Dear all, I would like to count the number of citations that a cousin patent receives (cousin patent is another external from a different firm patent that cites the same patent I cited). And I need this for year t and t+5. I will explain next:
Let`s say that firm A develop a patent (pat1) in 1990 that cites two patents (notice that firm A in 1990 could have done more than one patent), one from firm B in 1985 (pat2) and another from firm C in 1980 (pat3). Assume then that my base knowledge in 1990 (the year firm A made the citation) comes from these two patents (pat2 and pat3). I need to create a var1 that count the number of external patents that also cite pat2 and pat3 (external means excluding citations from A or B when A cites B; or from A and C when A cites C). Thus, assuming that in the same year 1990 firm D (with pat4) and E (with pat5) also cite the same patents pat2 and pat3 (cited by A), then for firm A it will be 2 external citations in the same year (1990). However, it could be also that in the same year firm D cited two times patents pat2 or pat3 (with two patents in that year, say pat41 and pat42). Then it would be 3 counts for A in 1990 (pat41, pat42, and pat5).
Thereafter, I will need the same but for 5 years after. So, in case of firm A, it will be counting the number of citations received by the patents cited in 1990 (pat2 and pat3) but this time in 1995. So, if there is another new firm F that cite one of these patents (pat2 and/or pat3) in 1995, then it will count as 1.
However, this process is cumulative. I mean, the base knowledge of firm A in 1991 will be those cited patents in 1990 (pat2 and pat3) plus the same process for 1991. Maybe in these two tables is easier to understand, the first one is what I have, while the second one is what I need.
An important idea is that while knowledge is cumulative the citations received are not. So, t0 for A in 1991 could be 0 (meaning that patents pat2, pat3 and patXX did not receive any external citation in 1991), but maybe they receive 20 citations 5 years after (t5).
I really hope you can help me with this. I was thinking in counting first the number of citations received by patents as in the command below, but then I realize that I probably will need first to build the base cumulative knowledge (something I do not know how to start with).
If something in the explanation is not clear, please let me know, and thanks in advance for your help!!!
citing_firm is the firms citing (this would be A)
cited_firm is the firm being cited (this would be B or C)
citing_appln_id is the id for the citing patent (this would be pat1)
cited_appln_if is the is for the cited patent (this would be pat2 or pat3).
Notice that what would be pat4 and pat5 is in fact other citing patents (look the first 5 observations in the dataex below).
Firm (2.5e+06) have a patent iin 2007 (16300347) that cites patent (1) from firm (2.1e+06). However, this patent (1) even though cited by others, is not cited by an external firm (different from (2.5e+06 and 2.1e+06) in 2007. So, this will be 0. Look that patent (15937040) cites patent (1) in 2003, and there i s another external citation to the same patent 1 in 2003 coming from firm (20671) with patent (15914251). That would count as 1 in 2003 for that obs. Fot the t5, it would be to count is any other patent cites (1) in 2012 for the first obs. IF any other external citation cites patent (1) in 2008 in the group and put it in the obs from 2003 (second obs)... Maybe starting by this, and then aggregate for the cumulative base knowledge is easier.
Let`s say that firm A develop a patent (pat1) in 1990 that cites two patents (notice that firm A in 1990 could have done more than one patent), one from firm B in 1985 (pat2) and another from firm C in 1980 (pat3). Assume then that my base knowledge in 1990 (the year firm A made the citation) comes from these two patents (pat2 and pat3). I need to create a var1 that count the number of external patents that also cite pat2 and pat3 (external means excluding citations from A or B when A cites B; or from A and C when A cites C). Thus, assuming that in the same year 1990 firm D (with pat4) and E (with pat5) also cite the same patents pat2 and pat3 (cited by A), then for firm A it will be 2 external citations in the same year (1990). However, it could be also that in the same year firm D cited two times patents pat2 or pat3 (with two patents in that year, say pat41 and pat42). Then it would be 3 counts for A in 1990 (pat41, pat42, and pat5).
Thereafter, I will need the same but for 5 years after. So, in case of firm A, it will be counting the number of citations received by the patents cited in 1990 (pat2 and pat3) but this time in 1995. So, if there is another new firm F that cite one of these patents (pat2 and/or pat3) in 1995, then it will count as 1.
However, this process is cumulative. I mean, the base knowledge of firm A in 1991 will be those cited patents in 1990 (pat2 and pat3) plus the same process for 1991. Maybe in these two tables is easier to understand, the first one is what I have, while the second one is what I need.
| Firm | Year | Citing Patent | Cited patent | Cousin | |
| A | 1990 | A1_1 | Pat2, Pat3 | Pat4 | |
| A | 1990 | A1_2 | Pat3 | Pat5 | |
| A | 1991 | A2_1 | |||
| A | 1992 | A3_1 | |||
| Firm | Year | Citing Patent | Knowledge | t0 | t5 |
| A | 1990 | A1_1 | pat2, pat3 | 2 | 1 |
| A | 1991 | A2_1 | pat2, pat3, patXX | ||
| A | 1992 | A3_1 | pat2, pat3, patXX, patYY | ||
An important idea is that while knowledge is cumulative the citations received are not. So, t0 for A in 1991 could be 0 (meaning that patents pat2, pat3 and patXX did not receive any external citation in 1991), but maybe they receive 20 citations 5 years after (t5).
I really hope you can help me with this. I was thinking in counting first the number of citations received by patents as in the command below, but then I realize that I probably will need first to build the base cumulative knowledge (something I do not know how to start with).
If something in the explanation is not clear, please let me know, and thanks in advance for your help!!!
Code:
sort cited_appln_id citing_appln_id egen count_t0 = count(citing_appln_id) if citing_firm_id!=. & cited_firm_id!=. & (citing_firm_id!=cited_firm_id), by(cited_appln_id citing_year)
cited_firm is the firm being cited (this would be B or C)
citing_appln_id is the id for the citing patent (this would be pat1)
cited_appln_if is the is for the cited patent (this would be pat2 or pat3).
Notice that what would be pat4 and pat5 is in fact other citing patents (look the first 5 observations in the dataex below).
Firm (2.5e+06) have a patent iin 2007 (16300347) that cites patent (1) from firm (2.1e+06). However, this patent (1) even though cited by others, is not cited by an external firm (different from (2.5e+06 and 2.1e+06) in 2007. So, this will be 0. Look that patent (15937040) cites patent (1) in 2003, and there i s another external citation to the same patent 1 in 2003 coming from firm (20671) with patent (15914251). That would count as 1 in 2003 for that obs. Fot the t5, it would be to count is any other patent cites (1) in 2012 for the first obs. IF any other external citation cites patent (1) in 2008 in the group and put it in the obs from 2003 (second obs)... Maybe starting by this, and then aggregate for the cumulative base knowledge is easier.
Code:
* Example generated by -dataex-. To install: ssc install dataex clear input long citing_firm_id float citing_year long(citing_appln_id cited_firm_id cited_appln_id) 2537543 2007 16300347 2125445 1 540830 2003 15937040 2125445 1 3449971 2006 16122608 2125445 1 20671 2003 15914251 2125445 1 1663023 2009 55416440 2125445 1 20671 2003 15914251 441693 436742 2125445 2000 1 . 13388690 2537543 2007 16300347 . 13542654 20671 2003 15914251 . 13611712 3449971 2006 16122608 . 13646192 3449971 2006 16122608 . 15104837 20671 2003 15914251 106640 15711936 1663023 2009 55416440 2601218 15842637 1663023 2009 55416440 7569 16218412 1663023 2009 55416440 2949820 16979978 2125445 2000 1 85705 16980819 2125445 2000 1 2125445 17000979 20671 2003 15914251 . 17234050 540830 2003 15937040 . 25465531 1663023 2009 55416440 . 38796668 2537543 2007 16300347 . 47318336 540830 2003 15937040 . 47884703 20671 2003 15914251 . 48138997 2537543 2007 16300347 . 48482791 20671 2003 15914251 . 48841637 2537543 2007 16300347 . 48934132 2537543 2007 16300347 . 49597833 3449971 2006 16122608 . 50456586 3449971 2006 16122608 . 52051158 2537543 2007 16300347 . 52205374 3449971 2006 16122608 . 52661224 2537543 2007 16300347 . 53412433 2537543 2007 16300347 . 53417459 3449971 2006 16122608 . 53759040 2537543 2007 16300347 . 54006334 2537543 2007 16300347 . 54198352 20671 2003 15914251 . 54477036 end
