation output to improve productivity (sales per employee). Analyses carried out in the IGNOREd research project centre mainly around the fourth innovation survey (CIS4), which was carried out in 2005 and covers the period from 2002 to Some of the analyses make use of information from the third innovation survey (CIS3), which was carried out in 2001 and covered the period from 1998 to The central idea behind innovation analyses is to increase understanding of the transformation process from inputs to impacts on productivity. Our framework allows us to analyse four different stages of the innovation process: the decision to innovate; investments in innovation; the impact of innovation activities on innovative performance (or the knowledge production function); and the impact of innovation output on overall productivity. There are two broader implications for policy design. First, highlighting and analysing the full innovation process is important both in confirming the need for a broader approach to innovation policy and in identifying important factors for each stage of the innovation process. The measures needed for example to increase the share of innovative firms, increase innovation activity or improve the gains of innovation may be quite different from one another. This emphasizes the need for analysis of each of these four stages. Second, to strengthen and inform the case for a broader focus of innovation policy, reliable, internationally comparable data is needed. The successful analysis across six countries lends strong support to the usefulness of innovation data for policy. There are positive, significant impacts of innovation expenditures on innovation output in all countries. However, coefficient estimates vary more here than for innovation input, with Denmark having the lowest impact and Norway the highest. At the same time, the analysis also points out limitations to innovation data in its current state. The common model estimated in all countries focuses on determinants of innovation, such as hampering factors and cooperation. Hampering factors have a positive impact on the decision to innovate. At first glance, this result would seem counter-intuitive, though there may actually be a reasonable explanation to this. What this result appears to indicate is that knowledge; cost and market barriers are not major factors that prohibit firms from innovating. X

11 On the contrary, the experience of these barriers is greater for innovative firms, hampering the size of innovation investments and their success. This may also reflect a lack of awareness of these factors among noninnovating firms. An implication is that the targets in tackling these barriers for policy should be to increase innovative effort and gains of innovative firms, and less to encourage innovation among non-innovating firms. Hampering factors become in general negative when looking instead at effects on innovation output. In particular cost factors have a strong negative impact on innovation performance. This suggests that key factors slowing innovation performance are lack of financing (internal or external) and high innovation costs. Coefficients are negative for market barriers across countries, but only significant for Finland. In contrast no evidence is found of an effect of knowledge barriers. This can be contrasted with the high policy focus on increasing the supply of knowledge intensive labour, and on improving access to information on technology and markets. However, one possible interpretation is that knowledge barriers may impact innovation investments, but have less impact on the productivity of those investments. Cooperation is an important policy target in the Nordic countries, with many R&D programmes designed to promote networking and cooperation. Our results were mixed, which may The Nordic common model reflect that it appears to have little impact on innovation performance. However, there appears to be a strong, positive relationship between cooperation and innovation intensity. Cooperation thus appears to have an indirect impact on innovation performance through increases in investments. But the evidence is a bit more mixed across the Nordic countries. This may be due to national differences in the innovation system and it may depend on the cooperation partner. Participation in foreign markets is an important positive factor. This suggests that to increase the share of innovative firms, policies to promote internationalisation may be a route of action. XI

12 More detailed analysis of the data provides some interesting results. Further analysis of the Danish data finds that the impact of innovation output depends to a large degree on how innovative firms are. For firms with product innovations that are new to international markets, impacts on innovation performance and productivity are greatest. This may indicate both that operating on international markets provides greater potential for innovations and that it may provide greater incentives to engage in novel innovation activities. By contrast, firms with novel innovations that only operate on national markets, impacts are much lower. Finally, both customer driven and supplier driven innovation have a positive impact on innovation output. Close focus on value chain partners thus seems to lead to more successful development and implementation of product innovations. Both these drivers also have a positive impact on productivity. Comparing innovation surveys across time would also allow the results to be set in the context of macroeconomic conditions and see how innovation is affected by business cycle fluctuations. A study of Estonia shows that only process innovations had a positive significant effect on labour productivity, and not product innovation in the CIS4 survey, whereas the CIS3 survey provides the opposite result. One reason for this may be that during the first period product innovation might have been necessary for firms to restructure and enter new export markets after the loss of traditional export markets in the Russian crisis, while in the 2nd period growing labour costs made it more important to reduce production costs through process innovation. In the early 2000s, process innovation might have also been necessary to increase production in order to meet the growing demand during the period of strong macroeconomic growth. The analysis would benefit from having more information about non-innovators. Another important issue that would need to be addressed is simultaneity. At present the innovation questionnaire asks about innovation inputs and innovation output that all occur at the very same time. In reality outputs follow inputs in time. Hence, without the possibility to account for the necessary time lag needed for inputs to have an effect on outputs the analysis suffers from this inability to control for time-related dynamics. To obtain additional information there might be two possible solutions: one is to ask additional questions in the survey; and another would be to merge additional data. There are pros and cons to both. Changing the coverage, frequency and timing of the survey is a difficult and time consuming process, but would improve the quality and quantity of information gathered, thus enabling better estimations. Merging CIS data with data gathered from external sources is a good solution if statistical offices are willing to match the data with other sources at the firm level and if these other types of innovation-related data are available. The next question, and too lengthy to discuss here, is what to add. The most obvious candidates are however investment, use of ICTs, skill level of the workforce, exports and financial accounts. XII

13 1. Introduction While the full potential of innovation surveys based on the Oslo Manual is far from being realised, they prove to be fundamental tools in the effort to improve our understanding of the relationship between innovation and firm performance. Innovation surveys provide both researchers and policy makers with a wealth of information and data related to firms innovative activities, including inputs, outputs and other determinants. The main objective of this project is to use firm level information collected in these surveys to study the link between innovation and productivity growth in the Nordic region. We use this analysis to formulate policies that promote better economic performance. To fulfil this objective, the research addresses the nature of the relationship between innovation and productivity; examines the factors that may affect the outcomes; investigates some of the puzzles in the relationship emerging out of the estimates; and identifies possible policy measures that can improve the effect of science, research and innovation on productivity and economic growth. This chapter provides an introductory background to some of the issues being discussed in the report. It starts by outlining some of the basic ideas underlying the IGNOREd project. Section two describes some of the important literature underlying the analysis and section three provides a brief overview of the Nordic region. 1.1 The idea Using the example of a pin factory, Adam Smith (1776) described how the division of labour could increase the performance of an enterprise and hence the wealth of nations. While this illustration may be found in almost every industry today, advancement of the idea that innovation may lead to productivity growth has been sometimes opposed and often discussed under various theoretical frameworks throughout the history of economic thought. Building on the idea of Smith, Joseph Schumpeter (1911; 1942) described how product, process and organizational innovation could affect the profitability of the firm and market structure. More recent developments in the theory of economic growth by Paul Romer (1986; 1990) established a link between research and development (R&D) activities and productivity growth. Wesley M. Cohen and Daniel A. Levinthal (1989) complemented this idea by suggesting that firms should carry out R&D activity to be able to absorb the technical knowledge developed by other firms. Further interesting contributions followed, often along more evolutionary lines of thought. 1 Applied economists have also shown considerable interest in the innovation productivity link over the past three decades. 2 Zvi Griliches (1979) was one of the first economists to introduce R&D capital stock as a factor of production into the production function approach to productivity pioneered by Robert Solow (1957). In this approach, R&D activities add to the existing stock of knowledge accumulated by the firms. This in turn leads to productivity growth, through product and process innovation. 1 See for example, Nelson and Winter (1974; 1982), Dosi et al. (1988), Freeman (1995), and Metcalfe (1998). 2 See for example, literature reviews by Nadiri (1991a), Griliches (1992), Mairesse and Mohnen, (1995), Cincera (1998), Kleinknecht and Mohnen (2002), Wieser (2005) and Knell and Rojec (2007). 1

14 Zvi Griliches Early models developed by economists affiliated with the National Bureau of Economic Research (NBER) incorporated variables into the production function intended to capture what Griliches (1979) called the knowledge capital. Little was said about what knowledge is, or of how it becomes important for innovation and growth, but it did provide a way to link innovation and productivity. Pakes and Griliches (1984) made one of the most important contributions to this literature, by developing a variant of the Griliches framework that allowed for the inclusion of several interrelated innovation inputs over time. These authors studied the relationship between R&D expenditures and patenting behaviour between 1968 and 1975 for a large number of firms. They found the sum of the contemporaneous and lagged effects to be positive and significant. 3 Crèpon, Duguet, and Mairesse (1998) adopted this methodology to link innovation to productivity. Known as the CDM model, it investigated the channels through which R&D activity influences innovation and productivity (for a cross-section of firms in the French manufacturing sector in the year 1992). Their 3 They also point out that patents are a flawed measure of innovation output, because not all innovations are patented. Nevertheless, Jaffe and Trajtenberg (2002) suggest that patent filings can be an important source of technology diffusion and spillovers. approach combined a knowledge production function relating R&D activity to patenting or innovative activities with economic performance as measured by labour productivity. It contains a system of three simultaneous equations where R&D activity and other factors generate new knowledge, which then propels innovation (output) and finally productivity growth. Other supply and demand factors, as well as sectoral differences and unobserved heterogeneity, are also included in the model to improve its explanatory power. These authors find evidence of a positive effect of R&D on innovation output, as measured by patents, as well as a positive and significant effect on the value-added per employee of the French firms. Their estimates are first based on a generalized Tobit that models firm research investment behaviours. They then adopt a two-stage procedure that relies on the generalised method of moments (GMM) and retrieve the estimates' coefficients through asymptotic least squares (ALS). Other papers have used alternative representations of firms' innovative activities, in an effort to use the innovation survey data in different ways. Chapter 2 and the appendix describes the modeling strategy followed in this study and describes how the method was used in the common model developed within the IGNOREd project. We visually schematise the basic Griliches CDM model on page 3. 4 In this figure, the green oval connects innovation inputs and innovation outputs, and represents the knowledge that is unobservable. Understanding what knowledge is has been an important issue for economic theorists, applied economists and policy-makers alike. 1.2 The literature The CDM model has influenced a new and burgeoning literature on the relationship between innovation and firm performance. Firm performance variables have included value-added, sales or exports per employee, the growth rate of value-added, sales, profitability or employment, sales margins, 4 The figure is adapted from Pakes and Griliches (1984) and Hall and Mairesse (2006). 2

15 and profit before and after depreciation (in both level and growth rate). The main finding of these studies is that, regardless of how performance is measured, innovation output positively and significantly affects firm performance. 5 Several important papers appeared recently that have either estimated the CDM model for other countries, or introduced variants of the econometric model itself. One of the first studies to use innovation survey data instead of R&D and patent data used in the original CDM model, Lööf, et al. (2002) shows that there is considerable variation in results between Finland, Norway and Sweden in the early 1990s. They argue that this variation may be due to data errors, the econometric model and its specifications (three stage least squares), and unobservable country-specific effects. Using survey data for France in 1993, Duguet (2000) shows that strongly innovative firms are much more likely to improve their total factor productivity than weakly innovating firms, and that the return to innovation increases with the degree of innovation opportunities firms have. The model also shows that the Solow residuals at the industry level are linked to radical innovations at the firm level. Janz, et al. (2004) was one of the first studies to pool data from different countries. Using the third CIS they pooled observations from Germany and Sweden and showed the existence of a strong link between innovation output and sales per employee in knowledge intensive manufacturing firms, independent of the country. Mohnen and Therrien (2003) compared Canada with selected European countries in the late 1990s. They found Canadian firms to be more innovative as a whole, but with a lower share of sales from innovative products for its innovative firms. These results led the authors to suggest that the national samples may not be representative and that differences 5 Klomp and van Leeuwen (2001) are an exception to this rule, as they find a negative but insignificant effect of innovation output on employment growth. Pianta (2005) points out, that empirical studies of the relationship between innovation and employment identify both a positive and a negative effect of the former on the latter. Innovation and Firm Performance in the questionnaire or perceptions of the questionnaire did matter. In 2006 a special issue of Economics of Innovation and New Technology was devoted to the CDM model (Hall and Mairesse, 2006) In this issue, Mohnen, Mairesse and Dagenais (2006) estimate the relationship between innovation output and firm performance using micro-aggregated data from seven countries (Belgium, Denmark, Ireland, Germany, the Netherlands, Norway and Italy) for They use a generalized Tobit model together with a framework accounting for the variation of production across countries. They include size, industry, ownership type, continuous R&D, cooperative R&D, R&D intensity, proximity to basic research, and perceived competition as independent variables, and find that firms productivity correlates positively with higher innovation output, even when correcting for the skill composition of labour and for capital intensity. However, they also emphasize that simultaneity tends to interact with selectivity, and that both sources of biases must be taken into account. 3

16 In the same issue, Lööf and Heshmati (2006) perform a sensitivity analysis of the different measures of firm performance, and find the same pattern of positive and significant effect of innovation output on firm performance. Van Leeuwen and Klomp (2006) use data from the Netherlands in 1997 to show that the impact of innovation differs depending upon the specific measure of firm performance used, and that additional information on the technological environment of the firm can improve estimation. Extending the analysis beyond Europe, Benavente (2006) applies the CDM model and its estimating procedure to Chile during the period 1995 to They find that R&D and innovation activities are related to firm size and market power, but that innovation output (or R&D activity) does not influence firm performance. By contrast, Jefferson, et al. (2006) - correcting for size, industry, profitability, and market concentration - show that there is a strong relationship between R&D intensity and new product sales and returns to R&D expenditures. 6 In another important contribution, Griffith, et al. (2006) estimate a variation of the CDM model for four European countries (France, Germany, Spain, and the UK), using firm-level data from CIS3, carried out in This model differentiates between the labour displacement effect of process innovation and the compensation effect determined by higher demand. They find that job loss due to process innovation is partly compensated by the displacement effect and that there is no evidence of a displacement effect when there is product innovation, even when old products are no longer produced. Although they find that, in general, results are similar across the four countries considered, employment effects differ. For example, there is no sign of a displacement effect from process innovations in Spain, whereas product innovation generates more employment in Germany and to a lesser extent in the UK. Finally, Criscuolo, et al. (2005) developed a similar type model using CIS3 data for the UK. Taking into account the nationality of ownership, they show that multinational firms and exporters generate more innovations than their domestic counterparts, not only because they use more knowledge inputs, but also because they seem more able to access the necessary knowledge through the links they have in the global economy. The literature on the relationship between innovation and firm performance is still at an early stage where econometric and measurement issues dominate the discussion. Yet, much has been accomplished, and obtaining quantitative evidence has been and will continue to have an important influence on shifting innovation policy discussions away from a narrow focus on R&D. Moreover, the quality and comparability of innovation survey data has improved greatly since the early 1990s, which increases the possibility for policy relevant analysis. Data confidentiality, however, remains one obstacle that makes it difficult to merge national database and to pool data across countries. 1.3 The Nordic region Denmark, Estonia, Finland, Iceland, Norway and Sweden are all part of the Nordic region. The region has a common history, strong economic interdependencies, and many common institutional arrangements, which is often described as the Nordic model. All the countries in the Nordic region are relatively small, with Sweden being the largest in terms of population, with just over 9 million inhabitants, and Iceland being the smallest, 6 Some analysis of less developed countries also exist. Using data from the World Bank Investment Climate Survey covering the years 2000 to 2002, Goedhuys, et al. (2006) shows that innovation output (or R&D activity) did not influence firm performance in Tanzania, but that the institutional arrangements had an important impact. 4

17 with just over 300,000 inhabitants. Estonia is often considered part of the region because it has strong linguistic, cultural and historical ties with Finland, Sweden and Denmark. Yet, the industrial structures of these economies are quite diverse, as well as the Gross Domestic Product per inhabitant in Purchasing Power Standards (see figure). Norway has one of the highest levels of income in Europe, whereas Estonia has one of the lowest. These Nordic countries are at the centre of the analysis carried out within the IGNOREd project, which attempts to verify the existence of and to quantify the innovation - productivity link. In this project, we mainly rely on the fourth Community Innovation Survey (CIS4), which was carried out in 2004 and covers the period Some of the analysis makes use of information from the third innovation survey (CIS3), which was carried out in 2000 and covered the period from 1998 to Norway is an exception in this instance because it carried out the survey in 2001 and covered the period from 1999 to Economic growth was rather strong during the two periods, but there was a slowdown in GDP growth in every country except Estonia when the CIS4 was carried out (Estonia instead experienced negative growth during the period covered by the CIS3. See figure below). might influence the analysis of innovation and productivity is the diverse industrial (and market) structures observed in the Nordic region. R&D intensity and other knowledge inputs into the innovation process can vary considerably across industries and will depend to a large extent on the strategic behaviour of firms (Sutton, 1998). With the exception of Norway, the Nordic countries were near or above the EU average in terms of the share of firms that are innovative. The industrial and market structure may have an important influence on the overall innovative activity of firms. Differences in the sampling method may also be a factor. The figure below shows the share of firms introducing products only, new processes only and both new products and processes. The Nordic countries lead EU R&D intensity. EU 27 Member States R&D intensity was on average about 1.8 per cent in 2004 (see figure). Norway was somewhat below average and Estonia was about half of the average. Sweden and Finland were the only countries to already reach the Barcelona target of spending 3 per cent of GDP on R&D activities by One important factor that 5

18 2. The extended CDM model The analytical core of the IGNOREd project was to design an econometric model based on firm-level data; the results of which could be used to target policies aiming to enhance productivity and economic performance. To accomplish this objective, the project experimented with different ways to use information available from enterprises about their innovative behaviour. It also investigated the applicability of the different econometric methods that can handle this type of survey data. In doing so, the project modified the CDM model to make it better suited to the Nordic region and to improve comparability across the six countries included in the study. The IGNOREd project has been running in parallel with and gained synergies from the OECD microdata project (OECD 2008), which in part has been addressing similar questions. This chapter describes some of the core issues that the project attempted to address. Some preliminary issues related to the innovation survey and indicators are addressed in the first section. Section two presents the results of the common model in a simple straightforward way. An appendix to this section describes the econometric model and the estimation methods used. Section 3 discusses some of the limitations of the model and some of the possible ways to improve the econometric model, especially in the light of policy evaluation. A final section summarizes some research done within the project that attempts to overcome these limitations. 2.1 Some preliminaries The fourth community innovation survey (CIS4), covering the period from 2002 to 2004, contains a rich amount of information on the inputs, the outputs and the characteristics of innovation activities. To provide a more comprehensive view of firms innovation processes over time, some of the individual country papers make use of the third community innovation survey (CIS3), as well as other intermediate surveys. Within the IGNOREd project we also merged detailed financial information about the firms included in the innovation survey and used these data in some of the country papers. CIS4 and CIS3 were designed to obtain information on innovation activities within the enterprises and cover various aspects of these processes, such as the sources of information used, the possible obstacle encountered when innovating, the costs of innovation, and the effects of innovation. CIS surveys are based on the second (1997) and third (2005) editions of the Oslo Manual, which provide definitions and methodological guidelines for the construction of the questionnaire. In the CIS questionnaire, firms were asked some general questions on their identity, total turnover, number of employees, industry classification and membership of an enterprise group. The survey then includes a series of questions that aim to distinguish innovating firms from non-innovators. All firms are asked questions related to hampering factors, that is, if they faced cost, knowledge and market barriers to their innovation activities; if and to what extent they make use of intellectual property rights; and organizational innovations that were implemented by the firm during the period considered. Firms that declared themselves to be innovative were asked additional questions. These relate to the knowledge inputs utilised throughout the innovation process; the investments in innovation firms made during the period considered, including own R&D activities and the acquisition of knowledge from outside sources; and finally about the innovationspecific cooperation firms had with different actors, such as suppliers, customers and universities. Information is also provided about the financial support firms might have received to innovate through public financing. Moreover, it also contains information on innovation outputs, such as the introduction of new products and processes, the share of 6

19 turnover due to new products, the effects of innovative activity on the enterprise, and the effects of organizational innovations. The requirement that only innovative firms complete the majority of the survey, that is, all the innovation-specific questions, hinders researchers ability to carry out a more precise assessment about the determinants of innovation and the innovation process itself. The amount of information available on noninnovators is of particular relevance for econometric analyses, both for modelling the decision to innovate and to control for selection biases (see below). In some cases, panels of innovative firms are maintained in samples over time. If firms are not randomly selected, for example, if some of the firms included in the sample are known to have been innovative in the past, innovators might end up being overrepresented in the sample. This would happen because innovative activities are persistent (Geroski, et al., 1997; Cefis, 2003), meaning that past innovators are more likely to be innovators in the future. Weighting procedures take account of this is tabulating aggregate statistics. However, this is more difficult to remedy in firm level econometric analyses. Econometric analyses focus on innovative firms, and the impact that their innovation activities have on economic performance. However, relying on a sub-sample (in this case, innovative firms) that does not correctly represent what happens in the entire economy may create a bias in estimates. To remove this bias, econometric models using CIS data need to take into account the firms decision to innovate and the possible systematic differences that may exist between innovators and non-innovators. This can be done by controlling for what is in econometrics defined as selection, that is by using a generalized Tobit model, which was proposed by James Tobin (1958) to address biases of this kind. In our model, we account for selection through a two-stage Heckman procedure, which appears as equations 1 and 2 in the appendix and in box 1 and 2 in the figure. The appendix describes the Tobit model and the Heckman procedure in more detail. To make better use of the survey statistics, we construct several composite variables resembling composite indicators that aim to capture aspects of the innovation process that are also and especially relevant for policy. Many questions in the survey, such as those related to the type and intensity of hampering factors to innovation firms encountered, as well as the cooperation activities put in place when innovating, and the factors that drive firms innovation, are not suited to be used as they are in the econometric analyses. However, as they contain very interesting and policy relevant information, we constructed composite indicators based on them. Among these are: (1) an index of overall cooperation; (2) a measure of private cooperation, defined as cooperation with suppliers, customers and competitors; (3) a measure of public cooperation, defined as cooperation with universities, government research institutes, and private consultants; and (4) the degree of international cooperation, relative to cooperation with national partners. 2.2 The Nordic common model The econometric model underlying the results obtained for the Nordic common model has a similar structure as the CDM model. It is a three-stage, four-equation model that starts from the decision to innovate, considers the factors that generate new knowledge (including investment in R&D), verifies the link between innovation input and innovation output and finally encompass the passage from innovation output to productivity. We estimate the model using STATA software. Many different variants of the CDM model were tested during the course of the project. The figure on this page summarizes the variant of the Nordic common model described in this section. This specification was chosen because of its more solid theoretical basis. It fit the data reasonably well across all countries, except for Iceland. In the case of Iceland this is likely due to the very small number of observations, which also made it necessary to streamline the industrial breakdown used in this very specific case. In this section we do not emphasize the comparable size of coefficients between countries, but rather focus on the level of significance and the sign of the contribution of each variable considered. 7

20 The first step of our model estimates the probability to invest in innovation and the amount being invested. We consider a firm to be innovative if it has introduced a new product from 2002 to 2004 and has positive sales deriving from this innovation sold on the market. This equation, shown as equation 1 in the appendix and the blue boxes in the figure on this page, captures the decision to innovate. Factors that are important to this decision include the firm size, membership in an enterprise group, being active in foreign markets, and various circumstances that might hamper the decision to innovate. We also control for industry differences, by introducing dummies in the estimated equation. This equation also estimates the inverse Mills ratio, which is then used as an additional regressor in the second and third step of the model, to control for selectivity. The inverse Mills ratio enables consistent estimates about the subpopulation considered in our case the innovating firms and corrects for the bias that may exist because we only observe activities of innovative firms and not from those who do not innovate. Results about the variables influencing the likelihood to innovate are shown to be very similar across the Nordic countries. The figure on this page summarizes the importance and signs of the factors that influence the decision to innovate and table 1 provides the estimates. Firm size and presence in foreign markets show to be important determinants of the decision to innovate. Hampering factors also appear to be positively related to the likelihood of being innovators. Although this result might seem somewhat counter-intuitive as it implies that the higher the intensity of the obstacles faced the more likely a firm is to indeed innovate it might be easily explained as follows. Innovative firms also face certain obstacles, but they are also more likely to be aware of them than firms that have not innovated. This may reflect either lesser awareness of potential for innovation among non-innovative firms or less incentive to innovate based on factors not observed here. Since innovating firms are more aware than non-innovators about the obstacles encountered, innovators are also more likely to be affected by them, thus yielding a positive coefficient in our estimates. The pattern The Nordic common model observed is consistent across the Nordic countries apart from the negative but insignificant coefficient for Iceland (because of too few observations). An interesting difference between the Nordic countries is that belonging to an enterprise group is positive in Denmark, Estonia and Sweden, but negative for Norway. The parameter is insignificant for Iceland and Finland. 8

21 Table 1. Heckman Selection equation: The innovation decision FINLAND NORWAY DENMARK ESTONIA ICELAND SWEDEN Firm size 0.253*** 0.240*** 0.247*** 0.200*** 0.385*** 0.107*** (0.027) (0.027) (0.029) (0.038) (0.069) (0.025) Enterprise group ** 0.236*** 0.184** *** (0.070) (0.058) (0.084) (0.080) (0.22) (0.061) Foreign market 0.495*** 0.569*** 0.607*** 0.215** *** (0.086) (0.054) (0.080) (0.092) (0.22) (0.057) Hampering factors 1.240*** 2.157*** 2.007*** 1.086*** *** (0.14) (0.12) (0.17) (0.14) (0.32) (0.12) Constant *** *** *** *** *** *** (0.18) (0.14) (0.22) (0.21) (0.15) Observations Notes: *** p<0.01, ** p<0.05, * p<0.1. Standard errors in parentheses. 16 Industry dummies not reported. In equation 2, shown as the purple box on page 8 and in the appendix, we consider the level of investment in innovation, contingent on having decided to innovate in the first place. Inputs are measured as the log of innovation expenditure per employee. In our model, the amount of resources invested in innovation inputs depends on: belonging to an enterprise group; being active in foreign markets; the financial support received; the intensity of innovation cooperation with other partners; and industrial dummies. The figure on page 8 summarizes the importance and signs of the inputs into innovative activities and table 2 provides the estimates. We find the same difference between the countries on the effect of belonging to a group, with it being negative for Norway and positive for Denmark. However, for the other countries group affiliation is insignificant for innovation investment. Being active on a foreign market also affects innovation costs positively in Denmark, Norway and Sweden, but has no significant effect for the other countries. Financial support for innovation from public authorities is also positive and significant for all the countries, except for Iceland, where it is not significant. There is also no data available on innovation funding for Sweden. Overall, public support appears to have a positive influence on innovation expenditures in the Nordic region. A similar positive relationship is found for innovation cooperation, including in Iceland. Equation 3, shown in the green box on page 8 and in the appendix, contains what is generally known as the knowledge production function. It models the relationship existing between the investment in innovative inputs and the innovative outputs firms obtain. Our dependent variable is the log of innovative sales per employee. As variables explaining firms innovative sales per employee we use: firm size; belonging to a group; the fact that the firm is also carrying out process innovation Table 2. Heckman equation: Innovation inputs FINLAND NORWAY DENMARK ESTONIA ICELAND SWEDEN Enterprise group * 0.299* (0.13) (0.082) (0.17) (0.16) (0.68) (0.13) Foreign market *** 0.603*** ** (0.20) (0.099) (0.21) (0.20) (0.3) (0.20) Financial support 0.417*** 0.651*** 0.638*** 0.432** 0.85 (0.12) (0.087) (0.17) (0.20) (0.7) Cooperation 2.099*** 1.609*** 1.419*** 1.685** 5.637*** 2.433*** (0.40) (0.32) (0.43) (0.68) (1.9) (0.33) Constant *** *** 1.518*** *** 9.674*** (0.34) (0.22) (0.40) (0.40) (0.3) (0.47) Observations Notes: *** p<0.01, ** p<0.05, * p<0.1. Standard errors in parentheses. 16 Industry dummies not reported. 9

22 Table 3. Innovation output equation: Log of innovative sales per employee FINLAND NORWAY DENMARK ESTONIA ICELAND SWEDEN Firm size *** *** * * (0.12) (0.075) (0.076) (0.17) (1.49) (0.057) Enterprise group *** 0.462*** 0.350* (0.14) (0.11) (0.14) (0.21) (0.87) (0.098) Process innovation 0.282** *** (0.12) (0.090) (0.12) (0.14) (0.65) (0.081) SME * * (0.21) (0.24) (0.16) (0.26) (1.19) (0.17) MILLS ratio ** ** * (0.57) (0.28) (0.32) (0.95) (4.98) (0.23) Cost factors ** * * *** (0.28) (0.22) (0.25) (0.32) (1.11) (0.18) Knowledge hurdles (0.33) (0.28) (0.32) (0.37) (1.24) (0.24) Market obstacles * (0.33) (0.30) (0.31) (0.38) (1.11) (0.23) Cooperation with firms *** (0.78) (0.50) (0.39) (0.69) (2.56) (0.42) Cooperation with pubic 2.005** (0.95) (0.77) (0.46) (0.99) (3.70) (0.41) Cooperation intensity * (0.70) (0.32) (0.19) (0.29) (1.20) (0.19) National cooperation *** ** (0.35) (0.38) (0.34) (0.61) (1.42) (0.29) Innovation input 0.188*** 0.459*** *** 0.141*** 0.336* 0.180*** (0.042) (0.039) (0.034) (0.039) (0.18) (0.026) Constant 5.061*** 3.664*** *** *** (1.13) (0.78) (0.78) (1.93) (10.70) (0.66) Observations R-squared Notes: *** p<0.01, ** p<0.05, * p<0.1. Standard errors in parentheses. 16 Industry dummies not reported. together with product innovation; whether it is a small or medium sized enterprise (SME); and the intensity of the various hampering factors encountered; that is, cost factors, knowledge hurdles, and market-related obstacles. In our model the log of innovative sales is also explained by the intensity of the cooperation links firms have with their suppliers and customers and competitors, the intensity of cooperation with universities, consultancies and research centres, the intensity of cooperation at the international level relative to national cooperation, the intensity of the national cooperation ties, and innovation input (from equation 2). In contrast to the original CDM model, we use firms total innovation expenditures as innovation input. This is done to capture the effect of the non-r&d activities that go into the innovation process. One feature of the estimates shown in table 3 that can appear striking at a first glance is that many of the variable coefficients in the equation are non significant. This is especially the case for the cooperation variables and the different types of hampering factors. This may to some degree reflect difficulties in measuring these factors and the relatively small number of observations in Nordic country samples. Nonetheless, the estimates provide very interesting results and country differences also emerge. We see, for instance, that cooperating with universities and research centres positively relates to Finnish firms innovative sales, but nothing can be said about the other Nordic countries. A somewhat more general result emerges with respect to the effect of the propensity to collaborate at the national level. Although the coefficients are statistically 10

23 significant only for Finland and Sweden, we see that, in general, concentrating on national cooperation may hurt firms innovative sales. Sweden is also the only country with a strong and significant positive effect between innovative sales and innovation cooperation with other firms. As for the cost related hampering factors, we see that they are significantly and negatively correlated with innovation sales, as expected. Also market-related difficulties seem to hurt firms innovative sales, even if such an effect is statistically significant only in the case of Finland. Evidence suggests that combining process and product innovations positively relates to higher innovative sales. As for belonging to a group, again we see a similarity between Denmark and Norway, whose estimates are characterised by positive and significant coefficients, as it is also the case for Estonia. For the other Nordic countries membership in an enterprise group remains positive, but is not statistically significant. Innovation output is negatively correlated with the size of the companies in all countries. This implies that the likelihood to innovate increases with firm size, as suggested in Table 1, but it also shows that the outcome is proportionally higher among smaller firms in terms of innovative sales. We also see a strong relationship between investment in R&D and innovative sales. Innovation input is always positive and significant in all the Nordic countries. Hence, it is reasonable to assume that the level of innovation activity, estimated in equation 2, affects output. The inverse Mills ratio, obtained from the selection equation, is negative for all countries and significant except for Norway (which has a compulsory questionnaire). This implies that accounting for selection bias is of great importance, and by not doing it would provide misleading results on the impact on innovation sales per employee. Innovation output should not be considered to be the final target of firms activity. Rather firms invest in R&D and innovation and in other activities in order to improve their overall performance and competitiveness. Hence, innovation should be considered as a tool to achieve such a goal a goal often referred to as productivity growth. This is exactly what we try to capture in the fourth and final equation of the model, which pictures the relationship between innovation output and productivity. The variables that enter this final equation are characteristics such as size, group affiliation, and whether or not process innovations have taken place, together with the Table 4. Productivity equation FINLAND NORWAY DENMARK ESTONIA ICELAND SWEDEN Firm size 0.101*** *** 0.132*** 0.138** *** (0.038) (0.034) (0.035) (0.056) (0.45) (0.026) Enterprise group 0.249*** 0.277*** 0.209** 0.238** 0.746** 0.108** (0.064) (0.057) (0.093) (0.12) (0.33) (0.047) Innovation output 0.334*** 0.318*** 0.324*** 0.708*** 0.493*** 0.438*** (0.058) (0.050) (0.11) (0.14) (0.18) (0.064) Process ** (0.056) (0.050) (0.056) (0.11) (0.47) (0.044) SME ** (0.093) (0.12) (0.078) (0.15) (0.089) MILLS ratio 0.290** ** * 0.152* (0.13) (0.096) (0.099) (0.19) (1.71) (0.085) Constant 3.576*** 5.172*** *** 9.182*** 10.55*** (0.33) (0.42) (0.26) (0.77) (3.32) (0.54) Observations R-squared Notes: *** p<0.01, ** p<0.05, * p<0.1. Standard errors in parentheses. 16 Industry dummies not reported. 11

24 dummy for the firm being an SME or not, the inverse Mills ratio, and innovative sales per employee. The selection-related parameter turns out to be positive for productivity when significant, with the exception of Iceland. This implies that the impact on productivity would be too high if we did not take the selection bias into account. Again, for Norway and Estonia it is not significant. In equation 4, shown in the appendix and in the red box on page 8, productivity is measured as log of turnover per employee. As expected, it is positively and significantly related to the amount of innovation output firms have (although negative but insignificant for Iceland). We find a consistent pattern among the countries: The size of the firm and belonging to a group relates to productivity in a positive way (except for Iceland where it is significant and negative). Process innovation appears to have a negative relationship to productivity. This may seem counter-intuitive, but we believe that the very broad type of innovations that can be encompassed by the term process might drive this. What this means is that firms may understand process innovations as minor, production-specific, and often product-related adjustments, as well as major organizational reshufflings. The latter being the case in the short run (that is, the period we observe) we would observe only the disruption caused by these major adjustments, and therefore end up observing negative coefficients, as we in fact do. The Nordic common model appears to work reasonably well for all countries except Iceland. As mentioned, this is likely due to having too few observations. The features captured by the equations appear to be sensible, and thus confirm the assumed relationship from innovation activity to economic outcome via innovative output. The analysis reveals more of a common structure between the countries than differences. Our results are partly driven by the underlying country-specific industrial structures, which we attempt to control for via the industry dummies included in the model. In particular, we see that the presence of public support has a positive effect on the level of innovative activity, which is positively correlated with productivity. Thus policy in this area appears to work from such a broad perspective. We think that further research would be needed to more carefully investigate the effect of the various cooperation patterns taking place among firms, and within firms and their surroundings, as this is among the target areas for public policies. From a more technical point of view the analysis would benefit from having more information about non-innovators. Another important issue that would need to be addressed is simultaneity. At present the questionnaire asks about innovation inputs and innovation output that all occur at the very same time. In reality outputs follow inputs in time. Hence, without the possibility to account for the necessary time lag needed for inputs to have an effect on outputs the analysis suffers from this inability to control for time-related dynamics. To obtain additional information there might be two possible solutions: one is to ask additional questions in the survey; and one to merge additional data. There are pros and cons to both. Changing the coverage, frequency and timing of the survey is a difficult and time consuming process, but would improve the quality and quantity of information gathered, thus enabling better estimations. Merging CIS data with data gathered from external sources is a good solution if statistical offices are willing to match the data with other sources at the firm level and if these other types of innovationrelated data are available. The next question, and too lengthy to discuss here, is what to add. The most obvious candidates are investment, use of ICTs, skill level of the workforce, exports and financial accounts. To account for the more complex dynamic aspects of the relationship between innovation input, innovation output and productivity, we would need panel data, that is, data that encompass the same firms and are collected in regular intervals over time. 2.3 Some econometric issues Applications of the CDM model to various innovation surveys give quite reasonable results, and certainly represent an important step forward in understanding the innovationproductivity link. However, results prove to be 12

25 highly sensitive to model specifications and to the way selectivity, simultaneity and endogeneity problems are addressed. For this reason it is extremely important to properly deal with these econometric problems and to develop and use a standardized model, especially if cross-country comparative work is to be undertaken. Some econometric and statistical issues at stake in these studies include: 7 1. Measurement issues Many necessary data are not available. For instance, we often have only sales and not value added or material input data. Moreover, when estimating the model, coefficients tend to be inflated because material inputs are held constant over time. The possible use of deflators, and how to account for depreciation of both tangible and intangible (including knowledge) assets should therefore be addressed, together with, such as problems like double counting of R&D. 2. Endogeneity and simultaneity biases. Problems arise because the causation link between innovation and productivity can be bi-directional: one may hold that innovation improves productivity, but at the same time it is likely that more productive firms are also those that succeed better when innovating. Moreover, simultaneity problems can be present, as inputs and outputs are observed as occurring simultaneously. These problems have to be taken into account in the model and can affect identification of the model parameters. Two-stage procedures or the use of instrumental variables are among the solutions suggested by the literature (in particular GMM methods), but choosing the wrong instrument can also cause additional problems (Arellano, 2005). 3. Sample selection bias Sample selection bias occurs when the dependent variable is observed only for a restricted, non-random part of the sample. To correct for sample selection, Heckman-type selection models are widely used, but their applicability depend on the fulfilment of the specific underlying conditions. 4. Autocorrelated disturbances The CDM model relies on the strong (and perhaps not too realistic) assumption that the disturbance terms in each of the four equations are not correlated. In order to fulfil this condition it would be better to have time series of innovative activity, thus creating a true panel of firms. Lööf and Heshmati (2006) introduce a multi-step approach. 5. Linearity Many of the models above are linear, when the relationship between innovation and productivity is probably non-linear. Testing for nonlinearity is also a test for heterogeneity, particularly since the returns to R&D activity is heterogeneous across firms. 6. Timing the internal timing problem of inputs and (intermediate) outputs in the existing innovation surveys can only be overcome by changing the survey design. At present including supplementary time series information on performance is the only solution. 7. Other issues Persistence of innovation and censoring. The persistence of innovation and timing issues are related to the way innovation surveys are carried out in Europe. In the design of CIS3 and CIS4, innovative activities take place over a three-year period, whereas the information about the effects of innovation is collected only for the last year. Alternatively, one should hold innovative activities to be constant over time and persistent, and assume 7 See Mohnen (2006) and Knell and Nås (2006; 2008) for a further discussion of the econometric issues. Also see Wooldridge (2001) and Arellano (2003) for an introduction to panel data methods. Kennedy (2003) provides a good guide to some of the issues important to econometrics. 13

26 that the results obtained in a considered year are the same that would be observed if we took the necessary time lag into account. This would be equivalent to assuming that current innovation efforts are representative of or a continuation of similar efforts in the past that actually lead to the results reported. Raymond et al. (2006) show that this assumption is generally not confirmed. It would be necessary to construct a time series panel data to overcome the problem. The ideal would be to have balanced panels of innovation surveys with a reasonable frequency, preferably annually. Both frequency and balancing of panels are, however, resource demanding and will take time before usable data can be in place. Differences in lag structures between industries will also represent a problem. Alternatively, innovation data can be supplemented with external performance data at the enterprise level, such as yearly balance accounts. Such a solution helps in balancing the panels, as all enterprises have to report results. Problems of time lags would still remain, though. There are also problems due to the differences in the accounting practices followed across countries. Lastly, the timing problem of including in the analysis intermediate results such as innovative sales is not solved this way. The latter would require linking frequent innovation surveys over several years. This would enable the application of GMM estimation techniques (Arellano, 2004), which at present are impossible to use in this context. These would greatly improve estimation accuracy and the robustness. As mentioned, selectivity problems arise because only a fraction of firms engage in innovation in a given period, and it is very likely that systematic differences exist between innovators and non-innovators. To correct for this problem, it is common to rely on Heckman selection models. It may be necessary to use variables that are supposed to affect the decision of firms to innovate to identify the parameters of the model, but not the amount then invested in innovation. The hypotheses made are known as an exclusion restriction and the identification of the model parameters crucially rely on them. At present, however, there is too little information that can be used for this purpose and that is collected for both innovators and non-innovators. To overcome this problem one option is to change the questionnaires and eliminate the filter question, thus making all firms respond to all questions. Another option is again to gather the information that is needed from supplementary statistical sources. The latter is the only possibility in the short run, and may constitute a solution if matching can be done at the micro level. Information that could be of help when estimating the innovation-productivity link include age of the firms, some indicators of technological opportunities in the given industry, and the composition of human resources in the firm, and industry-specific characteristics. Apart from technological opportunity- related data, many of the above variables are census-based information available in many countries. Technological opportunity is difficult to observe directly, but may be approximated by, for instance, the numbers of patents or patent applications in an industry. It is also related to the prevailing competition and market structures surrounding each firm. Such information is available in some cases, but usually limited to indicators such as share of national markets and export share of sales. In the increasingly open and free trading economies, such information should include the full world markets, even if doing so indeed constitute a difficult challenge and may introduce significant measurement problems Are innovative sales a good measure of innovative output? A central issue in innovation analysis is to gain a better understanding of how firms transform 14

27 knowledge inputs into marketable products and improvements in productivity. R&D and other investments in the creation of new knowledge are widely accepted as central factors in generating economic growth. Yet, the process by which R&D leads to growth may be very complicated. Kline and Rosenberg (1986) argue that innovation is often treated as a black box where economic analysis has largely neglected the highly complex processes through which certain inputs are transformed into certain outputs (Kline and Rosenberg, 1986: 278). Empirically, an important issue is how to measure knowledge output, that is, new knowledge produced by activities such as R&D that can be used to create new products and processes. Initial studies used patents as a measure of knowledge, however with the advent of innovation surveys such as the CIS4, researchers have sought alternative measures of innovative output. The most commonly used measure is the share of innovative sales, defined as the share of sales that can be attributed to product innovations. Yet, little work has been done to investigate the characteristics of this measure and how well it functions as a measure of knowledge, or what Griliches called knowledge capital. As with patents, the share of innovative sales may also have shortcomings, particularly as a measure of innovativeness. For example, one firm might have a truly novel invention that only comprises a small share of its sales while another firm may introduce a fairly minor change that impacts all its sales. However, these shortcomings, or variations in the actual activity that has gone into the innovations, may potentially be useful in analyzing the relation between innovation and productivity. For example, does novel innovative activity have a greater impact on performance than the adoption of products and technologies already on the market? By characterizing the relation between the share of innovative sales and a variety of other indicators, we hope to shed light on the relation between innovation activity and output that may be useful in econometric modelling. A paper written by Carter Bloch and Ebbe Graversen [2] for the project examines the share of innovative sales using a variety of approaches, with the aim of gaining a better understanding of what we are measuring with this indicator. By characterizing the relation between the share of innovative sales and a variety of other indicators, the paper hopes to shed light on the relation between innovation activity and output that may be useful in econometric modelling. The empirical analysis of the share of innovative sales as a measure of innovative output has shown several interesting patterns. Danish data illustrates the point that the output measure has to be used carefully as a proxy for productivity, but also that it is a usable proxy for knowledge. Shares of innovative sales vary much more greatly across service sectors than across manufacturing sectors. Electronics and optical instruments (ISIC 30, 32, and 33) have the highest shares, while food and beverages (ISIC 15), chemicals and pharmaceuticals (ISIC 24) and machinery and equipment (ISIC 29) all have shares around manufacturing averages. In contrast, service sectors such as transport (ISIC 34 and 35) and financial intermediation (ISIC 65-67) have very low averages shares of innovative sales, while the highest shares among all sectors are found within knowledge intensive services. Somewhat surprisingly, shares of innovative sales are the same for SMEs and large firms. However, shares are much larger for very small firms with less than 10 employees. Firms with the most novel innovations are generally considered the most innovative, and thus also expected to have the highest innovative output. This is also generally found to be the case: firms with new to market international innovations have on average the highest shares of innovative sales. 15

28 Organisational innovation is also given an important role and has been found in some studies to have an important impact on the overall productivity of innovation activities. In terms of innovative sales, the paper found little difference among the most novel innovators when comparing those with and without organisational innovations. One possible explanation for this is that the organizations of the most novel innovators are very geared towards innovation, including those that have not recently implemented organisational changes. In contrast to this, we found very large differences for less novel innovators: shares of innovative sales were almost twice as high for firms having also implemented organisational innovations. It might be expected that shares of innovative sales are highly correlated with R&D expenditures. However, a strong relationship is not found in the Danish data. While the paper finds a relation for extreme values - firms with very low (high) innovative sales also tend to have very low (high) R&D intensity for firms with between 10 and 50 percent innovative sales, average R&D intensity is the same. Correlation coefficients between R&D intensity and shares of innovative sales are also quite low What is really inside the knowledge production function Knowledge and knowledge creation appears at the centre of the NBER family of models, including the CDM model. Yet these models consider knowledge as an unobservable. The main reason is that we have only a rudimentary understanding of the factors that shape the rate, direction and effects that the learning process has on the creation of knowledge. Even among philosophers, the theory of knowledge remains controversial. Epistemological discussions have focused on analyzing the nature of knowledge, how it is acquired and how it relates to truth, belief, justification, and perception. For the early philosophers, including Plato, Aristotle, and Descartes, knowledge was purely objective in that it was impersonal, explicit and permanent. The Cartesian method, which was based on the search for certainty and universality through the use of conjecture, influenced the subsequent discussion on the limitations of knowledge and the meaning of experience. Descartes and Newton viewed knowledge as something that can be obtained not only from physical existence or empirical observation, but also from mental or imaginary constructs. The separation of abstract reasoning from physical events was vital to the Newtonian method, and to the way Newton was able to recognize the hierarchy of causes and distinguish mathematically derived laws from the physical properties of forces while still maintaining close interaction between mathematical reasoning and physical causes. This ability to link the abstract or mathematical reasoning with empirical or physical science to generate knowledge became the main feature of the Principia, as well as the source of much confusion by those advocating a more mechanistic interpretation of the method in the 19th and 20th centuries. It also became the most important feature of the Scottish Enlightenment, and Adam Smith s view of classical political economy. We can place Adam Smith (1776) at the centre of this discussion; both in terms of the philosophical debate and in defining how this epistemology can be incorporated into economic theory (cf. knowledge creation through the division of labour in chapters 1-3 of Book 1 and the assumption of known knowledge in chapters 5-7 of Book 1). 16

29 Perception, in the form of probabilistic statements, remained trivial in the theory of knowledge until the twentieth century. This changed when Merleau-Ponty (1945) argued for the primacy of perception over conjecture and experience, providing justification for moving away from the view that knowledge must be, at least, true and justified toward one where only beliefs are warranted. Epistemology thus becomes biological in the sense that knowledge must contain a tacit dimension as pointed out by Polanyi (1958). This recognition of two different kinds of knowledge is closely related to Ryle s (1947) distinction between knowing how (procedural knowledge), and knowing what (descriptive knowledge). While tacit knowledge is difficult to define, it is important for evolutionary economic models and innovation policies based on these models. Building on the previous ideas of Ryle, Lundvall and Johnson (1994) identify four different types of knowledge: (1) knowing how, (the ability to do something); (2) knowing what (knowledge about facts); (3) knowing why (knowledge about principle and laws); and (4) knowing who (knowledge about who knows what). This idea of knowledge fits very closely to Polanyi s (1958) distinction between tacit and codified knowledge and Nelson s (1991) distinction between generic and specific knowledge. A paper by Knell and Nås [5] for the project argues that one implication of this line of thought is that knowledge capital should be biological in the sense that it considers the tacit dimension. Discussions in the philosophy of knowledge suggest that knowledge appears in many different forms and localizations and interacts with other factors. In economics, knowledge is usually poorly understood and it often appears in simple forms; in neoclassical theory usually reduced to information that can be transferred without transaction costs or time lags. The relevant forms of knowledge vary according to the circumstances and over time, depending on the form of knowledge and the intended use. To be able to utilize knowledge, in any form, it must be learned by the relevant actors; firms and their management and employees, and customers. Knowledge forms are different when it comes to the time and effort it takes to learn and transfer them, the risk of copying, and the cost of utilizing them. The community innovation survey as it is now constructed does not deal with these different types of knowledge very well. Even when they distinguish between different components of innovation activities and costs, the breakdown is still difficult to comprehend both for respondents and analysts. It mixes types of knowledge with sources of knowledge and methods of producing it, and leaves out questions about learning. Learning is necessary to be able to implement and utilize knowledge, which may not be trivial Does history matter? One issue that is often overlooked in the literature is the extent to which pathdependency shapes innovative activities and their outcomes. Using composite indicators of input, process, output and the economic environment, it may be possible to identify a set of innovation indicators in which firms are similar in their behaviour for responding to general changes in competitiveness of an industry. Path dependency is a concept that has been criticised by being a fashionable label for the intuition that history matters without a clear and convincing account of decisionmaking over time, explaining only stability and not change; and its normative implications are confused and mostly left unexplored. In a paper written for the project, Ukrainski, Masso, and Varblane [8] capture some dynamic aspects of path dependency by assessing it through the level and dynamics of value added created in production in different industries. Using Estonian CIS4 data, indicators were grouped into three categories: innovation input, process and outcome. With the help of factor analysis the whole set of innovation indicators was compressed into 8 synthetic factors. These factors were labelled in the case of innovation inputs innovation cooperation, innovation barriers and capability to invest, in the case of process innovation nontechnological process innovation and technological process innovation and in the case of innovation output and outcome variables respectively impact of innovations, product innovations and commercialization 17

30 towards different groups of sectors. For instance, e.g. compared with sectors having both low initial levels of value added and rapid growth, innovation cooperation needs to be addressed much more in the case of sectors with low initial levels, but rapid growth of value added. 2.4 Some alternative models and exporting. The initial level of value added and its growth were taken as proxies for path dependency. Sectors were divided into four groups according their initial level and growth of value added. Results of the analysis supported the idea that the innovative behaviour of different industries was very heterogeneous and the initial level as well the speed of growth of value added matters. In very general terms the innovative behaviour of industry groups with low initial level of value added was weak. But looking more precisely, the two industry groups with initially low value added levels (previous development path has produced weak starting position) behave rather differently. In one group of industries the high level of cooperation and low barriers of innovation combined with high propensity to export resulted in the rapid growth of value added, but it occurred despite that they were very weak in the innovation process. In the second group of industries with low initial level of value added, all of the innovation input factors were extremely low and their major attempt was to execute process innovations in order to be able to continue to s The conclusion from these two groups is that initial levels of value added combined with innovation input indicators gives a rather good understanding about the different speed of growth of the value added in those groups. From a policy implications point of view, it emphasizes the need for a targeted approach One objective of the IGNOREd project was to experiment with different econometric models without having to be restricted to a common model. This exercise opened up the possibility for modelling the relationship between innovation and productivity growth in different ways by using peculiarities of the individual country surveys. A conscious attempt was made to address some of the econometric issues raised during the course of the project. But as the reader can see from the discussion, some of the issues are far beyond the scope of the project Innovation indicators and performance An analysis for Danish firms Policy documents, reports and analyses show that there is a need to develop a more accurate and detailed understanding of the role and dynamics of different types of innovation for firm performance and economic growth. Yet, the influence of indicators derived from innovation surveys on policy has so far been rather minor (Arundel, 2007). R&D indicators are still the most widely used indicators of innovative activity. One important factor may be that they should be better known and accepted as measures of innovation activity. This requires extensive analysis, both econometric and otherwise to examine the properties of the indicators. An additional factor that may have reduced policy use is an under exploitation of innovation survey data. Many potentially useful indicators of direct relevance to policy concerns have not been developed. Almost all publicly available indicators from innovation surveys are simple indicators based on a single question, such as the share of enterprises that applied for one or more patents, or the percentage of firms that have engaged in innovation cooperation. Although these 18

Hans-Olof Hagén, Caroline Ahlstrand, Martin Daniels, Emma Nilsson and Adrian Adermon

Hans-Olof Hagén, Caroline Ahlstrand, Martin Daniels, Emma Nilsson and Adrian Adermon Innovation matters An empirical analysis of innovation 2002 2004 and its impact on productivity Hans-Olof Hagén, Caroline Ahlstrand, Martin Daniels, Emma Nilsson and Adrian Adermon Conclusions In this

More information

Accounting for Innovation and Measuring Innovativeness: An Illustrative Framework and an Application

Accounting for Innovation and Measuring Innovativeness: An Illustrative Framework and an Application Accounting for Innovation and Measuring Innovativeness: An Illustrative Framework and an Application By JACQUES MAIRESSE AND PIERRE MOHNEN* The purpose of this paper is to propose and illustrate an accounting

More information

The determinants and effects of technological and nontechnological innovations Evidence from the German CIS IV

The determinants and effects of technological and nontechnological innovations Evidence from the German CIS IV The determinants and effects of technological and nontechnological innovations Evidence from the German CIS IV Tobias Schmidt 1 Christian Rammer 2 This (shortened) version: 06 September 2006 Abstract In

More information

The Innovation Union Scoreboard: Monitoring the innovation performance of the 27 EU Member States

The Innovation Union Scoreboard: Monitoring the innovation performance of the 27 EU Member States MEMO/12/74 Brussels, 7 February 2012 The Innovation Union Scoreboard: Monitoring the innovation performance of the 27 EU Member States This MEMO provides an overview of the research and innovation performance

More information

The Impact of Firm s R&D Strategy on Profit and Productivity

The Impact of Firm s R&D Strategy on Profit and Productivity CESIS Electronic Working Paper Series Paper No. 156 The Impact of Firm s R&D Strategy on Profit and Productivity Börje Johansson* and Hans Lööf** (*CESIS and JIBS, **CESIS and Division of Economics, KTH)

More information

Fourth European Community Innovation Survey: Strengths and Weaknesses of European Countries

Fourth European Community Innovation Survey: Strengths and Weaknesses of European Countries Fourth European Community Innovation Survey: Strengths and Weaknesses of European Countries Funda Celikel-Esser, Stefano Tarantola and Massimiliano Mascherini EUR 22799 EN - 2007 The mission of the JRC

More information

Investigation of ICT Firms' Decisions on R&D Investment

Investigation of ICT Firms' Decisions on R&D Investment Investigation of ICT Firms' Decisions on R&D Investment Wojciech Szewczyk, Juraj Stančík, Martin Aarøe Christensen 2 0 1 3 Report EUR 26230 EN European Commission Joint Research Centre Institute for Prospective

More information

A rm s-length transfer prices for the remuneration of

A rm s-length transfer prices for the remuneration of 1164 (Vol. 25, No. 20) BNA INSIGHTS Cost Plus Markups for Manufacturing Entities: The Effect of Size on Fully Loaded Versus Variable Costs The authors analyze the statistical and economic significance

More information

information exchange. To achieve this, the process must be managed through a partnership between the European Commission and the member states.

information exchange. To achieve this, the process must be managed through a partnership between the European Commission and the member states. As a result of an increasingly global economy based on knowledge and innovation, and in the wake of the achievement of the internal market and the EMU, Europe is in the process of adjusting its policies

More information

Submission to Inquiry into Economic Statistics Economy, Jobs and Fair Work Committee, Scottish Parliament

Submission to Inquiry into Economic Statistics Economy, Jobs and Fair Work Committee, Scottish Parliament Submission to Inquiry into Economic Statistics Economy, Jobs and Fair Work Committee, Scottish Parliament Written Evidence by Professor Catia Montagna and Dr Daniel Kopasker September 2017 Background We

More information

Title of manuscript: Patterns of innovation in the SMEs of the Hungarian agri-food industry

Title of manuscript: Patterns of innovation in the SMEs of the Hungarian agri-food industry Title of manuscript: Patterns of innovation in the SMEs of the Hungarian agri-food industry Author: Áron Török (corresponding), József Tóth, Jeremiás Máté Balogh University: Corvinus University of Budapest,

More information

Introduction. Methodology

Introduction. Methodology Introduction The EUROCHAMBRES Economic Survey 2018 (EES 2018) is the 2th consecutive annual assessment of the European business community s expectations. The study is based on responses from over 0,000

More information

Peter Nedergaard Jean Monnet Lecture: 11 February, 2008 EU s Internal Market Policy: Results and Problems

Peter Nedergaard Jean Monnet Lecture: 11 February, 2008 EU s Internal Market Policy: Results and Problems Peter Nedergaard Jean Monnet Lecture: 11 February, 2008 EU s Internal Market Policy: Results and Problems The power points can be found on www.cbs.dk/staff/pne 1 Overview: 1) An Economic Portrait of Europe

More information

The micro-evidence of innovation: data and research applications. Micheline Goedhuys. UNU-MERIT, Maastricht, Netherlands

The micro-evidence of innovation: data and research applications. Micheline Goedhuys. UNU-MERIT, Maastricht, Netherlands The micro-evidence of innovation: data and research applications Micheline Goedhuys UNU-MERIT, Maastricht, Netherlands Overview of lecture 1. The rise of survey methods 2. OECD Oslo manual and innovation

More information

TOTAL FACTOR PRODUCTIVITY DETERMINANTS IN DEVELOPED EUROPEAN COUNTRIES

TOTAL FACTOR PRODUCTIVITY DETERMINANTS IN DEVELOPED EUROPEAN COUNTRIES TOTAL FACTOR PRODUCTIVITY DETERMINANTS IN DEVELOPED EUROPEAN COUNTRIES Bogdan Florin FILIP Alexandru Ioan Cuza University of Iaşi, Faculty of Economics and Business Administration Iaşi, Romania bogdan.filip@feaa.uaic.ro

More information

Measuring innovation West Africa Regional Science, Technology and Innovation Policy Reviews and Statistics Workshop Bamako, Mali May 2010

Measuring innovation West Africa Regional Science, Technology and Innovation Policy Reviews and Statistics Workshop Bamako, Mali May 2010 Measuring innovation West Africa Regional Science, Technology and Innovation Policy Reviews and Statistics Workshop Bamako, Mali 10-13 May 2010 Measuring Innovation Oslo Manual - 2005: (Guidelines for

More information

Innovation, Productivity and Exports: Firm-Level Evidence from Malaysia

Innovation, Productivity and Exports: Firm-Level Evidence from Malaysia Innovation, Productivity and Exports: Firm-Level Evidence from Malaysia Cassey Lee Nottingham University Business School University of Nottingham Malaysia Campus Working Paper Series Vol. 2008-06 March

More information

Mergers and Sequential Innovation: Evidence from Patent Citations

Mergers and Sequential Innovation: Evidence from Patent Citations Mergers and Sequential Innovation: Evidence from Patent Citations Jessica Calfee Stahl Board of Governors of the Federal Reserve System January 2010 Abstract An extensive literature has investigated the

More information

University of Pretoria Department of Economics Working Paper Series

University of Pretoria Department of Economics Working Paper Series University of Pretoria Department of Economics Working Paper Series Social Rate of Return to R&D on Various Energy Technologies: Where Should We Invest More? A Study of G7 Countries Roula Inglesi-Lotz

More information

Introduction to innovation surveys and some very basic econometrics. Pierre Mohnen August 2016

Introduction to innovation surveys and some very basic econometrics. Pierre Mohnen August 2016 Introduction to innovation surveys and some very basic econometrics Pierre Mohnen August 2016 Main references Mairesse, Jacques and Pierre Mohnen, «Using innovation surveys for econometric analysis», in

More information

The Economic and Social Review, Vol. 33, No. 1, Spring, 2002, pp

The Economic and Social Review, Vol. 33, No. 1, Spring, 2002, pp 08. Girma article 25/6/02 3:07 pm Page 93 The Economic and Social Review, Vol. 33, No. 1, Spring, 2002, pp. 93-100 Why are Productivity and Wages Higher in Foreign Firms?* SOURAFEL GIRMA University of

More information

Innovation and Firms Performance in the Rwandese Manufacturing Industry. A firm Level Empirical Analysis

Innovation and Firms Performance in the Rwandese Manufacturing Industry. A firm Level Empirical Analysis Innovation and Firms Performance in the Rwandese Manufacturing Industry. A firm Level Empirical Analysis Etienne Ndemezo 1 and Charles Kayitana 2 January 2017 Abstract The main objectives of this paper

More information

A Cross-Country Analysis of the Employment Intensity of Economic Growth. William Seyfried Rollins College

A Cross-Country Analysis of the Employment Intensity of Economic Growth. William Seyfried Rollins College A Cross-Country Analysis of the Employment Intensity of Economic Growth William Seyfried Rollins College Abstract In this paper, we examine the relationship between economic growth and employment in six

More information

Potential Gains from Trade Liberalisation in the Baltic Sea Region

Potential Gains from Trade Liberalisation in the Baltic Sea Region 2007-05-22 Potential Gains from Trade Liberalisation in the Baltic Sea Region Executive Summary Executive Summary Potential Gains from Trade Liberalisation in the Baltic Sea Region A Simulation of the

More information

I. Introduction. Knowledge Flows and Innovative Performance: Evidence from Italian rms

I. Introduction. Knowledge Flows and Innovative Performance: Evidence from Italian rms Knowledge Flows and Innovative Performance: Evidence from Italian rms Martina Aronica a, Giuseppe De Luca b, Giorgio Fazio c, Davide Piacentino d a Martina Aronica, SEAS, University of Palermo, Viale delle

More information

Policy Note August 2015

Policy Note August 2015 Unit Labour Costs, Wages and Productivity in Malta: A Sectoral and Cross-Country Analysis Brian Micallef 1 Policy Note August 2015 1 The author is a Senior Research Economist in the Bank s Modelling and

More information

Estimating the Indirect Economic Benefits from Research

Estimating the Indirect Economic Benefits from Research Estimating the Indirect Economic Benefits from Research Bruce A. Weinberg Ohio State University, IZA, & NBER www.bruceweinberg.net National Academies April 18-19, 2011 Background Governments are major

More information

Distributed Access to Linked Microdata: the Example of ICT and Exports

Distributed Access to Linked Microdata: the Example of ICT and Exports Distributed Access to Linked Microdata: the Example of ICT and Exports Eva Hagsten* Statistics Sweden July 2014 Abstract This paper describes the novelty of how the relationship between ICT usage in firms

More information

R&D and performance in modern economies: survey R&D versus broader R&D measures Prof. Hannu Piekkola University of Vaasa

R&D and performance in modern economies: survey R&D versus broader R&D measures Prof. Hannu Piekkola University of Vaasa R&D and performance in modern economies: survey R&D versus broader R&D measures Prof. Hannu Piekkola University of Vaasa Challenges in measuring productivity, growth and intangibles EUROSTAT BRYSSELS 17.3.2017

More information

R&D Investments, Exporting, and the Evolution of Firm Productivity

R&D Investments, Exporting, and the Evolution of Firm Productivity American Economic Review: Papers & Proceedings 2008, 98:2, 451 456 http://www.aeaweb.org/articles.php?doi=10.1257/aer.98.2.451 R&D Investments, Exporting, and the Evolution of Firm Productivity By Bee

More information

Digitalization, Skilled labor and the Productivity of Firms 1

Digitalization, Skilled labor and the Productivity of Firms 1 Digitalization, Skilled labor and the Productivity of Firms 1 Jóannes Jacobsen, Jan Rose Skaksen and Anders Sørensen, CEBR, Copenhagen Business School 1. Introduction In the literature on information technology

More information

Determinants and Evidence of Export Patterns by Belgian Firms

Determinants and Evidence of Export Patterns by Belgian Firms Determinants and Evidence of Export Patterns by Belgian Firms Jan Van Hove, Sophie Soete and Zuzanna Studnicka University of Leuven Document Identifier D5.10 Case Study on Belgian business succession practices

More information

On the contribution of innovation to multi-factor productivity growth

On the contribution of innovation to multi-factor productivity growth On the contribution of innovation to multi-factor productivity growth George van Leeuwen a and Luuk Klomp b a* CPB Netherlands' Bureau for Economic Policy Analysis, P.O. Box 80510, NL-2508 GM The Hague,

More information

Communication Costs and Agro-Food Trade in OECD. Countries. Štefan Bojnec* and Imre Fertő** * Associate Professor, University of Primorska, Faculty of

Communication Costs and Agro-Food Trade in OECD. Countries. Štefan Bojnec* and Imre Fertő** * Associate Professor, University of Primorska, Faculty of The 83rd Annual Conference of the Agricultural Economics Society Dublin 30th March to 1st April 2009 Communication Costs and Agro-Food Trade in OECD Countries Štefan Bojnec* and Imre Fertő** * Associate

More information

User innovators and their influence on innovation activities of firms in Finland Jari Kuusisto, Mervi Niemi and Fred Gault

User innovators and their influence on innovation activities of firms in Finland Jari Kuusisto, Mervi Niemi and Fred Gault Working Paper Series #2014-003 User innovators and their influence on innovation activities of firms in Finland Jari Kuusisto, Mervi Niemi and Fred Gault Maastricht Economic and social Research institute

More information

TMD Working Paper: TMD-WP-63

TMD Working Paper: TMD-WP-63 ISSN 2045-5119 TMD Working Paper: TMD-WP-63 Government Support, Innovation and Productivity in the Haidian (Beijing) District Can Huang Yilin Wu Pierre Mohnen Yanyun Zhao School of Management, Zhejiang

More information

Innovating in the Manufacturing Sector 1 in Latin America and the Caribbean 2. Enterprise Surveys e

Innovating in the Manufacturing Sector 1 in Latin America and the Caribbean 2. Enterprise Surveys e Enterprise Surveys e Innovating in the Manufacturing Sector 1 in Latin America and the Caribbean 2 WORLD BANK GROUP LATIN AMERICA AND THE CARIBBEAN SERIES NOTE NO. 12/13 Basic Definitions Countries surveyed

More information

Irina Levina. Decentralization of decision-making at the firm: comparative analysis of firms in 7 European countries and Russia.

Irina Levina. Decentralization of decision-making at the firm: comparative analysis of firms in 7 European countries and Russia. Irina Levina Institute for Industrial and Market Studies, National Research University Higher School of Economics Decentralization of decision-making at the firm: comparative analysis of firms in 7 European

More information

National Competitiveness Council s Productivity Statement Productivity Statement 2018

National Competitiveness Council s Productivity Statement Productivity Statement 2018 National Competitiveness Council s Productivity Statement 2018 Productivity Statement 2018 1 November 2018 Summary Productivity growth is a key determinant of national competitiveness, enabling firms to

More information

Understanding UPP. Alternative to Market Definition, B.E. Journal of Theoretical Economics, forthcoming.

Understanding UPP. Alternative to Market Definition, B.E. Journal of Theoretical Economics, forthcoming. Understanding UPP Roy J. Epstein and Daniel L. Rubinfeld Published Version, B.E. Journal of Theoretical Economics: Policies and Perspectives, Volume 10, Issue 1, 2010 Introduction The standard economic

More information

2 The structure of the ICT sector in the Nordic Countries

2 The structure of the ICT sector in the Nordic Countries 14 2 The structure of the sector in the Nordic Countries 2. Introduction This chapter gives a description of the overall economic importance of the sector in the Nordic countries measured by the number

More information

Determinants of Firms Cooperation in Innovation

Determinants of Firms Cooperation in Innovation Determinants of Firms Cooperation in Innovation Flávio Lenz-Cesar and Almas Heshmati 1 Department of Industrial Engineering TEMEP, College of Engineering #37-306, Seoul National University San 56-1, Shilim-dong,

More information

FORECASTING LABOUR PRODUCTIVITY IN THE EUROPEAN UNION MEMBER STATES: IS LABOUR PRODUCTIVITY CHANGING AS EXPECTED?

FORECASTING LABOUR PRODUCTIVITY IN THE EUROPEAN UNION MEMBER STATES: IS LABOUR PRODUCTIVITY CHANGING AS EXPECTED? Interdisciplinary Description of Complex Systems 16(3-B), 504-523, 2018 FORECASTING LABOUR PRODUCTIVITY IN THE EUROPEAN UNION MEMBER STATES: IS LABOUR PRODUCTIVITY CHANGING AS EXPECTED? Berislav Žmuk*,

More information

Innovation and Productivity in SMEs. Empirical Evidence for Italy *

Innovation and Productivity in SMEs. Empirical Evidence for Italy * Innovation and Productivity in SMEs. Empirical Evidence for Italy * Bronwyn H. Hall, # Francesca Lotti, * and Jacques Mairesse Abstract Innovation in SMEs exhibits some peculiar features that most traditional

More information

The cyclicality of mark-ups and profit margins: some evidence for manufacturing and services

The cyclicality of mark-ups and profit margins: some evidence for manufacturing and services The cyclicality of mark-ups and profit margins: some evidence for manufacturing and services By Ian Small of the Bank s Structural Economic Analysis Division. This article (1) reviews how price-cost mark-ups

More information

ScienceDirect. The Economics and Politics of Process Innovation and The Sustainable Urban Development. Marek Vokoun a, *

ScienceDirect. The Economics and Politics of Process Innovation and The Sustainable Urban Development. Marek Vokoun a, * Available online at www.sciencedirect.com ScienceDirect Procedia Engineering 161 (2016 ) 2229 2233 World Multidisciplinary Civil Engineering-Architecture-Urban Planning Symposium 2016, WMCAUS 2016 The

More information

Chapter 5 FIRM GROWTH, INNOVATION AND THE BUSINESS CYCLE

Chapter 5 FIRM GROWTH, INNOVATION AND THE BUSINESS CYCLE Chapter 5 FIRM GROWTH, INNOVATION AND THE BUSINESS CYCLE The economic crisis that started in 2008 and is still ongoing in many European countries has significantly affected the ability of the EU economy

More information

Researcher Mobility and Innovation: The Effect of Researcher Mobility on Organizational R&D Performance in the Emerging Nations' Companies

Researcher Mobility and Innovation: The Effect of Researcher Mobility on Organizational R&D Performance in the Emerging Nations' Companies Asian Culture and History; Vol. 10, No. 2; 2018 ISSN 1916-9655 E-ISSN 1916-9663 Published by Canadian Center of Science and Education Researcher Mobility and Innovation: The Effect of Researcher Mobility

More information

GLOBAL VALUE CHAINS INTRODUCTION AND SUMMARY DIRECT AND INDIRECT EXPORTS

GLOBAL VALUE CHAINS INTRODUCTION AND SUMMARY DIRECT AND INDIRECT EXPORTS GLOBAL VALUE CHAINS Peter Beck Nellemann and Karoline Garm Nissen, Economics INTRODUCTION AND SUMMARY A final product is created through a chain of activities such as design, production, marketing and

More information

Working Party No. 1 on Macroeconomic and Structural Policy Analysis

Working Party No. 1 on Macroeconomic and Structural Policy Analysis For Official Use ECO/CPE/WP1(2005)2 ECO/CPE/WP1(2005)2 For Official Use Organisation de Coopération et de Développement Economiques Organisation for Economic Co-operation and Development 24-Feb-2005 English

More information

Productivity and R&D sources: evidence for Catalan firms

Productivity and R&D sources: evidence for Catalan firms Economics of Innovation and New Technology Vol. 20, No. 8, November 2011, 727 748 Productivity and R&D sources: evidence for Catalan firms Agustí Segarra and Mercedes Teruel* Industry and Territory Research

More information

Aspects of Statistics on Innovation in Latvia and Some Guidelines for Its Effective Use

Aspects of Statistics on Innovation in Latvia and Some Guidelines for Its Effective Use doi: 10.1515/eb-2016-0019 Aspects of Statistics on Innovation in Latvia and Some Guidelines for Its Effective Use Svetlana Jesiļevska University of Latvia, Central Statistical Bureau of Latvia Abstract

More information

Government at a Glance 2009

Government at a Glance 2009 Government at a Glance 2009 Summary in English Government at a Glance 2009 identifies several key governance challenges and raises fundamental questions facing governments as they reassess their roles,

More information

APPENDIX I. Data Appendix. B. US Patent Awards Data.

APPENDIX I. Data Appendix. B. US Patent Awards Data. APPENDIX I. Data Appendix B. US Patent Awards Data. a. Choice of US patent awards as outcome variable over the WIPO patent counts. There are two main patent measures available in the data I gathered. One

More information

Modeling technology specific effects of energy policies in industry: existing approaches. and a concept for a new modeling framework.

Modeling technology specific effects of energy policies in industry: existing approaches. and a concept for a new modeling framework. Modeling technology specific effects of energy policies in industry: existing approaches and a concept for a new modeling framework Marcus Hummel Vienna University of Technology, Institute of Energy Systems

More information

German participation in the Sixth European Framework Programme for Research and Technological Development

German participation in the Sixth European Framework Programme for Research and Technological Development German participation in the Sixth European Framework Programme for Research and Technological Development Imprint Published by Federal Ministry of Education and Research Orders Federal Ministry of Education

More information

R&D AND THE DYNAMICS OF FIRMS AND SECTORS

R&D AND THE DYNAMICS OF FIRMS AND SECTORS Knowledge for Growth Industrial Research & Innovation (IRI) The Role of Firm Size and Sector Belonging. CONTRIBUTED PAPER FOR THE 007 CONFERENCE ON CORPORATE R&D (CONCORD) R&D AND THE DYNAMICS OF FIRMS

More information

The Fourth Community Innovation Survey (CIS IV)

The Fourth Community Innovation Survey (CIS IV) The Fourth Community Innovation Survey (CIS IV) THE HARMONISED SURVEY QUESTIONNAIRE The Fourth Community Innovation Survey (Final Version: October 20 2004) This survey collects information about product

More information

An Empirical Study of the Relationships between Different Types of Innovation and Firm Performance Shouyu Chen

An Empirical Study of the Relationships between Different Types of Innovation and Firm Performance Shouyu Chen An Empirical Study of the Relationships between Different Types of Innovation and Firm Performance Shouyu Chen Zhejiang Yuexiu University of Foreign Languages, Shaoxing, P.R.China chenshyu@zju.edu.cn Keywords:

More information

Programme Society and Future

Programme Society and Future Programme Society and Future Final report Research Summary RESEARCH CONTRACT: TA/00/23 PROJECT ACRONYM: REFBARIN TITLE: Product market reform, labour bargaining and innovativeness of Belgian firms TEAM

More information

The Role of Education for the Economic Growth of Bulgaria

The Role of Education for the Economic Growth of Bulgaria MPRA Munich Personal RePEc Archive The Role of Education for the Economic Growth of Bulgaria Mariya Neycheva Burgas Free University April 2014 Online at http://mpra.ub.uni-muenchen.de/55633/ MPRA Paper

More information

Foreign Direct Investment and Innovation in Central and Eastern Europe: Evidence from Estonia

Foreign Direct Investment and Innovation in Central and Eastern Europe: Evidence from Estonia Eesti Pank Bank of Estonia Foreign Direct Investment and Innovation in Central and Eastern Europe: Evidence from Estonia Jaan Masso, Tõnu Roolaht, Urmas Varblane Working Paper Series 5/2010 The Working

More information

QUANTIFYING FIRM LEVEL INNOVATION IN PAKISTAN AND ITS CONSEQUENCES FOR PUBLIC POLICY

QUANTIFYING FIRM LEVEL INNOVATION IN PAKISTAN AND ITS CONSEQUENCES FOR PUBLIC POLICY QUANTIFYING FIRM LEVEL INNOVATION IN PAKISTAN AND ITS CONSEQUENCES FOR PUBLIC POLICY Dr. Izza Aftab September 18, 2017 Abstract In this paper we aim to study, understand and analyze firms level innovation

More information

SECTORAL AND FIRM-LEVEL DIFFERENCES IN INNOVATION PERFORMANCE: EVIDENCE FROM FINNISH MANUFACTURING FIRMS

SECTORAL AND FIRM-LEVEL DIFFERENCES IN INNOVATION PERFORMANCE: EVIDENCE FROM FINNISH MANUFACTURING FIRMS Knowledge for Growth Industrial Research & Innovation (IRI) SECTORAL AND FIRM-LEVEL DIFFERENCES IN INNOVATION PERFORMANCE: EVIDENCE FROM FINNISH MANUFACTURING FIRMS CONTRIBUTED PAPER FOR THE 2007 CONFERENCE

More information

Growth, Productivity, and Wealth in the Long Run

Growth, Productivity, and Wealth in the Long Run General Observations about Growth Growth, Productivity, and Wealth in the Long Run Growth is an increase in the amount of goods and services an economy produces. Chapter 7 Growth is an increase in potential

More information

GA No Report on the empirical evaluation of the impact of the EU ETS

GA No Report on the empirical evaluation of the impact of the EU ETS GA No.308481 Report on the empirical evaluation of the impact of the EU ETS Antoine Dechezleprêtre, London School of Economics and Political Science, LSE Executive Summary How have firms responded to the

More information

This policy brief addresses the issue of the complementarity of policies

This policy brief addresses the issue of the complementarity of policies f briefing paper No. 8/October 6, 2014 The promotion of renewable energy innovation When State intervention and competition go hand in hand 1 This policy brief addresses the issue of the complementarity

More information

Patterns of innovation diffusion and technological competition in Portuguese manufacturing and service industries

Patterns of innovation diffusion and technological competition in Portuguese manufacturing and service industries Patterns of innovation diffusion and technological competition in Portuguese manufacturing and service industries Maria Fraga Oliveira Martins Paulo Anciaes ISEGI, Universidade Nova de Lisboa, Portugal

More information

Employment in the EU based on Farmed Norwegian Salmon Short version

Employment in the EU based on Farmed Norwegian Salmon Short version Employment in the EU based on Farmed Norwegian Salmon Short version SINTEF Fisheries and Aquaculture SINTEF Technology and Society Fafo Institute for Labour and Social Research June 2005 Introduction SINTEF

More information

A taxonomy of innovation: How do public sector agencies innovate? Results of the 2010 European Innobarometer survey of public agencies

A taxonomy of innovation: How do public sector agencies innovate? Results of the 2010 European Innobarometer survey of public agencies A taxonomy of innovation: How do public sector agencies innovate? Results of the 2010 European Innobarometer survey of public agencies This report was prepared by Anthony Arundel 1 & Hugo Hollanders 2

More information

Innovation, IP choice, and productivity: Evidence from UK firms

Innovation, IP choice, and productivity: Evidence from UK firms Innovation, IP choice, and productivity: Evidence from UK firms Bronwyn H. Hall 1 Vania Sena 2 December 2011 This report is module 3b for the project The choice between formal and informal intellectual

More information

Economic and Social Council

Economic and Social Council United Nations E/CN.3/2018/28 Economic and Social Council Distr.: General 20 December 2017 Original: English Statistical Commission Forty-ninth session 6-9 March 2018 Item 4 (h) of the provisional agenda*

More information

Micro Data Linking Creating new Evidence by Utilising. existing Statistical Registers. Case: International Sourcing

Micro Data Linking Creating new Evidence by Utilising. existing Statistical Registers. Case: International Sourcing Int. Statistical Inst.: Proc. 58th World Statistical Congress, 2011, Dublin (Session STS055) p.3247 Micro Data Linking Creating new Evidence by Utilising existing Statistical Registers. Case: International

More information

BUSINESS STRATEGIES OF. SMEs AND LARGE FIRMS IN

BUSINESS STRATEGIES OF. SMEs AND LARGE FIRMS IN WORKING PAPER SERIES BUSINESS STRATEGIES OF SMEs AND LARGE FIRMS IN CANADA Working Paper Number 16 October 1997 WORKING PAPER SERIES BUSINESS STRATEGIES OF SMEs AND LARGE FIRMS IN CANADA by Gilles McDougall

More information

MONTHLY REVIEW OF ACADEMIC LITERATURE ON ASPECTS OF HUMAN CAPITAL AND SKILLS IN RESEARCH AND INNOVATION POLICY

MONTHLY REVIEW OF ACADEMIC LITERATURE ON ASPECTS OF HUMAN CAPITAL AND SKILLS IN RESEARCH AND INNOVATION POLICY Issue 10 /May 2016 MONTHLY REVIEW OF ACADEMIC LITERATURE ON ASPECTS OF HUMAN CAPITAL AND SKILLS IN RESEARCH AND INNOVATION POLICY Contact: DG RTD, Directorate A, A.4, Eva Rückert, Tel. 89319, eva.rueckert@ec.europa.eu

More information

Mastering Productivity Growth- Where Canadian Food processing is lagging behind

Mastering Productivity Growth- Where Canadian Food processing is lagging behind Mastering Productivity Growth- Where Canadian Food processing is lagging behind Main takeaways Productivity in food processing is declining Driven by a decrease in technical progress Main drivers differ

More information

Beyond balanced growth: The effect of human capital on economic growth reconsidered

Beyond balanced growth: The effect of human capital on economic growth reconsidered Beyond balanced growth 11 PartA Beyond balanced growth: The effect of human capital on economic growth reconsidered Uwe Sunde and Thomas Vischer Abstract: Human capital plays a central role in theoretical

More information

DETERMINANTS OF THE DECISION TO IMPORT: A CROSS-COUNTRY COMPARISON

DETERMINANTS OF THE DECISION TO IMPORT: A CROSS-COUNTRY COMPARISON DETERMINANTS OF THE DECISION TO IMPORT: A CROSS-COUNTRY COMPARISON Antonio Rodríguez Banco de España Joint work with Cristina Fernández, Coral García and Patry Tello Paper available at http://www.bde.es/f/webbde/ses/secciones/publicaciones/informesboletinesrevistas/boletineconomico/12/oct/files/art2e.pdf

More information

Transitions of creatives? empirical evidence on occupation and industry specific human capital

Transitions of creatives? empirical evidence on occupation and industry specific human capital Paper to be presented at the DRUID Academy conference in Rebild, Aalborg, Denmark on January 21-23, 2015 Transitions of creatives? empirical evidence on occupation and industry specific human capital Cecilie

More information

Online Appendices to Appropriability Mechanisms, Innovation and Productivity: Evidence from the UK

Online Appendices to Appropriability Mechanisms, Innovation and Productivity: Evidence from the UK Online Appendices to Appropriability Mechanisms, Innovation and Productivity: Evidence from the UK Bronwyn H. Hall 1 Vania Sena 2 March 2015 Disclaimer: This work contains statistical data from UK ONS

More information

TUAC Comments on the OECD Employment Outlook Making the case for coordinated and multi-employer collective bargaining systems

TUAC Comments on the OECD Employment Outlook Making the case for coordinated and multi-employer collective bargaining systems TUAC Comments on the OECD Employment Outlook 2018 - Making the case for coordinated and multi-employer collective bargaining systems Table of contents Arguments in favour of decentralised bargaining...

More information

Are different forms of innovation complements or substitutes?

Are different forms of innovation complements or substitutes? MPRA Munich Personal RePEc Archive Are different forms of innovation complements or substitutes? Justin Doran School of Economics, University College Cork 2012 Online at https://mpra.ub.uni-muenchen.de/44580/

More information

Human Capital Mobility into and out of Research Sectors in the Nordic Countries

Human Capital Mobility into and out of Research Sectors in the Nordic Countries Preliminary version Do not quote Ebbe K. Graversen The Danish Institute for Studies in Research and research Policy Human Capital Mobility into and out of Research Sectors in the Nordic Countries The Danish

More information

R&D and productivity growth: evidence from firm-level data for the Netherlands

R&D and productivity growth: evidence from firm-level data for the Netherlands R&D and productivity growth: evidence from firm-level data for the Netherlands Eric Bartelsman *, George van Leeuwen, Henry Nieuwenhuijsen and Kees Zeelenberg 1. Introduction This article presents evidence

More information

Assessing and Measuring Macroeconomic Imbalances in the EU

Assessing and Measuring Macroeconomic Imbalances in the EU Assessing and Measuring Macroeconomic Imbalances in the EU From a Macro to a Micro foundation Carlo Altomonte Bocconi University and Bruegel Rome - 21 May 2011 Altomonte (Bocconi University and Bruegel)

More information

Monitoring, evaluation and impact assessment of innovationenhancing

Monitoring, evaluation and impact assessment of innovationenhancing MLE on Innovation-Enhancing Procurement Vienna, 21st September 2017 Monitoring, evaluation and impact assessment of innovationenhancing procurement Jon Mikel Zabala-Iturriagagoitia Deusto Business School

More information

Working Party on Innovation and Technology Policy

Working Party on Innovation and Technology Policy Unclassified DSTI/STP/TIP(2004)4/FINAL DSTI/STP/TIP(2004)4/FINAL Unclassified Organisation de Coopération et de Développement Economiques Organisation for Economic Co-operation and Development 14-Oct-2005

More information

Summary Report. Question Q183. Employers rights to intellectual property

Summary Report. Question Q183. Employers rights to intellectual property Summary Report Question Q183 Employers rights to intellectual property The environment in which the intellectual property rights are exerted, knew significant changes since the Congress of Venice of 1969

More information

COMMISSION STAFF WORKING DOCUMENT IMPACT ASSESSMENT. Accompanying the document. Proposal for a Council Regulation

COMMISSION STAFF WORKING DOCUMENT IMPACT ASSESSMENT. Accompanying the document. Proposal for a Council Regulation EUROPEAN COMMISSION Brussels, 11.1.2018 SWD(2018) 6 final PART 3/4 COMMISSION STAFF WORKING DOCUMENT IMPACT ASSESSMENT Accompanying the document Proposal for a Council Regulation on establishing the European

More information

research paper series

research paper series research paper series Globalisation, Productivity and Technology Research Paper 2007/26 Productivity spillovers through vertical linkages: Evidence from 17 OECD countries by Jürgen Bitzer, Ingo Geishecker

More information

Working Party on Statistics

Working Party on Statistics For Official Use For Official Use Organisation de Coopération et de Développement Economiques Organisation for Economic Co-operation and Development 10-Jul-2002 English - Or. English DIRECTORATE FOR SCIENCE,

More information

Appendix to Gerlagh, Mathys and Michielsen, Energy Abundance, Trade and Specialization (ej371_02)

Appendix to Gerlagh, Mathys and Michielsen, Energy Abundance, Trade and Specialization (ej371_02) Appendix to Gerlagh, Mathys and Michielsen, Energy Abundance, Trade and Specialization (ej371_02) Appendix 1. Data Description Table A. Countries in the sample. Abbreviation Country AUS* Australia BEL

More information

The relationship between innovation and economic growth in emerging economies

The relationship between innovation and economic growth in emerging economies Mladen Vuckovic The relationship between innovation and economic growth in emerging economies 130 - Organizational Response To Globally Driven Institutional Changes Abstract This paper will investigate

More information

CONSUMER ENVIRONMENT IN THE COMMON EUROPEAN MARKET

CONSUMER ENVIRONMENT IN THE COMMON EUROPEAN MARKET Trakia Journal of Sciences, No 3, pp 330-334, 2013 Copyright 2013 Trakia University Available online at: http://www.uni-sz.bg ISSN 1313-7069 (print) ISSN 1313-3551 (online) Original Contribution CONSUMER

More information

11 EQ7: Structure of dairy industry

11 EQ7: Structure of dairy industry 11 EQ7: Structure of dairy industry To what extent have the CAP measures applicable to the dairy sector influenced structural changes in the processing sector? 11.1 Interpretation and comprehension of

More information

International sourcing and employment effects a micro data linking approach

International sourcing and employment effects a micro data linking approach International sourcing and employment effects a micro data linking approach 8 International sourcing and employment effects a micro data linking approach 8.1 Introduction 8.2 Literature review on international

More information

EDUCATION POLICY ANALYSIS (Phillip McKenzie, 3 November 2003)

EDUCATION POLICY ANALYSIS (Phillip McKenzie, 3 November 2003) EDUCATION POLICY ANALYSIS 2003 (Phillip McKenzie, 3 November 2003) Purpose of the Series Improving the quality of education is a key policy objective in OECD countries. Major education reforms are underway

More information

Determinants of innovation diffusion in the EU: A microeconometric analysis of firms innovation adoption choices 1

Determinants of innovation diffusion in the EU: A microeconometric analysis of firms innovation adoption choices 1 WP4/09 SEARCH WORKING PAPER Determinants of innovation diffusion in the EU: A microeconometric analysis of firms innovation adoption choices Corinne Autant-Bernard, Jean-Pascal Guironnet, Nadine Massard

More information

R&D and Productivity: Evidence from large UK establishments with substantial R&D activities

R&D and Productivity: Evidence from large UK establishments with substantial R&D activities R&D and Productivity: Evidence from large UK establishments with substantial R&D activities Stephen R Bond a and Irem Guceri b a Nuffield College, Department of Economics, and Centre for Business Taxation,

More information