Factors affecting innovation revisited

1 Factors affecting innovation revisited José Molero y Antonio García WP05/08 2 3 Resumen El propósito de este trabajo es contribuir a un mejor conocimiento de los factores que afectan a la innovación mediante el análisis de los microdatos de la encuesta de innovación de las empresas españolas de 2003. El estudio se aborda desde la elaboración de una taxonomía de sectores combinando las Ventajas Tecnológicas Reveladas de la industria española con el dinamismo tecnológico mundial; además se introduce una clasificación de las empresas en función de la pertenencia o no a un grupo de empresas y de si esos grupos son de nacionalidad española o extranjera. Se utilizan técnicas de Análisis Factorial para reducir y organizar la abundante información disponible en Factores con significado económico que después son empleados como variables explicativas de la innovación de producto y de proceso. Se encuentran diferencias entre ambos tipos de innovación tanto por el número de factores significativos como en la intensidad de su capacidad explicativa. La taxonomía elaborada muestra su importancia al mostrar patrones de comportamiento distintos entre los cuatro tipos de casos construidos. Abstract The aim of this paper is to contribute to a better understanding of factors affecting innovation by analysing the Spanish manufacturing sector using microdata of the 2003 Spanish Innovation Survey. To enrich the analysis a self developed sectoral taxonomy is used coming from the combination of both of the sectoral Revealed Technological Advantages (RTA) and worldwide technological dynamism of the sectors; moreover firms are classified according to the type of capital ownership: independent companies, companies belonging to a national group and subsidiaries of multinational enterprises. The abundance and heterogeneity of variables advised us to use Factor analysis to reduce and organise the original variables into a number of consistent and theoretically significant factors. We found differences between product and process innovation, both in number of explicative variables (significant independent variables) and in relative effect of independent variables (even, in some cases, a sign change from product to process innovation). Taxonomy matters because of some differences in explanatory (independent) variables for each sector and model explanatory power differences between sectors, and, on the other hand, because of the “non significance” of some significant variables once we control by sectoral taxonomy. Key words: Innovation, Factors affecting innovation, Multinational enterprises, Sectoral taxonomies, Spain. José Molero, Antonio García Instituto Complutense de Estudios Internacionales, Universidad Complutense de Madrid. Campus de Somosaguas, Finca Mas Ferre. 28223, Pozuelo de Alarcón, Madrid, Spain. Facultad de Ciencias Económicas y Empresariales. Universidad de Sevilla. Avda Ramon y Cajal, 1. 41018, Sevilla, Spain. © José Molero, Antonio García ISBN: Depósito legal: El ICEI no comparte necesariamente las opiniones expresadas en este trabajo, que son de exclusiva responsabilidad de sus autores. 4 5 Índice 1. Introduction…………………………………………………………………………….……..7 2. Theoretical background……………………………………………………………………….8 3. Methodology and data analysis………………………………………………...………...…..11 4. Some stylised facts of the innovation in the Spanish economy………………………….….13 5. Results…………………………………………………………………………………………15 5.1 Factor analysis ..…………………………………….………………………………..15 5.2 Regression analysis………...…………………………………………………….…..16 6. Concluding remarks…………………………………………………………………….……19 7. Annexe A……………………………………………………………………………………...21 Annexe B………………………………………………………………………………………22 8. References……………………………………………………………………………………..26 6 7 1. Introducción Innovation has experienced a remarkable change in recent years as a consequence of a number of factors including the advance of science and technology and the increasing globalisation of a number of markets and activities. The growing heterogeneity of sources affecting the process of firms’ innovation has led to the knowledge created out of the companies themselves achieving greater importance, and therefore to the central role to be played by the capacity of integrating inner and outer sources of technological capabilities with other competitive forces. Similarly, the acceleration of internationalisation at most economic and social levels has increased the necessity for exploiting firms’ advantages at international (sometimes world) level and seeking new competitive (technological) assets in a multinational framework. Moreover, the specialised research has reached a common conclusion that sectoral features have a remarkable influence on the possibilities and organisational modes of innovatory activity. From Pavit’s seminal sectoral taxonomy on innovation, there is a long tradition of using several taxonomies or “classificatory list” of productive sectors according to their innovative characteristic or intensity. Both academic and institutional taxonomies can be described as “closed aprioristic lists” of sectors, built on rigorous studies but without flexibility to allow country differences on sector characteristics. A main contribution of this paper consists of grouping companies in different categories of sectors in a non-aprioristic way: each sector is “self-classified” in a specific type and not in any other according to particular characteristics and “innovative behaviour” in a given country (in our case, Spain). To this end we have developed a taxonomy from the combination of (RTA1) of each sector of activity and the evolution of its world weight between 1993-1998 and 19992003. Original sectoral data come from patents granted by the USPTO by priority year at the national level by sector of economic activity (NACE class derived through concordance with International Patent Classification). 1 RTAs for a sector of a country is calculated as follows: RTAij = (Pij / Pwj )/ (PTi / PTw) where i is the country, j the sector, w the world total for j sector, T is the total of the country and tw is the absolute world total. All referred to patents in a period of time. Using microdata from the Spanish Innovation Survey2, we had the advantage of their quality and statistical significance. These data are of a great statistical validity insofar as an expert group coordinated by the Spanish National Statistics Institute has, on the one hand, drawn up a permanent sample of firms with the intention of creating a stable panel of data and, on the other, has controlled the statistical significance of the anonymous data vis-à-vis the original micro data. In addition, this information allows us to separate firms according to their independent feature or belonging to a group, including the country of origin of the mother house of MNCs. Our topic to explain is the innovative activity (product and process innovation) of Spanish manufacturing firms (both national and multinational ownership), using as explanatory variables those included in the Innovation Survey. The abundance and heterogeneity of variables advised us to use Factor analysis to reduce and organise the number of original variables in a number of consistent and theoretically significant factors. We have made a series of sectoral analyses in an iterative way using both PITEC original variables and its transformations. Non-included variables have been rejected on the basis of KMO and MSA values. Once relevant factors are retained, we have made two series of logistics regressions for both process and product innovation implementation, with factors and firms ownership as independent variables. In each series, firstly we regress once for the whole sample and secondly we regress four times more controlling by sectoral taxonomy. In the next section, we present the main theoretical background, in the third section we discuss our data and methodological guidelines, in the fourth section we present some stylised facts of innovation activities in Spain for a better understanding of the context in which our empirical work is done, in the fifth section we show the empirical results and finally in the sixth section we enhance the main conclusion 2 It is called the PITEC panel (Panel de Innovacion Tecnologica). See: www.technociencia.es. Due to the manipulation of the original survey data, the information of this panel cannot be compared in absolute terms with data directly derived from the innovation survey used in the former section. 8 2. Theoretical background It is not our intention to develop a complete theory of factors affecting innovation. First of all this is because such a theory is not available, as demonstrated by both the attempts to gather the most relevant empirical investigations (Tidd, Bessant and Pavitt, 1997), and the list of issues included in seminal books as “issues affecting innovation” (Rothwell and Dodgson, 1995; Fagerber et al, 2006). On the contrary, our aim is to briefly review the most significant contributions to find some theoretical guidelines for our empirical research. In this regard, to briefly summarise the panorama of theoretical visions of the conditions which promote successful innovative activities of firms, we can cluster them together in two fundamental groups. On the one hand are those belonging to the long tradition of industrial organisation theory which basically search for a few determining factors of the capacity of innovation and, on the other, those more representative of the relatively recent approach of the evolutionary theory which is based on the analysis of variety and diversity of innovative modes and frequently uses the classification of cases as a basic theoretical tool. Starting from the tradition of industrial organisation studies, one must remember their central methodological characteristics consisting of the combination of the three classical steps of structure-conduct-performance (Scherer and Ross, 1990). As formerly for other economic issues, the analytical work is oriented to finding out one or a few factors (never a large number) which can explain the innovatory activity of the firms in a satisfactory way3. Without a doubt these two are the most frequently researched elements: the size of the firm and the concentration level of the markets in which firms carry out their innovative activities. In spite of the huge number of studies made about both of theses issues, the current situation is that there are no conclusive results allowing us to assert the sign and intensity of the impact those factors have to induce innovation. In fact, as far as the size of the firm is concerned, we can share the argument of 3 Orsenigo (1989) suggests they use a dichotomous perspective: large versus small size, concentrated versus non-concentrated markets, etc. Cohen (1995) calls them studies in the Schumpeterian tradition. Freeman and Soete (1997, page 193) that the size “certainly influences what kind of projects can be attempted in terms of technology, complexity and costs but does not in itself determine the outcome”. This inconclusive conclusion is the result of a large amount of empirical research relating to the size of the firms and innovative activities of the firms. Perhaps the most classic one is the association of size with R&D expenditure. In this case, the available research shows there is a concentration of R&D expenditure in large companies, basically determined by the size of R&D programmes instead of the size of the firm (Freeman and Soete, 1997). Nevertheless, it is more difficult to find a clear association between the increase of the size and the intensity of R&D expenditure. In fact, mainly after controlling by sector, the association seems to follow a growing trend (the larger the size the more intense is R&D effort) but just to a certain extent; from this point onward the dominant relation is a proportional one (Cohen, 1995). With regard to small firms, the evidence points to a twofold situation: whereas a vast majority of small firms do not perform any specialised R&D programme, in several countries those small firms that do perform R&D have above average R&D intensities (Freeman and Soete, 1997, p. 232). The situation is even less clear if innovation replaces R&D activities; more qualifications have to be incorporated, such as, for example, the possible advantage of small companies in early stages of innovative work and the less expensive but more radical innovation, whereas large firms have advantages in the later stages and improvement and scaling up of early breakthroughs (Freeman and Soete, 1997, 234). More generally, Rothwell and Dodgson (1995, 323) arrive at the following conclusions: I. Innovatory advantage is unequivocally associated neither with large nor small companies. Small firm advantages are mainly behavioural while those of large firms are mainly material. II. Available data suggests that the firm size/innovation share relationship is Ushaped. III. Small firms´ innovatory contribution varies significantly from sector to sector. 9 IV. Small and large firms do not operate in isolation from each other and they enjoy a variety of complementary relationships in their technological activities”. V. Any study of the roles of small and large firms in innovation should be dynamic: their relative roles vary considerably over the industry cycle (Shepherd, 1991; Utterback, 1994). The situation is more confusing if the aim is to associate levels of concentration (or monopolistic power) with a superior innovative performance. The reason is that there are two different angles to approach that relationship: one is the passive and direct association of higher levels of concentration with more intensive behaviour and the second the more dynamic and complex one that postulates the need for a monopolistic reward to encourage innovative activities (Orsenigo, 1989; Cohen, 1995). In the first case no strong conclusive result can be shown and the second is very difficult to test; therefore, some idea of a simultaneous determination of concentration and innovation can be proposed (Cohen, 1995). To complete that uneven perspective we can conclude that to generalise about size of the firms, scale of R&D, inventive output and innovation needs to be heavily qualified. Industry, technology and history matter4 (Freeman, 1982; Freeman and Soete, 1997). There are not many other features so deeply investigated. In general, progress has been poor in topics such as the influence of cash flows or diversification (Cohen, 1995). Nevertheless, it is important to highlight the importance given to some organisational characteristics in the seminal SAPPHO project (Rothwell et al, 1974) and the renaissance of those elements in the literature and policy practices (Nelson&Sampat, 2001; Nelson, 2008; OECD, 2005). Following a methodology of comparing success and failure of innovation, the SAPPHO project spotlighted a number of organisational variables, mainly “marketing related”, “external communication” and “firm management”. 4 It is important to remember a number of empirical limits which seriously make it difficult to arrive at general conclusions. According to Cohen (1995), we can mention the fact that samples are mostly not random, in many cases there are other firms variables out of control and the multisectoral character of most large firms. Freeman cleverly summarised it by saying: “the fact that the measures which discriminated between success and failure include some which reflected mainly on the competence on R&D, others which reflected mainly on efficient marketing and some which measured characteristics of the business innovator with good communications, confirms that view of industrial innovation as essentially a coupling process” (Freeman, 1982, 125). The former notwithstanding, the innovation theory reoriented the approach to investigate the role of the firms. Precisely the cited book of Freeman, together with seminal works such as Rosenberg 1976, Nelson and Winter, 1982, Dosi, 1984 and Pavitt, 1984 made claims for a better understanding of technology and innovation that enables to create a more accurate theoretical approach. Apart from other general considerations (Orsenigo, 1979; Dosi, 1984) as far as this paper is concerned, the central change has to do with the introduction of variety and diversity as opposed to the main stream perspective of general determinants of firms’ innovation; furthermore, the notion of “learning” occupied a central position as a consequence of considering technology and innovation as specific kinds of knowledge (Pavitt, 2006). As a consequence of this new orientation the approach to firms’ innovatory behaviour incorporates a number of remarkable modifycations. Four of them are particularly significant for the objectives of this investigation: the role of systems of innovation, the importance of non-technological innovation, the relevance of the sector of activity and the introduction of taxonomies to map the heterogeneity of conducts. The concept of systems of innovation (Freeman, 1987; Lundvall, 1992) basically reinforced the idea of systemic interaction among firms and a large amount of institutions. Thus, in the context of this paper, the relevant issue is the necessity of incorporating internally and externally sources of knowledge to the firm, either on a collaborative mode or through market mechanisms. In other words, the isolated consideration of firms´ characteristics is not enough to fully understand the innovative practice of enterprises (Rothwell, 1995; Tidd, Bessant and Pavitt, 1997); hence variables measuring such an interaction must be incorporated. 16 2 factors between 9.5% and 10.5%; level II, 3 or 2 factors between 7% and 8.5%; level III, 3 factors between 5.5% and 6,5%; level IV, 2 factors around 4,5%). The more solid Factor structures are: • NESTR, NGEST, NORG, NMARK, NESTE, always in the first level and usually the first factor. • GMAQUI, GTECNO, GPREP, usually in the second level, we find a tendency to incorporate GFORM also. • PATNUM, FPRO, usually in the second level. GINTID has a tendency to link this factor. • INNOVEM, INNPROD, sometimes in the second level, others in the third. • PIDTEJpw, FAP, sometimes in the second level, others in the third. • TAMMED, GEXTID, GMARKET, usually in the third level. • INNNOVET, INNPROC, always in the last level. • INNOVEC, INFUN, always in the last level too. Our selected Factor structure is shown in Table 1 and its quite reasonable possibility of being interpreted in economic and innovative terms must be underlined; each factor is accompanied by the variables that cluster together and the name we have assigned to make explicit its economic or innovative significance. TABLE 1: extracted factors. EXPLAINED VAR (%) Factor DESCRIPTION Each factor Accumulated 1 Organizational Innovation (ORGINNV) 10.53 10.53 2 Non-R&D innovative expenditures (NRDEXPEND) 10.42 20.95 3 Own R&D and results (RD&PAT) 7.5 28.45 4 Product innovation on inner effort (PRODINNER) 7.15 35.6 5 Human resources and Public funds (HUMCAP&FUNDS) 6.15 41.75 6 Size and external knowledge integration (SIZE&INTEGRATION) 6.05 47.8 7 EU funds attraction (UEFUNDS) 5.81 53.61 8 Basic research and cooperation (BASICR&COOP) 4.82 58.43 9 Process innovation on external sources (PROCEXTER) 4.81 63.24 Source: Own elaboration from PITEC; principal components method. 5.2 Regression analysis. This phase of the analysis has been done through the estimation of regression models using the former factors as independent variables to explain the innovative activity of the firms. Following the research path proposed, we have estimated a number of logistic regressions with several characteristics: • We have separated product and process innovation on the understanding that there can be significant differences between them. In both cases models try to explain the probability of any firm belonging to the innovative group or not. • Apart from the factors we have included as independent variables two dummies trying to control the fix effect of firms belonging to any of the three groups in which we have divided the sample: independent Spanish companies (EIN), firms belonging to a national group (GN), and firms belonging to foreign multinational groups (GMN). The idea is to consider the importance of the growing interna- 17 tionalisation of the innovation in the Spain. • In addition to the general regressions for products and processes we have estimated in each case four other regretssions corresponding to the types of sectors formerly explained. The results are shown in tables 2 and 3 from which we can deduce as the main findings the following. TABLE 2: Logit coefficients for “innprod” (product innovation) Variable GLOBAL DYNAMIC STATIC LOST OPP. RETREAT ein .226** gmn basicr_coop .209*** .241*** .182* humcap_funds .981*** .977** .526*** 1.29*** nrdexpend .406** -5.26*** .804** -6.13*** orginnv .424*** .435*** .402*** .439*** .393*** rd_pat .449*** 1.57*** .65*** Size_integr .238** 2.06*** .324*** .146* 2.12*** uefunds .195** .46*** .437* constant .677*** 1.22*** .618*** .915*** .582*** N 3691 805 1362 1086 438 ll -2213 -427 -852 -634 -250 ll_0 -2395 -500 -900 -695 -294 chi2 363 144 96.1 123 87 r2_p .0758 .144 .0534 .0881 .148 aic 4445 869 1713 1284 509 bic 4500 902 1739 1324 525 legend: * p<0.05; ** p<0.01; *** p<0.001 Source: Own elaboration from PITEC data. TABLE 3: Logit coefficients for “innproc” (process innovation) Variable GLOBAL DYNAMIC STATIC LOST OPP. RETREAT ein .383*** .666*** gmn basicr_coop .886*** .973*** 1.12*** .971*** .768*** humcap_funds -.471*** -3.76*** -.969*** -.436*** -3.12*** Nrdexpend .499*** -8.55*** -1.06*** .886*** -10.6*** orginnv .597*** .475*** .65*** .532*** .43*** rd_pat .128* -1.75*** -.297** .291** Size_integr .502** Uefunds 1.89*** 1.87*** 1.09*** constant -.213*** -.803*** -.196* -.0482 -.455*** N 3691 805 1362 1086 438 Ll -2033 -390 -726 -598 -225 ll_0 -2552 -551 -944 -752 -297 chi2 1037 322 436 309 145 r2_p .203 .292 .231 .205 .244 aic 4083 791 1470 1210 459 bic 4132 820 1517 1245 480 legend: * p<0.05; ** p<0.01; *** p<0.001 Source: Own-elaboration from PITEC data. 18 GENERAL FINDINGS. 1. Generally speaking we obtain a better explanation for process than for product innovation. It is reflected in the higher pseudo R2 values and in the higher homogeneity found across types of sectors. Nevertheless, significant factors are rather similar, although size & integration is more important in product innovation. From a more speculative point of view we ask if those differences are a reflection of the Spanish innovation characterised, among other elements, by a lesser presence of radical and strategic innovations and a heavier weight of adoptions and by a lesser dependence on foreign inputs in process technologies (Buesa& Molero, 1992; Molero, 2007). 2. The association between human capital (personnel) and the access to public funds changes its signs: positive for product innovation and negative for process. All the remaining factors maintain the same sign in both cases. 3. The kind of sector emerges as much more important than the type of firms´ ownership, confirming what was expected. Once we control for the first, we only find a case in which the type of firm is relevant: ICs in process innovation of Stationary Specialisation sectors. Importantly enough it must be said that this group gathers a non negligible number of so called traditional industries in which independent companies are still the bulk of the activity. 4. Organizational innovation arises as the most important factor explaining the behaviour of innovative firms. Its universal presence in both types of innovation and in all kinds of sectors, always with a positive sign, allows us to assert that this new evidence confirms and reinforces what theory and previous studies predicted. 5. After controlling for type of sector, a number of differences can be found. It is so both for the explanatory capacity of the models and for concrete factors which are significant in some cases but not in others (sometimes changing the sign). In our opinion this confirms and justifies the relevance of the taxonomy used. Moreover the level of adjustment varies significantly between the sectors better adapted to the international dynamism (Dynamic Specialisation and Retreat) and the other two more imbalanced sectors (Lost Opportunities and Stationary Specialisation). PRODUCT INNOVATION 1. The type of firm is not significant in any of the cases. However the size & integration factor has a favourable effect in all cases. 2. Expenditures on non R&D innovative activities have a positive impact in Lost Opportunities, negative in Dynamic Specialisation and Retreat and are nonsignificant in Stationary Specialisation. This points out to a greater necessity of proper R&D activities in more dynamic sectors to follow the world evolution whereas in dynamic sectors in which Spain has technological disadvantages, non R&D activities can be considered as an alternative to build some capacities to be completed afterwards through R&D. 3. The already mentioned role of organisational innovation shows a higher effect in the two sectors with weaker adaptation to the world dynamic: Stationary Specialisation and Lost Opportunities. In other words, it is for those sectors that this non technological innovation seems to play its most important role. 4. The joint effect of R&D expenditure and Patents is a factor which positively influences product innovation in internationally dynamic sectors, regardless of having technological advantages or disadvantages. That is to say, in mature sectors the technological effort measured through R&D and patents is not a discriminating factor between innovative and non innovative firms; whereas it positively dis- 19 criminates in dynamic sectors. In these cases R&D and patents appear as a requirement to get technological advantages. A similar comment can be made with regard to UE funds; this factor can be understood as an index of innovative dynamism. 5. Basic research and Cooperation is only significant in the two sectors which reveal a bad international adaptation: Stationary Specialisation and Lost Opportunities. Therefore we can send a message for policy considerations: to strengthen those activities is essential to improve the situation of the firms in Lost Opportunities sectors. PROCESS INNOVATION 1. Independent companies are more active in Stationary Specialisation sectors, together with Size & Integration factor. The weight of traditional sectors with independent companies –some of them of noticeable sizeallows us to interpret this result as a situation in which foreign companies come to “follow” or “adapt” the behaviour of national companies with remarkable technological advantages. 2. Basic Research & Cooperation is favourable in the four cases, albeit its effect is relatively higher in sectors with bad international adjustment: Lost Opportunities and Stationary Specialisation. This result supports what was said in the analysis of product innovation: financing basic research and cooperative activities positively influences innovation in product and process. 3. Non R&D innovative expenditures have a negative impact in all sectors except in Lost Opportunities. Moreover, the size of the effect is much greater in sectors positively adapted to international evolution. Again, most dynamic sectors are much more dependent on R&D while in other cases non R&D activities can be a preliminary step to create R&D capabilities. 4. Organisational innovation, as always, favours innovation and its effect is greater in internationally dynamic sectors, particularly if it is a sector with technological disadvantages. 5. The positive sign of Human Capital and Funds suggests in processes with more systematic and incremental innovation, the stability of human resources is important and they seem to be strongly connected with the availability of public funds. A complementary way of exploiting the findings can be made through the point of view of the kinds of sectors to highlight some relevant regularities. The starting point is the existence of more similarities between sectors with good fitness concerning international dynamism, on the one hand, and between those badly adapted, on the other. As far as the first two are concerned we can point out that in the Dynamic Specialisation category two factors are common to product and process innovation: Organisational Innovation, with a positive impact, and non R&D expenditures, with a negative one. Another two show different signs in product and process: Human capital and Funds and R&D and Patentsboth have a positive effect in product innovation and a negative effect in process. Regarding Retreat Sectors, only two factors arise as common to product and process: The salient fact is that Organisational Innovation and Non R&D expenditures are common to both Dynamic and Retreat and both types of innovation, product and process. The first has a regular positive influence and the second a negative one. In Dynamic sectors the role of human capital and funds and R&D and patents is also remarkable. The two are significant for both types of innovation, although with different signs; positive for product and negative for process. Coming to sectors with negative international adjustment, it is noticeable that there is less regularity than in the former cases. The most outstanding are the following. Apart from the systematic presence of organisational innovation, there also arises the 20 regular presence of Basic Research and Cooperation. This, on the one hand, explains the insufficient role played insofar as it has not been enough to guarantee a better international adjustment. This is why we insisted on the idea that policy has to foster those activities in order to upgrade the situation. A second regularity has to do with Human Capital and Funds. This factor is common to the two sectors and types on innovation, although for the last one, the sign changes from positive in product to negative in process. 6. Concluding remarks. As on many occasions, the distance between the theoretical debate and empirical findings has been non-negligible in this research exercise. In fact, the theory about factors affecting firms´ innovation has still a long way to go because the analytical object is complex and difficult to set limits for, as our review of the literature has shown. In spite of those difficulties we have tried to cast some light on the bases of the study of the Spanish case with some noticeable features: first, the importance given to the category of sector by constructing a new taxonomy through the combination of sectoral technological RTAs and the international dynamism; second, the utilisation of a very reliable database, which, based on the CIS has been statistically improved and ready to be used prepared by a group of specialists; third, the combination of a very large number of variables in order to capture the richness and complexity of the innovatory process; fourth, the inclusion of a separate study of product and process innovation; and fifth, the incorporation of the type of company by differentiating between national independent enterprises from either national or multinational groups. The results obtained allow us to make explicit the following remarks: 1. First of all, the methodology has demonstrated its usefulness because both the sectoral taxonomy and the separation of product and process innovation show significant differences. We should like to underline the importance of typological analysis in the overall theoretical effort. 2. Particularly important are some findings about differences across categories of sectors. If in a preliminary approach we differentiate between the positive diagonal (dynamic and retreat sectors) and the negative one (lost opportunities and stationary specialisation), we arrive at the conclusion that our estimations are stronger for the positive one, both for product and process innovation. In other words, the influence of the variables grouped in factors, allows us to assert that there is a better integration and functioning in the cases in which the Spanish specialisation coincides with the world dynamic. 3. Similarly, although less clearly, the separation of product and process has shown its usefulness. In general terms the better adjustment of the model for explaining process innovation is adequate to Spanish innovative specialisation which is more process oriented. Even though with the necessary caution, we can remember the demonstrated bias of the Spanish pattern towards less intensive and more adoption based kinds of innovations. Furthermore, the extraordinary importance of machinery and equipment purchases as sources of innovation for a majority of firms fits well into our findings. 4. On the contrary, the type of firms according to the distinction between national and foreigners does not shed much light. Very much in accordance with previous works (Molero & Garcia, 2008), our results do not find extraordinary differences between companies belonging to national or multinational groups. Although a number of qualifications must be introduced regarding the organisation of innovative process, our study finds more similarities than differences regarding the innovatory efficiency. Thus notwithstanding, we find out some differences for companies which are not members of a group (regardless of its nationality) particularly in sectors named as “traditional” or of low to medium technological intensity. 5. To explain differences sectoral taxonomy is clearly more relevant than firm ownership typology. This fact 21 point to, both, on the relevance of taxonomy and on the adaptative strategies of MNCs. 6. Public grants enhance product innovation, perhaps because they help to solve problems with human resources expenditures on concrete project development. But, on the other hand, they hamper process innovation, revealing weakness in human resources expenditure financing while for process innovation stability and temporal continuity they are urgently required. ANNEXE A: SECTORAL CLASSIFICATION Two criteria have been combined: the position of the sector in the Spanish economy according to its Revealed Technological Advantage (RTA) and the international dynamism of the sector in terms of its percentage in world total patents. The source has been the US patent office and the period selected 1993-2003, divided into two sub periods. 1993-1998 and 1999-2003. Patents have been obtained at two digit level of NACE classification as provided by Eurostat. RTA has been calculated for the sub period 1999-2003 while the technological dynamism of the sectors has been estimated through the difference between the percentages each sector has in total patents in the second period compared with the same percentage in the first one. Combining the two criteria we arrive at a typology with four cases: 1: sectors with RTA > 1 and an increase of its percentage in world patents between the two periods ( Dynamic Specialisation ). 2: sectors with RTA<1 and a decreasing participation in total patenting ( Retreat ). 3: sectors with RTA > 1 and a decreasing participation in world technological dynamism ( Stationary Specialisation ) and 4: sectors with RTA < 1 and an increasing participation in total patents ( Lost Opportunities ). SECTORAL CLASSIFICATION Lost Opportunities Sectors Tanning, dressing of leather; manufacture of luggage Manufacture of fabricated metal products, except machinery and equipment Manufacture of office machinery and computers Manufacture of electric motors, generators and transformers Manufacture of accumulators, primary cells and primary batteries Manufacture of lighting equipment and electric lamps Manufacture of electrical equipment n.e.c. Manufacture of electronic valves and tubes and other electronic components Manufacture of television and radio transmitters and apparatus for line telephony and line telegraphy Manufacture of television and radio receivers, sound or video recording or reproducing apparatus and associated goods Manufacture of industrial process control equipment Manufacture of optical instruments, photographic equipment Manufacture of watches and clocks Manufacture of motor vehicles, trailers and semi-trailers Manufacture of other transport equipment Dynamic Specialisation Sectors Manufacture of textiles Manufacture of wearing apparel; dressing; dyeing of fur Manufacture of basic metals Manufacture of machinery for the production and use of mechanical power, except aircraft, vehicle and cycle engines Manufacture of other general purpose machinery Manufacture of agricultural and forestry machinery Manufacture of machine-tools (split into DK2941, DK2942 and DK2943 in NACE Rev.1.1) Manufacture of other special purpose machinery Manufacture of domestic appliances n.e.c. Manufacture of electricity distribution and control apparatus, manufacture of insulated wire and cable 22 Manufacture of furniture; manufacturing n.e.c. Retreat Sectors Manufacture of tobacco products Manufacture of wood and of products of wood and cork, except furniture; manufacture of articles of straw and plaiting materials Manufacture of other non-metallic mineral products Manufacture of medical and surgical equipment and orthopaedic appliances Manufacture of instruments and appliances for measuring, checking, testing, navigating and other purposes, except industrial process control equipment Stationary Specialisation Sectors Manufacture of food products and beverages Manufacture of pulp, paper and paper products Publishing, printing, reproduction of recorded media Manufacture of coke, refined petroleum products and nuclear fuel Manufacture of basic chemicals Manufacture of pesticides and other agrochemical products Manufacture of paints, varnishes and similar coatings, printing ink and mastics Manufacture of pharmaceuticals, medicinal chemicals and botanical products Manufacture of soap, detergents, cleaning, polishing Manufacture of other chemical products Manufacture of man-made fibres Manufacture of rubber and plastic products Manufacture of weapons and ammunition Source: Own elaboration ANNEXE B TABLE B1: Calculated Variables. VARIABLE CALCULUS FROM PITEC VARIABLES AND DESCRIPTIÓN EGTINN (GTINN/CIFMED*100) Innovation effort GTINNpw (GTINN/TAMMED) Expenditure in innovation per worker PIDTEJCpw (PIDTEJ/TAMMED*100) R&D staff per 100 workers. PGINTID (GINTID/GTINN*100) % Internal R&D expenditure PGEXTID (GEXTID/GTINN*100) % External R&D expenditure PGID (PGINTID+PGEXTID) % R&D expenditure PGMAQUI (GMAQUI/GTINN*100) % Expenditure in acquisition of machines, equipment and software PGTECNO (GTECNO/GTINN*100) % Expenditure in acquisition of external know-how PGPREP (GPREP/GTINN*100) % Expenditure in preparation for production/distribution. PGFORM (GFORM/GTINN*100) % Expenditure in training PGMARKET (GMARKET/GTINN*100) % Expenditure in introduction of innovations FPRO (F1) Own funds FEMP (F2+F3+F4) Funds from other firms FEMPEXT (F11+F12) Funds from other foreign firms FAP (F5+F6+F7+F8) Funds from AAPPs FUNI (F9+F15) University funds FIPSFL (F10+F16) Funds from IPSFLs Source: Own elaboration from PITEC. 23 TABLE B2: Variables. NAME DESCRIPTIÓN TAMMED Average number of workers in the sphere of activity to which the firm belongs. CIFMED Average figure of businesses in the sphere of activity to which the firm belongs EXPMED Average export volume in the sphere of activity to which the firm belongs. INVMED Average gross investment in material goods in sphere of activity it belongs to. INNOVE Carries out innovation activities. INNOVEM Innovation developed by firm or group. INNOVEC Innovation developer in cooperation with other firms or institutions. INNOVET Innovation developer by other firms or institutions. INNPROD Innovation products from (t-2) to t INNPROC Innovation process from (t-2) to t GINTID Figure for internal R&D expenditure. GEXTID Figure for external R&D expenditure. GMAQUI Figure for expenditure on acquisition of machinery, equipment and software. GTECNO Figure for expenditure on acquisition of external know-how. GPREP Figure for expenditure on product/distribution preparation.. GFORM Figure for expenditure on training. GMARKET Figure for expenditure on introduction of innovations. EGTINN Innovation effort PIDTEJpw EJC staff in R&D per 100 workers. INFUN Basic or fundamental research INAPL Applied research DESTEC Technological development. FPRO Own funds FEMP Other firms’ funds FEMPEXT Funds from other foreign firms. FAP AAPP funds FUNI University funds FIPSFL IPSFL funds FUE EU program funds FEXT Other funds from abroad COOPERA Cooperated from (t-2) a t with other firms PAT Request for patents PATNUM Number of requests for patents PATOEPM OEPM Patents PATEPO EPO Patents PATUSPTO USPTO Patents PATPCT PCT Patents PATINT International Patents (outside OEPM) NESTR Non-technological innovation strategy NGEST Non-technological innovation management NORG Non-technological innovation organization NMARK Non-technological innovation : marketing NESTE Non-technological innovation: aesthetic or subjective change Source: Own elaboration from PITEC. 24 TABLE B3: Correspondences of PITEC sectoral classification and RTA-Dynamism sectoral typology. PITEC SECTOR RAMAI D DESCRIPTION NACE_RE V_1 SECTORAL TYPOLOGY 02 Manufacture of food products and beverages DA_15 3 (lost opportunities) 03 Manufacture of tobacco products DA_16 2 (retreat) 04 Manufacture of textiles DB_17 1 (dynamic specialisation) 05 Manufacture of clothing / dressing and dyeing of fur DB_18 1 (dynamic specialisation) 06 Tanning and dressing of leather/ manufacture of luggage, handbags, saddlery, harness and footwear DC_19 4 (stationary specialisation) 07 Manufacture of wood and of products of wood and cork, except furniture/ manufacture of articles of straw and plaiting material DD_20 2 (retreat) 08 Manufacture of pulp, paper and paper products DE_21 3 (lost opportunities) 09 Publishing, printing and reproduction of recorded media DE_22 3 (lost opportunities) 10 Manufacture of coke, refined petroleum products and nuclear fuel DF_23 3 (lost opportunities) 11 Manufacture of chemicals and chemical products (except Manufacture of pharmaceuticals, medicinal chemicals and botanical products) DG_24 except 24.4 3 (lost opportunities) 12 Manufacture of pharmaceuticals, medicinal chemicals and botanical products DG_24.4 3 (lost opportunities) 13 Manufacture of rubber and plastic products DH_25 3 (lost opportunities) 14 Manufacture of ceramic tiles and flags DI_26.3 2 (retreat) 15 Manufacture of other non-metallic mineral products (except Manufacture of ceramic tiles and flags) DI_26 except 26.3 2 (retreat) 16 Manufacture of basic iron and steel and of ferrous alloys and ferrous products DJ_27.1, 27.2, 27.3, 27.51, 27.52 1 (dynamic specialisation) 17 Manufacture of basic precious and nonferrous metals and non-ferrous products DJ_27.4, 27.53, 27.54 1 (dynamic specialisation) 18 Manufacture of fabricated metal products (except machinery and equipment) DJ_28 4 (stationary specialisation) 19 Manufacture of machinery and equipment n.e.c DK_29 1 (dynamic specialisation) 20 Manufacture of office machinery and computers DL_30 4 (stationary specialisation) 21 Manufacture of electrical machinery and apparatus n.e.c. DL_31 4 (stationary specialisation) 22 Manufacture of electronic valves and tubes and other electronic components DL_32.1 4 (stationary specialisation) 23 Manufacture of radio, television and communication equipment and apparatus DL_32 except 321 4 (stationary specialisation) 24 Manufacture of medical, precision and optical instruments, watches and clocks DL_33 2 (retreat) 25 Manufacture of motor vehicles, trailers and semi-trailers DM_34 4 (stationary specialisation) 26 Building and repairing of ships and boats DM_35.1 4 (stationary specialisation) 27 Manufacture of aircraft and spacecraft DM_35.3 4 (stationary specia- 25 lisation) 28 Manufacture of other transport equipment DM_35, except 35.1, 35.3 4 (stationary specialisation) 29 Manufacture of furniture DN_36.1 4 (stationary specialisation) 30 Manufacture of games and toys DN_36.5 4 (stationary specialisation) 31 Manufacture of games and toys DN_36, except 36.1,36.5 4 (stationary specialisation) Source: Own elaboration from PITEC and OECD patent d