Assessing the Values of Global Patents

This study seeks to assess the value of global patent in term of probability of occurrence of patent litigation in each country. To obtain the purposes, two following research steps are adopted. The first step is litigation probability modeling- a holistic-scale analysis on both 31,992 litigated patents and 3,878,852 non-litigated patents issued between 1976 and 2010 by USPTO are conducted to understand the differences between the two types of patents in terms of different variables. It is found that there are statistically significant differences for the two types of patents in the following 11 variables: 1) Number of Assignee, 2) Number of Assignee Country, 3) Number of Inventor, 4) Inventor Country, 5) Number of Patent Reference, 6) Number of Patent Citation Received, 7) Number of IPC, 8) Number of UPC, 9) Number of Claim, 10) Number of Non-Patent Reference, and 11) Number of Foreign Reference. Subsequently, logistic regression is used for predicting the probability of occurrence of a patent litigation by fitting the 11 characteristics of 3,691,225 USPTO patents to a logistic function curve. The following Model of predicting litigation probability is obtained: Probability of Litigation = e(Z)/(e(Z)+1), z = -4.9311-0.3013*(No. of Assignee)-0.0760*(No. of Inventor)+0.1045*(No. of Inventor Country)+0.0031*(No. of patent Reference)+0.0142*(No. of Patent Citation Received)+0.0263*(No. of IPC)+0.0072*(No. of UPC)+0.0148*(No. of Claim)+0.0021*(No. of Non-Patent Reference) -0.0061*(No. of Foreign Patent). The second step is applying the obtained litigation probability model to patents owned by different countries to quantitative calculate averaged probability of litigation for each patent in different countries. The contributions of this study are not only providing the model of predicting litigation probability, but also providing a novel way of patent valuation as well as facilitating global assessment on patent valuation.