Industrial Property Econometrics

Grand Amphithéâtre Lumière - Lyon
May, 14 & 15 1998


Value of Intellectual Property

D. Harhoff, F. Narin, F.-M. Scherer, K. Vopel
Centre for European Economic Research - ZEW
PO - BOX 10 34 43
D-68161 Mannheim (Allemagne)

Through a survey, economic value estimates were obtained on 962 inventions made in the United States and Germany and on which German patent renewal fees were paid to full-term expiration in 1995. A search of subsequent U.S. and German patents yielded a count of citations to those patents. Patents renewed to full term were significantly more valuable than patents allowed to expire before their full term. The higher an invention's economic value estimate was, the more the relevant patent was subsequently cited. Results from two wide-ranging surveys, one in Germany and one in the United States, support Trajtenberg's conclusion that patents with greater economic value are more heavily cited in subsequent patents. Our evidence suggests at least a two-stage escalation of economic values and citation counts. First, patents that are renewed to full term expiration in environments such as in Germany with highly progressive annual maintenance fees are more valuable and more highly cited than run-of-the-mill patents allowed to expire before running to full term. Second, within the relatively exclusive cohort of full-term patents, citation frequency rises with economic value, although with considerable noise in the relationship. The most cited patents turn out to be very valuable indeed, with a single U.S. citation implying on average more than $1 million of economic value. These findings provide new support for research seeking to overcome the limitations of simple patent counts as a measure of innovative contribution by acquiring data on citation frequencies and the number of nations in which patent protection is sought.


M. Sanchez-Padron, V. Cano-Fernandez, E. Los Arcos, E. Perez-Moriana
Dept. Economica Applicada - Univ.
Cam. de la Hornera s/n,
La Laguna
38071 TENERIFE (Spain)

This paper attempts to contribute to a richer understanding of the different functions currently performed by the European patent system by rigorously analyzing some of the information provided in patent documents. We will accomplish our task by liberally drawing on the stocks of ideas put forward in previous works to overcome the limitations of simple patent counts as indicators of innovative output. Since the sheer numbers of innovations which are patented defy easy generalizations, we shall be primarily concerned with a set of Dutch and French EPO-granted patents which have been maintained up to their full legal term (20 years). It is our contention that the study of "old" patents will provide useful information to gauge the effects of the European patent system on the direction and pace of innovation efforts and hence contribute to the design of innovation policies. Moreover, to the extent that some statements one can read in reports on the current state of the European patent lack the necessary supportive empirical evidence, we hope to show the importance of our analysis for filling some of these gaps.

As a background for our study we shall provide information on the following two aspects:
Summary statistics describing general European patent data at large: averages of granting periods, designated states, Kaplan -Meier survivor functions and life expectancies for eight European states.
Costs of patenting in Europe. The frequency with which references to the high costs of European patenting are made in the literature as an obstacle to its wider use, contrasts with the paucity of the analysis made of said costs. We shall therefore undertake an analysis of these costs, distinguishing between fixed and variable costs (mainly renewal fees) and show the major differences existing in the case of the latter. Although one expert opinion on these fees is that "the levying of renewal fees is to be regarded as a purely fiscal measure, and corresponds in no way to the value which a viable patent represents to the proprietor; nor is in other respects related to it" (Østerborg, 1986), we think this interpretation should be qualified. In particular, we shall suggest that the great differences in renewal fees should be interpreted as evidence of the different (explicit or implicit) approaches that European states follow in their patent policies.

The information provided in (1) will be used to evaluate, in a first approximation, the particular and peculiar nature of old patents: are their features distinguishable from those of the "average" European patent? For its part, the information provided in (2) is a starting point to question the much-repeated idea that the first patentable invention frequently occurs years before the first significant commercial product. Is this idea applicable to innovations patented in such a costly system as the European?.

We will then be ready to undertake an analysis of our data base which consists of the whole set of European patents which, having designated France and Holland, have exhausted their full patent legal term. This analysis will start with the identification of major patents and is based on the assumption that the number of times a patent document is cited may be a measure of its technological significance. Hence, we begin this section by identifying major patents in our data base using a procedure in which the number of citations of such patents by other patents is counted. The patents thus (so) identified will then be subject to an in-depth examination to ascertain if the cross-citations encountered are indicative of whether the cited patent:
a- opens the way or is related to a technologically important development path: "new windows of opportunity";
b- has a wider-than-average scope? (The scope of these patents will also be subject to a separate analysis).
c- is being "protected" by a wall of the other patents that cite the main patent.

We also examine whether there exists a relationship between the number of citations and scope and if (a), (b) and (c) can be used as reliable indicators of patent values. These questions are merely indicative of the range of topics we would like to address. The reason we can not be more explicit is because the most powerful aspects of carrying out a patent screen as the one we are proposing is that we approach patents with humility -we allow patents to tell us which are their important functions rather than impose on them a priori roles derived from a theoretically established framework. Therefore, the real scope of our study will be determined as we go along and follow the most promising avenues (see the other abstract proposed by Cano Fernandez, V. et. al.) arising out of our patent screen. We will conclude by referring to the policy implications of our findings.


D. Bosworth, G. Jobome, M. Longland
Professor of Economics
Manschester School of Management
PO Box 88

This paper focuses on the ex ante risks of and the ex post distribution of returns. In doing so, it reviews the existing literature on the returns to R&D and other forms of innovatory activity from a different angle. The original literature concentrates on the marginal returns to, say, R&D, these returns are averaged across firms. While it is interesting to know whether, on balance the returns to R&D are positive, this disguises the issue of the distribution of the returns. Distributional issues concern the fact that a small proportion of companies do much better than the average and a high proportion tend to do somewhat worse. We demonstrate that, although some evidence about distribution can be found in the mainstream economies literature, much remains hidden. In order to throw more light on this issue therefore, we explore a panel data set of UK companies. The data set is relatively such, containing information about the financial accounts, as well as patents, designs and trade marks. We report evidence based upon patent R&D relationships and market valuation (Tobin q) type functions. The results indicate highly skewed levels of innovative effort as well as a highly unequal distribution of returns to innovation.


D. Harhoff, F.-M. Scherer, K. Vopel
Centre for European Economic Research - ZEW
L7,1 - D-68161 MANNHEIM (Deutchland)

Patent renewal studies reveal a lighly rightward-skewed distribution of values. Our approach elicits valuations approximating those of the patented invention. This paper focused on the full-term patents of the application year 1977 held by West German and U.S. residents. The tail of skewed distributions values account for a large fraction of the cumulative value over all observations. Several tests were conducted to pin down more precisely the nature of the high-value tail distribution. Three highly skew alternatives were evaluated by graphical and maximum likelihood techniques : the two-parameter log normal, the ne-parameter Pareto-Levy, and the three-parameter Singh-Maddala distribution. A tow-parameter log normal distribution appears to provide the best fit to our patented invention value data.

The whole text is available on:


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.


O. Cadot, S. A. Lippman
Bd de Constance

S. A. Lippman
CA 90095-1481 Los Angeles (USA)

This paper examines the relationship between the length l of the period during which inventors are legally protected from imitation and the speed of innovation. We show that this relationship is non monotone: for small values of l, the speed of innovation is an increasing function l, whereas the reverse is true for large values of l. As a result, very long protection (such as that conferred by copyright on software) may hinder technological advance.

JEL classification numbers; 030, 031
Key words : Innovation, Imitation, Patent length

(1) This research was supported in part by INSEAD and by the John E. Olin Center, UCLA.
We are grateful for helpful comments from Rabah Amir, Richard Rumelt, Bernard Sinclair-Desgagné, Michael Waldman, and participants at seminars at the University of Montreal, Laval University, the WZB, HEC, the European Economic Association and North American Econometric Society meetings, and the OECD Workshop on the Information Society. Special thanks are due to Eric Brynjolfsson, who directed our attention to the Software industry’s problem.