Industrial Organization
Econometrics
Applied Microeconomics
International Economics

Industrial Organization
Econometrics
Microeconomics

Industrial Organization and Econometrics (Oral), May 1998
Microeconomics and Macroeconomics (Written), September 1997

Essays on the Evaluation of European Patents

Professor Jean Lanjouw
Professor Steven Berry
Professor Ariel Pakes

May 2003

M.Phil., Economics, Yale University, 1999
M.A., Economics, Yale University, 1998
B.A., Economics, Nankai University, 1993

Yale University Dissertation Fellowship, 2001
Yale University Fellowship, 1997-2001

Teaching Fellow, Introductory Microeconomics, Yale University, 2001
Teaching Fellow, Econometrics, Yale University, 2000
Teaching Fellow, Intermediate Microeconomics, Yale University, 1999

Research Assistant to Professor Jean Lanjouw and Professor Ariel Pakes, Yale University, 1999-2000
   Performed econometric analysis on application and renewal of patents under the regime of
   European Patent Organization.
Research Assistant to Professor Steven Berry, Yale University, 1998
   Collected and collated data for research on the U.S. antidepressant drug industry

"A Dynamic Stochastic Analysis of International Patent Application and Renewal Processes", Yale University, 2002. Job market paper.

"Private Value of European Patents", Yale University, 2001. Job market paper.

"Market Structure of the Antidepressant Drug Industry", Yale University, work in progress.

"Patents, R&D and the Stock Market Rate of Return: Revisited ", Yale University, 1998.

"Trade and Migration", Master Thesis, Nankai University, 1996.

"The Market Economy System of the Association of the Southeast Asian Nations", Lanzhou University Press, Chapters 4 and 6, Coedited with Liu Zhongli, et al, 1994.

"Economic Development and Trade", Nankai University, 1993.

Patent data have long been used as an indicator of the output of innovative processes. While simple patent counts are publicly available and easily obtained, the private and social value of innovations protected by patents varies widely, making patent counts too noisy an indicator of innovative output in many cases. In finding good measures of the value of patents, it is essential to observe that patenting is an optimizing process. The value of holding a patent lies in the monopolistic rights that a patent grants the patentee of making, using or selling an invention, whereas to obtain and maintain patent protection, the inventor has to pay the cost of filing the patent application as well as the renewal fee. That is to say, the inventor makes a binary decision of whether to file a patent application by comparing the expected private value of the patent, which may be known only to him, and the application cost, which is publicly observable. Similarly, he decides whether to keep a patent in force by comparing its private value known to him and the renewal fee, and drops the patent when the cost of renewal exceeds the value of the patent. Therefore, observations on the countries where the patent application is filed as well as the length of time that the patent has been kept in force, together with filing and renewal costs, can reveal a lot about the private value of the patent.

I use in my paper a comprehensive patent data set collected from the European Patent Office (EPO), which not only contains information on patent applications and renewal, but also is unusually rich in detail with records of patent characteristics such as nationality of inventor, type of technology and designated countries of patent protection. I use data on the technological area that a patent falls into and the nationality of its owner to study whether systematic differences exist in the mean value of innovations protected by different groups of patents, or in other words, which technology groups or countries benefit disproportionately from the patent system. I also observe that the value of holding a patent is not the same across countries, because each country provides a different market size for the innovation, and have a different judicial and technological background that affect the scope of protection a patent receives. By decomposing the patent value into components that are common across countries and those that vary by country, both systematically and stochastically, I am able to look at the contribution of different country’s institutional regime to the private value of patents.

The first chapter of my dissertation analyzes the initial filing and the subsequent renewal decisions of European patent applicants. Under the EPO regime, a patent applicant can file a single patent application with the European Patent Office, and a unified examination process by the EPO determines the grant of a European patent. Once a European patent is granted, it has the same rights as a national patent in those European countries that have been designated for protection. In this chapter I first run a set of nonparametric tests which reveal that patent characteristics such as the nationality of inventor and the technological classification have significant influence on the application as well as renewal pattern of European patents. I then formulate a deterministic structural patent application model to analyze how different characteristics of patents result in different designation decisions of European patent applicants, and the model implications are similar to the filing patterns found in the nonparametric studies. For example, I find that patents originating from European countries other than Germany tend to designate more countries for protection than those from Germany, Japan or the U.S. Patent applicants in the "pharmaceutical and health" technology group on average file in more countries for protection than patentees in other technology groups, while the "electronic" patentees file in the fewest countries.

However, the nonparametric tests show that the "electronic" patent holders tend to pay the renewal fees and keep the patents alive for a longer life than patentees in other technology groups, while the "pharmaceutical and health" patents have the shortest life among different technology groups on average. When this renewal pattern suggests a higher average value of the electronic patents than the pharmaceuticals, the application pattern suggests the opposite. This disparity cannot be readily explained by the application model of this chapter alone, and calls for a joint examination of patent application and renewal decisions, which is performed in the second chapter of my dissertation.

While the above model makes the simplifying assumption of a deterministic depreciation rate of returns to the patent and focuses only on the application behavior of the inventor, the second chapter builds in stochastic learning of the value of patents and combines the study of filing and renewal decisions. Two models are formulated and estimated in this chapter. The first is a stochastic patent renewal model in which the representative patentee is assumed to receive updated information about the value of his patent through a stochastic learning process, and make decisions of whether to pay the renewal fee and keep the patent alive period by period through the patent’s life. I also assume that the patentee is aware of the possibility of the patent being infringed and the subsequent costs in patent litigation. The model is estimated upon the renewal data on patents that have been filed with the EPO and designated Germany, France or the United Kingdom for patent protection. The estimation and simulation results reveal that the European patents have substantially higher private value on average compared with national patents analyzed in previous literature.

On the basis of the above stochastic renewal model I construct the second model in this chapter which jointly examines the patent application and renewal behavior. The utilization of both the cross-sectional (multi-country filing) and the time-series (patent renewal) dimensions of European patent data in this model allows one to distinguish more aspects of patent value. The European patents in two industries — pharmaceutical and electronic — are examined. The estimation shows that pharmaceutical patents on average are endowed with higher initial values, and the patent holders seek for protection in more countries than the electronic patent holders. However, pharmaceutical patents depreciate faster than electronic patents, and consequently have lower renewal rates and shorter patent lives.