Patent family data and statistics at the European Patent Office

At the European Patent Office (EPO) a comprehensive data file called PRI is maintained of patent families. The file records are based on published patent documents, indexed by the priority number of the first patent filing, with information on subsequent patenting activities for that invention in the four major economic blocs: EPC contracting states, Japan, USA and Others. It is possible to filter the data in order to highlight the most important inventions, for example by selecting Trilateral patent families that lead to patenting activity in EPC contracting states (including EPO), Japan and USA. The relationship between patent families and subsequent filings is not one-to-one. In order to compare calculated figures from the EPO data set with an alternative system of consolidated families, it is suggested that bounds may be calculable for numbers of consolidated families by taking account of the overall numbers of network links between priority forming first filings and subsequent filings. The key to this methodology is the identification of all the links between first filings and subsequent filings in a family. There is a timeliness problem caused by a considerable delay between the date of first filing and the appearance of a publication that can index a patent family. A method is described by which more up-to-date counts of families (numbers of priorities) can be made by augmenting the database with information that is available in the distinct filings databases at the patent offices. The families data set can be used to investigate the patenting behaviour by individual companies, industries, countries or economic blocs, or to study changing patterns of technology in world-wide industrial research. Some representative data are presented over a series of years that show increasing trends for the numbers of world-wide first filings, for numbers of filings flowing from one country to another, for numbers of patent families making use of the PCT system, and for the numbers of families within the EPC contracting states area that make use of the EPO. Good forecasts for numbers of patent filings at the EPO are needed for the purpose of internal resource requirements planning. An initial attempt is described to set up an econometric model for the development of subsequent filings at the EPO, based on patent families information and on concomitant variables including source country R&D stock per worker and source country GDP per capita. It may eventually be possible to generalise a successful model of this type in order to predict filings flows to and from all the major patent offices.