Page 1 of 1

PATSTAT application count sql results

Posted: Wed Sep 24, 2014 2:18 pm
by pstat
Dear all,
I am running the following SQL on PATSTAT Apr.2014 edition.

SELECT date_Part('year', appln_filing_date) as year, count(appln_id), appln_auth
FROM tls201_appln
WHERE appln_auth in ('US', 'JP', 'EP', 'CN')
and date_Part('year', appln_filing_date) > 1970
and date_Part('year', appln_filing_date) < 2014
and ipr_type = 'PI'
GROUP BY 1, 3
ORDER BY 1;

I would like to count yearly applications for the US, JP, EP, CN patent offices. For the years 2013 and 2012 there is an important fall in the number of applications. How could you interpret this result? Or, is there any problem with the SQL?

Thanks.

Re: PATSTAT application count sql results

Posted: Tue Nov 11, 2014 3:41 pm
by Geert Boedt
Hello,
your query is a ranking based on the application filing date. Generally spoken, there are 18 months between the date of the patent application and the publication of the application. Add to this 6 months (maximum) between the publication and the PATSTAT production date, and you will have between 18 and 24 months overall delay. (this can even be more for PCT applications).
Because of this , you will always see a dip in filings for the last 2 years there is no data yet available on the non-published applications. This is inherent to most filing routes and you can not avoid it when you use the application fling date.
Another important factor of influence is the "data coverage of PATSTAT" which is directly linked to the data delivered to the EPO by the respective national patent authorities. In your query ( appln_auth in ('US', 'JP', 'EP', 'CN'), this is not relevant because the data sets are up-to-date for these authorities.

By using the "publication date" in your query this effect can be reduced (Example query for PATSTAT on-line)

SELECT tls201_appln.publn_earliest_year as year, count(appln_id), appln_auth
FROM tls201_appln
WHERE appln_auth in ('US', 'JP', 'EP', 'CN')
and tls201_appln.appln_filing_year > 1970
and tls201_appln.appln_filing_year < 2014
and ipr_type = 'PI'
GROUP BY 1, 3
ORDER BY 1,3

But one has to keep in mind that by taking the publication date, other external parameters can influence the figures. For example: productivity of the patent office, changes in the application & granting procedures, etc...
Researchers tend to use the earliest priority date if they need the date "closest" to the moment of invention.
They take the publication date if they want to quantify effects of the patent becoming known to the public, so called "becoming part of the prior art".

Geert BOEDT
PATSTAT team