Constructing a Chinese Patent Database of listed firms in China: Descriptions, lessons, and insights

• Author: Wenlong He,Yuchen Zhang,Tony W. Tong,Zi‐Lin He
• Date: 01 September 2018
• DOI: http://doi.org/10.1111/jems.12186
• Published date: 01 September 2018

Received: 2 June 2015 Revised: 12 August 2016 Accepted: 12 August 2016

DOI: 10.1111/jems.12186

ORIGINAL ARTICLE

Constructing a Chinese Patent Database of listed firms in China:

Descriptions, lessons, and insights

Zi-Lin He1Ton y W. Ton g 2,3 Yuchen Zhang4Wenl ong He5

1Tilburg Schoolof Economics and Man-

agement, TilburgUniversity, Tilburg, The

Netherlands (Email: z.l.he@uvt.nl)

2Krannert School of Management, Purdue

University,West Lafayette, IN, USA

3Leeds School of Business, Universityof

Colorado, Boulder, CO, USA

4Freeman School of Business, Tulane

University,New Orleans, LA, USA

(Email: yzhang54@tulane.edu)

5The Business School, Universityof Interna-

tional Business & Economics, Beijing,China

(Email: wenlong.he@uibe.edu.cn)

Correspondence

TonyW.Tong, Krannert School of Manage-

ment,Purdue University, WestLafayette, IN.

Email:tonytong@purdue.edu

Abstract

Although China is now the largest patent filing country in the world, there is lit-

tle firm-level research using Chinese patents due to difficulties in integrating Chi-

nese patent data with firm data. To partially address this gap, we construct a Chinese

Patent Database linking State Intellectual Property Office (SIPO) patents to all listed

firms and their subsidiaries in China, and we are making the database publicly avail-

able to the research community. We first develop a computer program to match the

assignee names of SIPO patents to the names of listed firms and subsidiaries based

on a similarity score, taking into account unique challenges associated with Chinese

firm-names and Chinese characters. High-scoring likely matches are then checked

manually to determine whether they are indeed true matches. The resulting database

includes 191,325 SIPO patents matched to listed firms in China from 1990 to 2010.

Using this database, we find a large amount of patenting heterogeneity across firms

of different geographic locations, technological foci, and ownership types. We also

leverage strengths of the data to conduct a detailed analysis of the patent examina-

tion process at the SIPO. Although there is not much difference in the examination

process between listed firms and their nonlisted counterparts in China, substantial

differences exist between Chinese firms and foreign firms. We find that foreign firms

experience substantial delay in publishing patent applications and requesting for sub-

stantive examination compared to Chinese firms. Such delay accounts for 40–60%

of the longer duration from application date to decision date for foreign firms. How-

ever, after accounting for suchdelay, foreign firms still face much longer pendency in

reaching an examination outcome (grant, withdrawal, or refusal) than Chinese firms.

We hope that the public database and such analysis will encourage new streams of

research on Chinese patents to improve our knowledge of China’s fast-changinginno-

vation landscape.

1INTRODUCTION

Innovation is widely seen as an important driver of economic growth. Because firms are the main engines of innovation and

technological change (Nelson & Winter,1982; Porter, 1990), it is import ant that academic scholars,policy makers, and business

We thank the State Intellectual Property Office (SIPO) of China and the National Natural Science Foundation of China (No. 71272131) for supporting this

research. Weare g rateful to seminar participants at CEIBS, Copenhagen Business School, Hong KongUniversity of Science and Technology, INSEAD, North-

western University’s Searle Center ResearchRoundtable on Patent and Technology Standard Datasets, Ohio State University, Peking University,Purdue Uni-

versity, Rensselaer PolytechnicInstitute, Shanghai University of Finance and Economics, Singapore Management University, Tsinghua University, University

of California Berkeley Innovation Seminar, and University of Hong Kong for helpful comments, as well as to Professor Bronwyn Hall and two anonymous

reviewers for their encouragement and valuablesuggestions.

J Econ Manage Strat. 2018;27:579–606. © 2017 WileyPeriodicals, Inc. 579wileyonlinelibrary.com/journal/jems

580 JOURNAL OF ECONOMICS & MANAGEMENTSTRATEGY

managers understand the firm-level underpinnings of innovation. However, innovation research has long been constrained by

the lack of firm-level data on innovative activities. Patents play a central role in this context (Griliches, 1990). Despite some

of their limitations, patents provide highly detailed information on the innovation concerned, including technical descriptions,

the assignee(s), the inventor(s), the time (application, grant, and expiry dates), the location (inventor and assignee addresses),

the technology domains to which it belongs (technology classes), the scope of property rights (claims), and so forth. Although

extremely valuable for innovation research, patent data do not come with firm-level identifiers that can be readily used to link to

other sources of data, such as firms’ financial and market data that are publicly available and systematically collected. To better

harness the power of patent data for innovation research, a number of important projects have been carried out to match patents

to the firms that own them, making it possible to link patent data to the wealth of public information already available on the

firms themselves. These include Hall, Jaffe, and Trajtenberg (2001), Bessen (2009), Balasubramanian and Sivadasan (2010),

Thoma et al. (2010), among others.

Although the resultant data sets from these matching projects have greatly expanded the breadth and depth of questions that

researchers are able to analyze using patent data, they all focused on patent offices in developed economies, notably the United

States Patent and Trademark Office (USPTO) and the European Patent Office (EPO). In this study,we develop a patent database

of Chinese listed firms based on patents filed with China’s State Intellectual Property Office (SIPO) to complement prior projects

focusing on patent offices in developed economies. In 1985, China adopted a patent law with featuressimilar to those of Europe

and Japan. China’s patent law has since undergone three amendments in 1992, 2000, and 2008, respectively (each went into

force in the following year), which progressively aligned the Chinese patent system with international norms (Park, 2008). The

SIPO grants three types of patents: invention patents, utility model patents (similar to the German Gebrauchsmuster), and design

patents. The SIPO has experienced a surge in patent applications and grants over the past 30 years, as illustrated in Figure 1.

Notably, applications of invention patents, the type of patents that is most comparable across countries, to the SIPO increased

from 14,409 in 1992 to 526,412 in 2011, overtaking the United States to become the world’stop patent filer (WIPO, 2012, p. 58).

A number of Chinese companies have emerged as world-classinnovation players during this period, including Huawei and ZTE

in information and communication technology, Haier in white goods, Lenovoin computing, BYD in electric cars and batteries,

and so forth.

Despite the patenting surge in China and the value of patents for innovation studies,fir m-level research using Chinese patent

data has been scarce, largely due to difficulties in linking such data with firm data. To meet the increasing interest in Chi-

nese firms’ innovation activities and reduce potentially duplicative matching efforts by scholars, we construct a Chinese Patent

Database that links Chinese patents directly from the SIPO to all listed firms on the Shanghai or Shenzhen Stock Exchange

(“Main Board”) and the firms’ subsidiaries. Our focus on listed firms in China follows Hall et al.’s (2001) pioneering NBER

Patent Data Project that linked USPTO patents to listed firms on the U.S. stock exchanges. A focus on listed firms allows

researchers to readily link patent data to other financial and market information available on listed firms from public sources,

thereby expanding the scope of the research questions that scholars can examine. Furthermore, we match the rawChinese patent

data provided by the SIPO that have rich information on all key patenting steps and alternative examination outcomes. Com-

pared with the Chinese patent data in Worldwide Patent Statistical Database (PATSTAT) that only give information about the

filing date and grant date (if granted) of a patent (e.g., Eberhardt, Helmers, & Yu, 2011; Liegsalz and Wagner,2013), we capture

and code, for each patent application, the date of filing, the date of publication, the date of request for substantive examination,

and the dates of grant (if granted), withdrawal (if withdrawn), or refusal (if refused). Such information allows researchers to

study the whole patenting process with greater granularity and to model competing outcomes of patent examination. Following

Hall et al. (2001), we are making the matched database publicly available to the research community via a Chinese Patent Data

Project web site (https://sites.google.com/site/sipopdb).

The rest of this paper is organized as follows. Section 2 describes the approach we developed to construct the Chinese Patent

Database by matching SIPO patents to listed firms and their subsidiaries. Section 3 assesses how our matching approach com-

pares with alternative approaches in terms of matching results. In Section 4, we use the constructed database to examine the

distribution of matched patents by geographic location, technology class, ownership type, and year. In Section 5, we exploit

features of the matched data to conduct a detailed analysis of the patent examination process for three types of firms with the

SIPO. Section 6 concludes this paper.

2MATCHING SIPO PATENTS TO LISTED FIRMS AND SUBSIDIARIES

Entity name matching is an important field of study in its own right. Scholars in many disciplines, including statistics, database

management, and machine learning, have contributed to the core question of how to efficiently and effectively match entity