User Guide to the PharmaNet Database: Interfirm Knowledge Networks of U.S. Pharmaceutical Industry, 1976-2006 John Qi Dong University of Groningen Chia-Han Yang National Cheng Kung University June 20, 2016

1

Introduction In this database, we construct interfirm knowledge networks for the U.S. pharmaceutical firms using patent citations made by them from 1976 to 2006. In the database, each dyad of firms represents a unique knowledge tie in the interfirm knowledge network as a focal firm cited another firm’s pre-existing patent in a specific year, namely the knowledge flows from other firm to the focal firm. The database was developed based on our earlier work (Dong and Yang, 2015; 2016), in which we suggest that interfirm knowledge networks consisting of knowledge flows among firms are different from interfirm collaboration networks based on alliance partnerships. While the latter may involve some knowledge flows between a focal firm and its partners, knowledge can also flow through learning from codified documents (e.g., patents) without formal collaboration. In the innovation literature, a common practice to capture knowledge flows among technological inventions, individual inventors or firms, is using patent citations (e.g., Dong and Yang, 2015; 2016; Li, Lai, D’Amour, Doolin, Sun, Torvik, Yu and Fleming, 2014; Sorenson, Rivkin and Fleming, 2006; Wang, Rodan, Fruin and Xu, 2014; Wang, Choi, Wan and Dong, 2016). Using patent citations to measure the knowledge flows among firms have several advantages. First, patent citations have been carefully assigned and examined by intellectual property regulations. The U.S. Patent and Trademark Office (USPTO) requires all patent applications to explicitly identify and cite relevant pre-existing patents that are the basis of new inventions. USPTO also recruits patent examiners in different technological domains to review and supplement the prior references in patents to ensure that accurate and comprehensive citations are made. Thus, patent citations are probably the most reliable data source for knowledge flows available over years. Second, the innovation literature has documented that new patented inventions can be conceptualized as recombination of pre-existing technological knowledge (Fleming, 2001). Patent citations therefore provide ideal records about the prior knowledge in pre-existing patents that have been searched and used by a focal firm to develop its new patented inventions, consistent with our ontology of a firm’s knowledge flows as the external knowledge transferred from other firms to the focal firm. Of course, patented knowledge is often technological and may not cover other kinds of knowledge. However, in specific industry contexts such as the U.S. pharmaceutical industry, technological knowledge is significant as the main source of competitive advantage (Grigoriou and Rothaermel, 2016). This database is freely available on https://sites.google.com/a/rug.nl/pharmanet. It should be cited as: Dong, J. Q., and Yang, C.-H. 2016. PharmaNet database: Interfirm knowledge networks of U.S. pharmaceutical industry, 1976-2006. Working Paper, University of Groningen, Groningen. To use this database, a request email should be sent to [email protected]. In addition, a copy of any publications or working papers using this database must be sent to [email protected] for archival purposes.

2

Database Overview To construct the database, we obtained almost 24 million patent citation pairs for all the U.S. utility patents that were granted between 1976 and 2006 from the National Bureau for Economic Research (NBER), which was originally derived from USPTO (Hall, Jaff and Trajtenberg, 2001). We then used the “match” file provided by NBER to match the patent assignees in each patent citation pair to their Global Company Keys (GVKEYs) of public firms included in the Standard and Poor’s Compustat database. This “match” file is a dynamic family tree of firms over time as firms may change their GVKEYs due to mergers and acquisitions (M&As). After this matching process, we obtained basic company information from Compustat database and merged the information to our data based on firm GVKEY and year. We derived the sample from the U.S. pharmaceutical industry by excluding firms with SIC code other than 2833 to 2836 that making a patent citation to any other firms’ preexisting patents. Note that this procedure still enabled us to maintain the firms from other industries if a pharmaceutical firm cites their patents, and thereby to construct a complete interfirm knowledge network. We aggregated patent-level citation data to the assignee level and then to the firm level, by grouping the number of each unique patent citation pair by firm and year. Self-citations were excluded, as a network does not allow a node with a tie to itself. Finally, we constructed interfirm knowledge networks for the U.S. pharmaceutical industry based on the dyads of firms in a yearly base and then calculated network variables to be described later. In the Appendix, we provide the computer algorithm that we used to calculate network variables. Variables PharmaNet Database with Dyad-Year Panel Structure In the data file with a dyad-year panel structure, each pair of companies represents a unique knowledge tie in the network as one company cited another company’s patent in a specific year. This data file contains a total of 19,045 dyad-year observations from 1976 to 2006. Below is a summary of the variables in the data file with a dyad-year panel structure. GVKEY: The variable “gvkey_citing” indicates the GVKEY of a company that cited another company’s patent and the variable “gvkey_cited” indicates the GVKEY of a company with the patent that was cited. Year: The variable “year” indicates the year when citations were made in patent applications. Weight: The variable “weight” indicates the number of knowledge flows in a unique knowledge tie of two companies. Degree Centrality: The variable “degree_citing” is the degree centrality of the company indicated by “gvkey_citing” and the variable “degree_cited” is the degree centrality of the company indicated by “gvkey_cited”.

3

Weighted Degree Centrality: The variable “wdegree_citing” is the weighted degree centrality of the company indicated by “gvkey_citing” and the variable “wdegree_cited” is the weighed degree centrality of the company indicated by “gvkey_cited”. Closeness Centrality: The variable “closeness_citing” is the degree centrality of the company indicated by “gvkey_citing” and the variable “closeness_cited” is the degree centrality of the company indicated by “gvkey_cited”. Weighted Closeness Centrality: The variable “wcloseness_citing” is the weighted degree centrality of the company indicated by “gvkey_citing” and the variable “wcloseness_cited” is the weighted degree centrality of the company indicated by “gvkey_cited”. Betweenness Centrality: The variable “betweenness_citing” is the degree centrality of the company indicated by “gvkey_citing” and the variable “betweenness_cited” is the degree centrality of the company indicated by “gvkey_cited”. Weighted Betweenness Centrality: The variable “wbetweenness_citing” is the weighted degree centrality of the company indicated by “gvkey_citing” and the variable “wbetweenness_cited” is the weighted degree centrality of the company indicated by “gvkey_cited”. Eigenvector Centrality: The variable “eigenvector_citing” is the degree centrality of the company indicated by “gvkey_citing” and the variable “eigenvector_cited” is the degree centrality of the company indicated by “gvkey_cited”. Weighted Eigenvector Centrality: The variable “weigenvector_citing” is the weighted degree centrality of the company indicated by “gvkey_citing” and the variable “weigenvector_cited” is the weighted degree centrality of the company indicated by “gvkey_cited”. PharmaNet Database with Firm-Year Panel Structure For the ease of use and linking to other databases, we also provide a data file with a firm-year panel structure. Additional firm identifiers and variables are included by augmenting the database with other data from Compustat database. This data file contains a total of 4,106 firm-year observations from 1976 to 2006. Below is a summary of the variables in the data file with a firm-year panel structure. GVKEY: The variable “gvkey” indicates the GVKEY of a company. Year: The variable “year” indicates the year when citations were made in patent applications. CUSIP: The variable “cusip” indicates the Committee on Uniform Securities IP (CUSIP) of a company. CIK: The variable “cik” indicates the Central Index Key (CIK) number of a company. Ticker: The variable “ticker” indicates the ticker symbol of a company. 4

Name: The variable “name” indicates the name of a company. City: The variable “city” indicates the city of a company’s registered address. County: The variable “county” indicates the county of a company’s registered address. State: The variable “state” indicates the state of a company’s registered address. Zip Code: The variable “zipcode” indicates the zip code of a company’s registered address. SIC Code: The variable “sic” indicates the Standard Industry Classification (SIC) code of a company. Assets: The variable “assets” indicates the total assets of a company. Number of Employees: The variable “employees” indicates the total number of employees in a company. Sales: The variable “sales” indicates the total sales of a company. Income: The variable “income” indicates the operating income before depreciation of a company. R&D Expenditure: The variable “rd” indicates the R&D expenditure of a company. Degree Centrality: The variable “degree” is the degree centrality of a company. Weighted Degree Centrality: The variable “wdegree” is the weighted degree centrality of a company. Closeness Centrality: The variable “closeness” is the degree centrality of a company. Weighted Closeness Centrality: The variable “wcloseness” is the weighted degree centrality of a company. Betweenness Centrality: The variable “betweenness” is the degree centrality of a company. Weighted Betweenness Centrality: The variable “wbetweenness” is the weighted degree centrality of a company. Eigenvector Centrality: The variable “eigenvector” is the degree centrality of a company. Weighted Eigenvector Centrality: The variable “weigenvector” is the weighted degree centrality of a company.

5

Discussion PharmaNet database is different from NBER’s original patent data as the former provides firm-level knowledge network data while the latter only contains patent-level information. It is also different from other patent databases such as the U.S. Patent Inventor database about coauthorship networks at the individual inventor level (Li et al., 2014). In the next step, we will update the database by including more recent data after 2006. It is also possible to include more network variables in the database upon the request of database users. To calculate additional network variables in need, users can import the database with dyad-year panel structure to network analysis software (e.g., R with statnet, Stata with netsis or UCINET), and use “gvkey_citing” and “gvkey_cited” as dyad and “weight” as the weight of dyad. It is important to note that network analysis should be carried out for each year by conducting subsample analysis. Should you have any comments and suggestions on the database, please send an email to [email protected].

6

Appendix We use Stata 14.1 with netsis to calculate degree cnetrality, closeness centrality, betweenness centrality and eigenvector centrality based on patent citation pairs and their weights. Below is the computer algorithm that we use: . generate degree_citing = . . generate degree_cited = . . generate closeness_citing = . . generate closeness_cited = . . generate betweenness_citing = . . generate betweenness_cited = . . generate eigenvector_citing = . . generate eigenvector_cited = . . netsis gvkey_citing gvkey_cited if year == 1976, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1976, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1976, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1976, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1976 . replace degree_cited = degree_target if year == 1976 . replace closeness_citing = closeness_source if year == 1976 . replace closeness_cited = closeness_target if year == 1976 . replace betweenness_citing = betweenness_source if year == 1976 . replace betweenness_cited = betweenness_target if year == 1976 . replace eigenvector_citing = eigenvector_source if year == 1976 . replace eigenvector_cited = eigenvector_target if year == 1976 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1977, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1977, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1977, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1977, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1977 . replace degree_cited = degree_target if year == 1977

7

. replace closeness_citing = closeness_source if year == 1977 . replace closeness_cited = closeness_target if year == 1977 . replace betweenness_citing = betweenness_source if year == 1977 . replace betweenness_cited = betweenness_target if year == 1977 . replace eigenvector_citing = eigenvector_source if year == 1977 . replace eigenvector_cited = eigenvector_target if year == 1977 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1978, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1978, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1978, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1978, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1978 . replace degree_cited = degree_target if year == 1978 . replace closeness_citing = closeness_source if year == 1978 . replace closeness_cited = closeness_target if year == 1978 . replace betweenness_citing = betweenness_source if year == 1978 . replace betweenness_cited = betweenness_target if year == 1978 . replace eigenvector_citing = eigenvector_source if year == 1978 . replace eigenvector_cited = eigenvector_target if year == 1978 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1979, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1979, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1979, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1979, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1979 . replace degree_cited = degree_target if year == 1979 . replace closeness_citing = closeness_source if year == 1979 . replace closeness_cited = closeness_target if year == 1979 . replace betweenness_citing = betweenness_source if year == 1979 . replace betweenness_cited = betweenness_target if year == 1979 . replace eigenvector_citing = eigenvector_source if year == 1979 . replace eigenvector_cited = eigenvector_target if year == 1979

8

. drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1980, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1980, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1980, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1980, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1980 . replace degree_cited = degree_target if year == 1980 . replace closeness_citing = closeness_source if year == 1980 . replace closeness_cited = closeness_target if year == 1980 . replace betweenness_citing = betweenness_source if year == 1980 . replace betweenness_cited = betweenness_target if year == 1980 . replace eigenvector_citing = eigenvector_source if year == 1980 . replace eigenvector_cited = eigenvector_target if year == 1980 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1981, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1981, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1981, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1981, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1981 . replace degree_cited = degree_target if year == 1981 . replace closeness_citing = closeness_source if year == 1981 . replace closeness_cited = closeness_target if year == 1981 . replace betweenness_citing = betweenness_source if year == 1981 . replace betweenness_cited = betweenness_target if year == 1981 . replace eigenvector_citing = eigenvector_source if year == 1981 . replace eigenvector_cited = eigenvector_target if year == 1981 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1982, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum)

9

. netsis gvkey_citing gvkey_cited if year == 1982, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1982, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1982, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1982 . replace degree_cited = degree_target if year == 1982 . replace closeness_citing = closeness_source if year == 1982 . replace closeness_cited = closeness_target if year == 1982 . replace betweenness_citing = betweenness_source if year == 1982 . replace betweenness_cited = betweenness_target if year == 1982 . replace eigenvector_citing = eigenvector_source if year == 1982 . replace eigenvector_cited = eigenvector_target if year == 1982 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1983, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1983, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1983, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1983, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1983 . replace degree_cited = degree_target if year == 1983 . replace closeness_citing = closeness_source if year == 1983 . replace closeness_cited = closeness_target if year == 1983 . replace betweenness_citing = betweenness_source if year == 1983 . replace betweenness_cited = betweenness_target if year == 1983 . replace eigenvector_citing = eigenvector_source if year == 1983 . replace eigenvector_cited = eigenvector_target if year == 1983 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1984, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1984, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1984, measure(betweenness) name(B,replace) weight(weight)

10

. netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1984, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1984 . replace degree_cited = degree_target if year == 1984 . replace closeness_citing = closeness_source if year == 1984 . replace closeness_cited = closeness_target if year == 1984 . replace betweenness_citing = betweenness_source if year == 1984 . replace betweenness_cited = betweenness_target if year == 1984 . replace eigenvector_citing = eigenvector_source if year == 1984 . replace eigenvector_cited = eigenvector_target if year == 1984 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1985, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1985, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1985, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1985, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1985 . replace degree_cited = degree_target if year == 1985 . replace closeness_citing = closeness_source if year == 1985 . replace closeness_cited = closeness_target if year == 1985 . replace betweenness_citing = betweenness_source if year == 1985 . replace betweenness_cited = betweenness_target if year == 1985 . replace eigenvector_citing = eigenvector_source if year == 1985 . replace eigenvector_cited = eigenvector_target if year == 1985 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1986, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1986, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1986, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1986, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1986

11

. replace degree_cited = degree_target if year == 1986 . replace closeness_citing = closeness_source if year == 1986 . replace closeness_cited = closeness_target if year == 1986 . replace betweenness_citing = betweenness_source if year == 1986 . replace betweenness_cited = betweenness_target if year == 1986 . replace eigenvector_citing = eigenvector_source if year == 1986 . replace eigenvector_cited = eigenvector_target if year == 1986 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1987, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1987, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1987, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1987, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1987 . replace degree_cited = degree_target if year == 1987 . replace closeness_citing = closeness_source if year == 1987 . replace closeness_cited = closeness_target if year == 1987 . replace betweenness_citing = betweenness_source if year == 1987 . replace betweenness_cited = betweenness_target if year == 1987 . replace eigenvector_citing = eigenvector_source if year == 1987 . replace eigenvector_cited = eigenvector_target if year == 1987 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1988, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1988, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1988, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1988, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1988 . replace degree_cited = degree_target if year == 1988 . replace closeness_citing = closeness_source if year == 1988 . replace closeness_cited = closeness_target if year == 1988 . replace betweenness_citing = betweenness_source if year == 1988 . replace betweenness_cited = betweenness_target if year == 1988 . replace eigenvector_citing = eigenvector_source if year == 1988

12

. replace eigenvector_cited = eigenvector_target if year == 1988 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1989, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1989, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1989, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1989, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1989 . replace degree_cited = degree_target if year == 1989 . replace closeness_citing = closeness_source if year == 1989 . replace closeness_cited = closeness_target if year == 1989 . replace betweenness_citing = betweenness_source if year == 1989 . replace betweenness_cited = betweenness_target if year == 1989 . replace eigenvector_citing = eigenvector_source if year == 1989 . replace eigenvector_cited = eigenvector_target if year == 1989 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1990, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1990, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1990, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1990, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1990 . replace degree_cited = degree_target if year == 1990 . replace closeness_citing = closeness_source if year == 1990 . replace closeness_cited = closeness_target if year == 1990 . replace betweenness_citing = betweenness_source if year == 1990 . replace betweenness_cited = betweenness_target if year == 1990 . replace eigenvector_citing = eigenvector_source if year == 1990 . replace eigenvector_cited = eigenvector_target if year == 1990 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1991, measure(adjacency) name(A,replace) weight(weight)

13

. netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1991, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1991, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1991, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1991 . replace degree_cited = degree_target if year == 1991 . replace closeness_citing = closeness_source if year == 1991 . replace closeness_cited = closeness_target if year == 1991 . replace betweenness_citing = betweenness_source if year == 1991 . replace betweenness_cited = betweenness_target if year == 1991 . replace eigenvector_citing = eigenvector_source if year == 1991 . replace eigenvector_cited = eigenvector_target if year == 1991 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1992, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1992, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1992, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1992, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1992 . replace degree_cited = degree_target if year == 1992 . replace closeness_citing = closeness_source if year == 1992 . replace closeness_cited = closeness_target if year == 1992 . replace betweenness_citing = betweenness_source if year == 1992 . replace betweenness_cited = betweenness_target if year == 1992 . replace eigenvector_citing = eigenvector_source if year == 1992 . replace eigenvector_cited = eigenvector_target if year == 1992 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1993, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1993, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum)

14

. netsis gvkey_citing gvkey_cited if year == 1993, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1993, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1993 . replace degree_cited = degree_target if year == 1993 . replace closeness_citing = closeness_source if year == 1993 . replace closeness_cited = closeness_target if year == 1993 . replace betweenness_citing = betweenness_source if year == 1993 . replace betweenness_cited = betweenness_target if year == 1993 . replace eigenvector_citing = eigenvector_source if year == 1993 . replace eigenvector_cited = eigenvector_target if year == 1993 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1994, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1994, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1994, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1994, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1994 . replace degree_cited = degree_target if year == 1994 . replace closeness_citing = closeness_source if year == 1994 . replace closeness_cited = closeness_target if year == 1994 . replace betweenness_citing = betweenness_source if year == 1994 . replace betweenness_cited = betweenness_target if year == 1994 . replace eigenvector_citing = eigenvector_source if year == 1994 . replace eigenvector_cited = eigenvector_target if year == 1994 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1995, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1995, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1995, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1995, measure(eigenvector) name(E,replace) weight(weight)

15

. netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1995 . replace degree_cited = degree_target if year == 1995 . replace closeness_citing = closeness_source if year == 1995 . replace closeness_cited = closeness_target if year == 1995 . replace betweenness_citing = betweenness_source if year == 1995 . replace betweenness_cited = betweenness_target if year == 1995 . replace eigenvector_citing = eigenvector_source if year == 1995 . replace eigenvector_cited = eigenvector_target if year == 1995 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1996, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1996, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1996, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1996, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1996 . replace degree_cited = degree_target if year == 1996 . replace closeness_citing = closeness_source if year == 1996 . replace closeness_cited = closeness_target if year == 1996 . replace betweenness_citing = betweenness_source if year == 1996 . replace betweenness_cited = betweenness_target if year == 1996 . replace eigenvector_citing = eigenvector_source if year == 1996 . replace eigenvector_cited = eigenvector_target if year == 1996 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1997, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1997, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1997, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1997, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1997 . replace degree_cited = degree_target if year == 1997 . replace closeness_citing = closeness_source if year == 1997 . replace closeness_cited = closeness_target if year == 1997 . replace betweenness_citing = betweenness_source if year == 1997

16

. replace betweenness_cited = betweenness_target if year == 1997 . replace eigenvector_citing = eigenvector_source if year == 1997 . replace eigenvector_cited = eigenvector_target if year == 1997 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1998, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1998, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1998, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1998, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1998 . replace degree_cited = degree_target if year == 1998 . replace closeness_citing = closeness_source if year == 1998 . replace closeness_cited = closeness_target if year == 1998 . replace betweenness_citing = betweenness_source if year == 1998 . replace betweenness_cited = betweenness_target if year == 1998 . replace eigenvector_citing = eigenvector_source if year == 1998 . replace eigenvector_cited = eigenvector_target if year == 1998 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 1999, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1999, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1999, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 1999, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 1999 . replace degree_cited = degree_target if year == 1999 . replace closeness_citing = closeness_source if year == 1999 . replace closeness_cited = closeness_target if year == 1999 . replace betweenness_citing = betweenness_source if year == 1999 . replace betweenness_cited = betweenness_target if year == 1999 . replace eigenvector_citing = eigenvector_source if year == 1999 . replace eigenvector_cited = eigenvector_target if year == 1999 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target

17

. netsis gvkey_citing gvkey_cited if year == 2000, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2000, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2000, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2000, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 2000 . replace degree_cited = degree_target if year == 2000 . replace closeness_citing = closeness_source if year == 2000 . replace closeness_cited = closeness_target if year == 2000 . replace betweenness_citing = betweenness_source if year == 2000 . replace betweenness_cited = betweenness_target if year == 2000 . replace eigenvector_citing = eigenvector_source if year == 2000 . replace eigenvector_cited = eigenvector_target if year == 2000 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 2001, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2001, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2001, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2001, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 2001 . replace degree_cited = degree_target if year == 2001 . replace closeness_citing = closeness_source if year == 2001 . replace closeness_cited = closeness_target if year == 2001 . replace betweenness_citing = betweenness_source if year == 2001 . replace betweenness_cited = betweenness_target if year == 2001 . replace eigenvector_citing = eigenvector_source if year == 2001 . replace eigenvector_cited = eigenvector_target if year == 2001 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 2002, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2002, measure(distance) name(D,replace) weight(weight)

18

. netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2002, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2002, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 2002 . replace degree_cited = degree_target if year == 2002 . replace closeness_citing = closeness_source if year == 2002 . replace closeness_cited = closeness_target if year == 2002 . replace betweenness_citing = betweenness_source if year == 2002 . replace betweenness_cited = betweenness_target if year == 2002 . replace eigenvector_citing = eigenvector_source if year == 2002 . replace eigenvector_cited = eigenvector_target if year == 2002 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 2003, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2003, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2003, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2003, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 2003 . replace degree_cited = degree_target if year == 2003 . replace closeness_citing = closeness_source if year == 2003 . replace closeness_cited = closeness_target if year == 2003 . replace betweenness_citing = betweenness_source if year == 2003 . replace betweenness_cited = betweenness_target if year == 2003 . replace eigenvector_citing = eigenvector_source if year == 2003 . replace eigenvector_cited = eigenvector_target if year == 2003 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 2004, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2004, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2004, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum)

19

. netsis gvkey_citing gvkey_cited if year == 2004, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 2004 . replace degree_cited = degree_target if year == 2004 . replace closeness_citing = closeness_source if year == 2004 . replace closeness_cited = closeness_target if year == 2004 . replace betweenness_citing = betweenness_source if year == 2004 . replace betweenness_cited = betweenness_target if year == 2004 . replace eigenvector_citing = eigenvector_source if year == 2004 . replace eigenvector_cited = eigenvector_target if year == 2004 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 2005, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2005, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2005, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2005, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 2005 . replace degree_cited = degree_target if year == 2005 . replace closeness_citing = closeness_source if year == 2005 . replace closeness_cited = closeness_target if year == 2005 . replace betweenness_citing = betweenness_source if year == 2005 . replace betweenness_cited = betweenness_target if year == 2005 . replace eigenvector_citing = eigenvector_source if year == 2005 . replace eigenvector_cited = eigenvector_target if year == 2005 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target . netsis gvkey_citing gvkey_cited if year == 2006, measure(adjacency) name(A,replace) weight(weight) . netsummarize A/(rows(A)-1), generate(degree) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2006, measure(distance) name(D,replace) weight(weight) . netsummarize (rows(D)-1):/rowsum(D), generate(closeness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2006, measure(betweenness) name(B,replace) weight(weight) . netsummarize B/((rows(B)-1)*(rows(B)-2)), generate(betweenness) statistic(rowsum) . netsis gvkey_citing gvkey_cited if year == 2006, measure(eigenvector) name(E,replace) weight(weight) . netsummarize E, generate(eigenvector) statistic(rowsum) . replace degree_citing = degree_source if year == 2006 . replace degree_cited = degree_target if year == 2006 . replace closeness_citing = closeness_source if year == 2006

20

. replace closeness_cited = closeness_target if year == 2006 . replace betweenness_citing = betweenness_source if year == 2006 . replace betweenness_cited = betweenness_target if year == 2006 . replace eigenvector_citing = eigenvector_source if year == 2006 . replace eigenvector_cited = eigenvector_target if year == 2006 . drop degree_source degree_target closeness_source closeness_target betweenness_source betweenness_target eigenvector_source eigenvector_target

21

Acknowledgements We thank Gloria Barczak, Wilfred Dolfsma, Dries Faems, Devi Gnyawali, Martin Kenney, Chang Won Lee and Jeffrey Reuer for their encouragement and comments. We are also grateful to the general support that we received from the Hong Kong University of Science and Technology and the University of Groningen when developing this database. References Dong, J. Q., and Yang, C.-H. 2015. Information technology and organizational learning in knowledge alliances and networks: Evidence from U.S. pharmaceutical industry. Information and Management 52(1): 111-122. Dong, J. Q., and Yang, C.-H. 2016. Being central is a double-edged sword: Knowledge network centrality and new product development in U.S. pharmaceutical industry. Technological Forecasting and Social Change, In Press. Fleming, L. 2001. Recombinant uncertainty in technological search. Management Science 47(1): 117-132. Grigoriou, K., and Rothaermel, F. T. 2016. Organizing for knowledge generation: Internal knowledge networks and the contingent effect of external knowledge sourcing. Strategic Management Journal, forthcoming. Hall, B. H., Jaffe, A. B., and Trajtenberg, M. 2001. The NBER patent citation data file: Lessons, insights and methodological tools. NBER Working Paper 8498, Cambridge, MA: National Bureau of Economic Research. Li, G.-C., Lai, R., D’Amour, A., Doolin, D. M., Sun, Y., Torvik, V. I., Yu, A. Z., and Fleming, L. 2014. Disambiguation and co-authorship networks of the U.S. patent inventor database (1975-2010). Research Policy 43(6): 941-955. Sorenson, O., Rivkin, J. W., and Fleming, L. 2006. Complexity, networks and knowledge flow. Research Policy 35(7): 994-1017. Wang, C., Rodan, S., Fruin, M., and Xu, X. 2014. Knowledge networks, collaboration networks, and exploratory innovation. Academy of Management Journal 57(2): 484-514. Wang, H., Choi, J., Wan, G., and Dong, J. Q. 2016. Slack resources and the rent-generating potential of firm-specific knowledge. Journal of Management 42(2): 500-523.

22