Industrial Property Econometrics

Grand Amphithéâtre Lumière - Lyon
May, 14 & 15 1998


Knowledge Spillovers

A. Jungmittag, K. Blind
Fraunhofer Institute for Systems and Innovation Research
Breslauer Strasse 48
76139 Karlsruhe (Deutchland)

The significance of technological activities as an essential determinant of the economic performance of industrialised economies is generally acknowledged today. It is also indisputed in the meantime that technical standards are very important for the fast diffusion of new technologies. In clear contradiction to the theoretical insight and economic relevance, however, is the consideration of the level of technology resp. the technological progress and the role of standardisation in macroeconometric production models. So when estimating production functions (e. g a Cobb-Douglas production function), technological progress is commonly approximated only by a linear time trend. This procedure reveals a series of weaknesses. On the one hand the inclusion of a time trend does not provide an explanation for technical changes, i.e the causes or sources underlying technical progress are not distinguishable. At the most, the order of magnitude of the technical progress can be estimated. On the other hand, no changes in the rate of technical progress can be identified, rather technical progress grows uniformly, as if dropping from heaven. Only a few authors have attempted to take technical progress into account by using more appropriate indicator variables. A formal record of standardisation influence in macroeconometric production functions by means of appropriate indicator variables is completely missing.

In the present study alternative sources of technical progress will be identified and approximated by means of indicator variables, wich will then be considered when estimating long-term production functions for the enterprise sector of the Federal Republic of Germany, without agriculture, forestry, and fishing and without flat rental from 1960 until 1990. We shall distinguish between technical progress wich is the result of domestic research and development activities, and the import of technological know-how through licensing agreements. The first source of technical progress will be approximated through the time lagged stock of patents at the German Patent Office (Deutsches Patentamt), the second by the real fees for licenses captured in the balance of payments of the Federal Republic of Germany. In addition, the role of standardisation will be integrated in the long-term production function in that it will be approximated by the indicator variables - the stock of effective technical standards published by semi-public standard development organisations.

For estimating the long-term production fuctions, the concept of the cointegration of time series introduced by Engle and Granger will be used. This concept allows the differentiation between actual long-term relations and merely spurious regressions if time series are trending. As in this study only the long-term relations and not the short-term dynamics between the output, the usual production factors and the indicator variables for technical progress as well as for the role of standardisation are being considered, first of all the first step of the Engle and Granger’s two-step procedure will be applied, in which existing long-term relations are identified and estimated without specifying the short-term dynamics. However, the distribution of the estimators of the cointegrating vector provided by such a static regression is generally non-normal and so inference cannot be drawn about the significance of the individual parameters by using the standard ‘t’ test. For this reason the three-step procedure, proposed by Engle and Yoo, is subsequently used to remedy this shortcoming. Their third step, added to the Engle-Granger two-step procedure, provides a correction to the parameter estimates of the first stage static regression wich makes them asymptotically equivalent to FIML and provides a set of standard errors which allows the valid calculation of standard ‘t’ tests. The superior long-term production function will then be used to at least roughly assess the effects of the technical progress approximated by the indicator variables and of the role of standardisation, approximated by the stock of technical standards registered at the database PERINORM edited by German, French and British standard development organisations, as well as the impact of the usual production factors on economic growth from 1961 until 1990.


S. Andersen, P. Hingley
European Patent Office
Erhardstrasse, 27
D-80331 MUNICH (Allemagne)

At the European Patent Office (EPO) it is considered advisable to forecast the level of future European patent applications, so that figures can be proposed for the purposes of planning and the construction of the internal budget. Since in recent years there has been a growth in popularity of the PCT route for applications at the expense of the European direct route, it is also important to predict future proportions of PCT applications. Traditionally forecasts have been carried out by modelling the annual development of filings, also taking into account information on first filings which later give rise to European filings. But we consider that it is also important to ask applicants directly what their intentions are towards future filings. The EPO has now carried out two annual surveys of filing intentions among applicants for European patents, and is currently setting up the third such survey.

The survey questionnaire is constructed in as simple a way as possible in order to elicit the required information from the sample members. It was considered important to approach a wide variety of applicants rather than just the biggest ones. On the other hand it was also a requirement to sample large applicants so that information on growth rates could be established on the basis of as many patent applications as possible. Two groups were sampled in order to achieve this. The first was a group of the largest applicants, while the other was a random sample taken on the basis of a simple random sample of applications rather than applicants. Estimation of growth rates in patent applications could be made by comparing, for each respondent, the known number of filings in the previous year with estimated future filing numbers, as reported in the questionnaires that were sent back. The construction of a suitable generalised index of patent growth requires however that details of the sampling scheme should be taken into account. Since the random sampling was done of applications rather than applicants, an appropriate method is to combine individual indices obtained from each applicant by a process of weighted averaging, with weights that take into account the probabilities of occurence of the applicants in the sample. The rationale for such indices is discussed in the paper, and also a method is suggested by which confidence limits can be given for the projections of future filings from the random sampled group.

The 1997 survey projected large increases in the levels of patent filings in the years 1997 - 1999 compared to 1996. So far the projections seem to be confirmed by actual developments. The results indicate that the European patent system will be used increasingly in the future, and suggest that recent reductions in fees by the EPO may have had the effect of increasing the proportion of applications being made by the Euro-direct route as opposed to the PCT route.


D. Rodriguez
MIT Department of Economics
1300 Clifton Road, Suite #554
GA 30322 Atlanta (USA)

In this paper, we employ a detailed panel data set representing the patenting patterns of 150 Belgian firms from 1981-96, to determine the effects of EU Technology Policy over this period. We find that up until recently, large firms in manufacturing oriented pursuits dominated patenting focusing on process patents as opposed to product patents. We also find that changes over the past seven years in EU patenting policy has influenced patenting strategies among the larger firms, and encouraged a greater degree of patenting by medium to small firms. We compare this analysis with an identical investigation of the 141 companies contained in the Georgia (USA) state Bloomberg Index over the same time period. Our comparative analysis provides stark contrasts in patenting patterns and strategies between two similarly placed groups of firms over the 1981-96 time period and suggests how differing intellectual property protection policies greatly impact firm strategies.


G. Abramo
National Research Council of Italy (CNR)
UTIBNOT - Via Tiburtina
770 00158 ROME (Italy)

This study covers CNR technological research outputs (in terms of patents) and its ability to transfer them to the production system (in terms of licenses), in a 15-year period which goes from the first actual year of operations of CNR Licensing Office, 1981, to 1995. First, a cross-section analysis will compare CNR performance and economic impact with those of the Massachussetts Institute of Technology (MIT), USA, which is reputed as one of the best-practices in the world. The outcome of the comparative analysis allows to clearly identify technology transfer the bottleneck of the Italian technology infrastructure. The negative socio-economic implications, in termsof missed opportunities for GNP and employment growth, are such to call for a micro-level analysis, second part of the study, in order to identify the main causes wich may explain the results shown in the first part. A time-series analysis is carried out to compare the main efficiency indicators of the start-up phase of CNR licensing activity with those of the steady-condition phase.

The outcome of this study is twofold. On the one side, findings provide valuable insights on the current state of the Italian technology infrastructure, and lead to recommendations of hopeful interventions at research-policy and -management levels. On the other side, a methodology to assess the socio-economic impact of public R&D institutions is provided. Its application then, may be extended to other components of the national technology infrasctructure.


S. Allegrezza, A. Guarda-Rauchs
Ministère of Economics

By creating a trademark, a firm may protect its market share by differentiating its products. This should be particularly important for small and medium sized firms which are more vulnerable to the increased competitive pressures generated by the achievement of the internal market and growing economic globalisation. This applied study seeks to explain the mechanisms underlying a firm’s decision to deposit a trademark.

While applied research is common in analysing patents and innovations, it is not frequently found in the analysis of trademarks. This study exploits data from a survey of 1,862 firms by the "Bureau Benelux des Marques". A logit regression relates a binary dependent variable (indicating whether a firm has deposited a trademark) to different categorical explanatory variables, such as R&D expenditure, exporting capacity, sector of production, country of operation and number of employees. The regression results emphasise the central role of the entrepreneur’s personal awareness of trademark protection. Not surprisingly, the findings also confirm that firms with more employees are more likely to register a trademark.


J. Plasmans, W. Pauwels
University of Antwerp
Prinsstraat 13
B-2000 ANTWERP (Belgique)

T. Thewys

In this paper strategic R&D and patent behaviour of private enterprises will be analysed under the occurrence of spillovers with an application on a panel of 22 sectors using OECD, EPO and EUROSTAT data for Belgium, Denmark, Germany, Italy, the Netherlands and the United Kingdom over the sample period 1989-1995.

D'Aspremont and Jacquemin (AJ - 1988) were the first to systematically present a strategic analysis on decisions of R&D expenditure and (homogeneous) output quantities under duopoly in a two step game. They showed that with sufficiently large (knowledge) spillovers R&D cooperation (e.g. in the form of an R&D Joint Venture - RJV) leads to more output, more R&D and more welfare.

If both domestic and foreign (knowledge) spillovers exist an international RJV will generally innovate less than competitive national RJVs (Brod and Shivakumar (1997)). De Bondt a.o. (1992) and Kamien a.o. (1992) have extended the above mentioned AJ analysis to differentiated products under output (Cournot) and price (Bertrand) competition. Kamien a.o. (1992) have shown that R&D cooperation with perfect (i.e. maximal) spillovers yields the highest benefit. Suzumura (1992) presented an extension to an output oligopoly. Considering an oligopoly on the various markets (domestic and foreign) we will derive optimal R&D and output strategies and will provide a (very) crude empirical test for various EU countries.

Since we do not possess (yet) sufficiently detailed patent-based R&D spillovers using a weighting matrix which measures the "proximity" between the innovator and the receiver of innovation (as a patent citation analysis (Jaffe a.o. (1993)), a co-classification method (Grupp (1996)), a patent matrix (Kortum and Lerner (1997)), etc.), we had to stick to a very crude measure for the absence of spillovers: the propensity to patent defined as the number of patents per million dollars of R&D expenditure.

Of all the above mentioned EU countries the lowest propensity to patent could be observed in Belgium. Germany and the Netherlands (and to a less extent) Denmark are characterized by relatively high (sectoral) propensities to patent. The following mechanism could be observed: a loglinear inverse relationship between (sectoral) R&D expenditure and the (sectoral) propensity to patent with negative elasticities in the neighbourhood of -1 so that small firms with lower R&D budgets tend to produce more patents per R&D dollar than large firms with higher R&D budgets; at the same time we found that a positive linear relationship exists between R&D expenditure and the number of patents (sometimes, as for the Dutch data, with the number of patents as causal variable for future R&D expenditure). These observations are internally consistent given the strong negative relationshp between the (sectoral) propensities to patent and (sectoral) R&D spillovers.


P.-Y. Badillo
Université d'Aix-Marseille III
Ecole de Journalisme et de la Communication de Marseille
Chemin de Saint-Martin

L’innovation est au coeur de la dynamique d’importants secteurs économiques, en particulier dans le domaine des hautes technologies et des réseaux -énergie, transport aérien, télécommunications- ; simultanément ces secteurs sont soumis à un puissant mouvement déréglementaire qui modifie profondément les processus décisionnels. Dans un article de 1989 sur les effets de la réglementation économique, P.L. Joskow et N.L. Rose déploraient que "si peu d’effort eut été consacré à la mesure des effets de la réglementation sur l’innovation et la croissance de la productivité". Depuis lors, l’impact de la réglementation sur l’innovation a été très peu exploré, notamment en termes de comparaisons internationales. Or la réglementation évolue de manière différenciée selon les pays. Il apparaît intéressant d’examiner s’il existe un lien entre le rythme différencié d’évolution de la réglementation et le rythme différencié de diffusion des innovations. Dans la communication une comparaison internationale est faite pour ce qui concerne les nouveaux services de télécommunications.

Nous cherchons à apprécier comment de nouveaux services se sont développés dans des pays différents, et s’il est possible d’établir une relation entre la réglementation et l’extension de ces services. Certes, il ne s’agit pas d’une causalité entre la réglementation et l’innovation. Il est particulièrement difficile de distinguer les effets de la réglementation d’autres effets, tels que l’impact de la recherche-développement ou la pression de la concurrence (voir le "Competition Effect Pitfall" souligné par D. Kridel, D. Sappington et D. Weisman (1996). Cependant notre approche permet de donner des éléments originaux de réflexion sur le parallélisme entre la réglementation et la diffusion de nouveaux services.

L’analyse appliquée repose sur la méthode DEA (Data Envelopment Analysis), qui est particulièrement bien adaptée pour mesurer l’efficacité de différentes entreprises ou pays dans des cas de technologies multi-facteurs et multi-outputs. Notre approche est originale puisqu’il ne s’agit pas d’une estimation classique de l’efficacité. Ainsi les outputs sont les nombres de Fax, de "pagers", et de téléphones mobiles, ce qui donne une bonne indication du niveau de développement des nouveaux services dans chaque pays ; les inputs sont constitués par le nombre de lignes téléphoniques et le chiffre d’affaires issu des services de télécommunication. En d’autres termes, nous estimons dans quelle proportion un opérateur (ou un ensemble d’opérateurs) de télécommunication disposant d’une infrastructure donnée au niveau des lignes téléphoniques et d’un certain revenu contribue au développement des nouveaux services de télécommunication.


K. Laursen
DRUID/IKE Group Private
Department of Business Studies
Aalborg University
Fibigerstraede 4 Svendsgade 5, 1th 9220, 9000 AALBORG (DK)

In previous research (Dalum, Laursen and Villumsen, 1996) it was shown that there is a general tendency for 20 OECD countries to de-specialise over the period from 1965-1992, for what concerns export specialisation. This finding is in contrast to findings made by other authors, working on technological specialisation. These authors found increasing technological specialisation from the late 1970s to the early 1980s measured as specialisation in US patents. The first aim of this paper is to investigate whether these contradictory findings are due to a 'real world' phenomenon, or whether the explanation is purely technical, by comparing the development of export specialisation to specialisation in terms of US patents, using the same methodology, and level of aggregation. The second aim is to analyse the extent to which countries and sectors display stable specialisation patterns over time, also both in terms of exports and in terms of technology. One of the contributions made by Dalum et al, was the distinction made between specialisation (or de-specialisation) in trade patterns on the one hand, and divergence (or on the contrary convergence) on the other. A specialisation process refers to a process in which specialisation intra-country becomes more dispersed (and counter-wise for de specialisation). In contrast, a divergence process refers to a process in which countries become more different in terms of specialisation in a particular sector, across countries (and counter-wise for convergence). However the estimations made by Dalum et al., were made making separate estimations for countries and sectors, respectively. In this paper, the stability characteristics of both export and technological specialisation patterns are estimated for both countries and sectors, in one single model. The patent data used in the paper is taken from the US patent office, and concerns the number of US patents held by the firms of the country in question. The export data are taken from the STAN database. When a concordance between the two sources of information is made, it is possible to get 19 comparable sectors. The analysis concerns the period between 1971-1991. The paper confirms that the OECD countries did in general de-specialise in terms of export specialisation. The evidence is less conclusive with regard to technological specialisation, as the results are mixed in the sense that just about half of the countries tend to increase in terms of the level of specialisation, while the other half tend to engage in de-specialisation. In terms of country and sectoral stability of specialisation patterns, it can be concluded that both trade specialisation and technological specialisation patterns are path-dependent in the sense that all country and sectoral patterns are correlated between seven three year intervals, within the period in question. In comparison however, trade specialisation patters are more stable than are technological specialisation patterns. Of the sectors, food, drink & tobacco; rubber and plastics; non-ferrous metals; shipbuilding; and other transport display a relative low stability across countries and time for both types of specialisation. Pharmaceuticals; non-electrical machinery, on the other hand, are stable to the extent that the hypothesis of an unchanged specialisation pattern cannot be rejected, in the case of both types of specialisation. Among the countries, Denmark; West Germany; Portugal; Spain; and United Kingdom display the highest degree of turbulence in the specialisation patterns, across sectors and time. In contrast, the specialisation pattern for countries such as Australia; Belgium; Italy; Japan; and Norway are path-dependent to the extent that the hypothesis of an unchanged specialisation pattern cannot be rejected, also both for what concerns technological and export specialisation.


Ch. Le Bas
Laboratoire d'Economie et Statistiques Appliquées
16 quai C. Bernard
69365 LYON Cedex 07

Nous exploitons les 2531 brevets déposés au Etats-Unis par les institutions publiques de recherche (IPR) françaises (y compris les centres techniques) de 1985 à 1994. Le volume des dépots atteste bien de l'existence d'un système post-moderne de recherche fondé sur une forte interactivité entre recherche fondamentale et recherche appliquée, entre laboratoires publics et firmes industrielles. Une tendance à la diminution de la production technologique est perceptible dans la seconde période (1990-94) mais elle est différenciée. Les champs technologiques de dépots donnent les images des compétences technologiques construites et accumulées par les institutions. Celles-ci possèdent les propriétés déjà identifiées pour les grands firmes (K. Pavitt, P. Patel 1996). Enfin il est montré qu'une interdépendance existe entre la distribution des brevets par champs technologiques des IPR prises globalement et celle des firmes françaises Ceci renforce les point de vue sur le système post-moderne de recherche et les systèmes d'innovation.


J. Schmidt-Szalewski
Professeur à l'Université Lyon 3
Faculté de droit
15 quai Claude Bernard - BP 0638
69239 LYON

Introduction: Objectif du règlement
Le règlement définit les catégories d'accords de transfert de technologie qui sont exemptés de sanctions afférentes à la violation du droit communautaire de la concurrence (art. 85 al. l du traité de Rome).

I. Domaine du règlement
Le règlement définit les contrats auxquels il s'applique :

- contrats de licence de brevets nationaux
- contrats de licence de savoir-faire
- contrats mixtes.

II- Contenu - du règlement

1. Clauses exemptées de sanction
Celles qui comportent des restrictions de concurrence énumérées l'art. l et 2 du règlement

2- Clauses non exemptées de sanction
Celles qui comportent des restrictions de concurrence énumérées à l'a1"l. 3 du règlement.

III- Procédure d’application du règlement

L'exemption est automatique, mais peut-être retirée par la Commission européenne dans les conditions
de l'art. 6 du règlement.