counting USPTO patent applications

[avival]

In a simple count of USPTO patent applications for 2010 using PATSTAT, I’m getting a very different answer than USPTO data provides.

I’m a new user so I’m going to lay out what I did and hope that someone can help me discover my error. I’m using the October 2012 snapshot file but I don’t expect things should have changed too much in the updated file.

1. I created a single file from all three tls201 files.
2. I counted all entries with jointly appln_year=2010, appln_kind= “A “ and appln_auth="US". There are a small number of appln_kind D, but it’s trivial. I would expect appln_kind A would match USPTO’s “Utility Patent Applications” and the design and plant patent applications (appln_kind F and P, repectively?) would match USPTO’s design and plant applications.
3. I am not including international filings through PCT (appln_kind=”W ”) in these counts as it is my understanding that they should only be counted when they reach the national level and should then appear as appln_kind “A” with an internat_appln_id linking back to the PCT application.

I’m counting 285,913 patent applications, much lower than the USPTO count of 490,226 for “Utility Patent Applications” (sorry, apparently I can't include a link).

My appln_kind=”P “ count is also much higher than USPTO’s Plant Patent Applications (74,804 vs. 992), but this discrepancy isn’t high enough to account for the discrepancy.

I’m getting reasonable counts for PCT applications (i.e. they match those from WIPO) using appln_kind=”W “ but am having similar issues with EPO counts.

Hope someone can help and thanks in advance!

[nico.rasters]

My first guess was the coverage but Contents and coverage of the DOCDB bibliographic file.xls but it does not list any known gaps for US. I'm using PATSTAT October 2011 and the "To date" in the coverage file is 9/20/2011 so that's quite up to date.

My query on TLS201_APPLN:
SELECT `APPLN_KIND` , COUNT( * )
FROM `TLS201_APPLN`
WHERE YEAR( `APPLN_FILING_DATE` ) =2010
AND `APPLN_AUTH` = "US"
GROUP BY `APPLN_KIND`;

APPLN_KIND; COUNT
23; 1
A; 168280
D; 6
F; 12436
P; 8837
W; 41654

So for the year 2010 you counted 285,913 patents in the October 2012 version.
I counted 168,280 patents in the October 2011 version. That's quite a difference.

I also have PATSTAT April 2009.

SELECT YEAR( `APPLN_FILING_DATE` ) , COUNT( * )
FROM `TLS201_APPLN`
WHERE YEAR( `APPLN_FILING_DATE` ) >=2000
AND `APPLN_AUTH` = "US"
AND `APPLN_KIND` = "A"
GROUP BY YEAR( `APPLN_FILING_DATE` );

Application year | 2009 April | 2011 October
2000 | 218234 | 219566
2001 | 265205 | 265864
2002 | 270703 | 271927
2003 | 283318 | 285628
2004 | 321994 | 326776
2005 | 343116 | 357550
2006 | 299393 | 341546
2007 | 236546 | 339921
2008 | 088574 | 322499
2009 | 000001 | 271810

Till 2005 the results are quite similar. But why is there a difference for 2006 and beyond? Probably the same reason as to why your results from USPTO are higher than from PATSTAT. My guess is that the USPTO does not publish their patents right away. Apparently the lag is 3 years.

[nico.rasters]

Until recently, US patents were only published after grant. This has changed, and patent applications in the United States are now published 18 months after their filing date, unless they have been withdrawn or have been filed with a non-publication request (if the applicant declares that he will not file a related application in another country that quotes the priority of the USPTO first filing).

Source: OECD Patent Statistics Manual