What can patent litigation tell us
about patent examination?
Alan Marco and Richard Miller
September 2016
OFFICE OF CHIEF ECONOMIST
www.uspto.gov/economics
Patent litigation conditions
1. Valuable enough to make litigation profitable
2. Some degree of uncertainty so that there is a dispute
 Patent value is a function of validity, scope, and the
underlying technology
 Patent examination affects scope and uncertainty, but
not the underlying technology
Caution: litigated patents are highly selected, so one must
be careful generalizing the results
2
Strategy
1. Match litigated patents to “similar” non-litigated
patents
2. Estimate how examination characteristics impact the
probability of litigation
Sample:
• Patents granted 2005-2011
• Patent litigation filings in federal district court, 20052015 (source: RPX). Examine first filing only.
3
Litigated Patents by Year of Issue and
Technology Area
2500
Number of patents
2000
CHEM
1500
DES
TRANS
SEMI
1000
MECH
BIO
500
COMP
0
2005
2006
2007
2008
2009
2010
2011
Year of patent issue
4
Methods
• Control group of non-litigated patents that match the
litigated patents on several characteristics
–
–
–
–
Exact match on examination art unit & year of issue
Exact match on maintenance history
Nearest neighbor match on 3-year forward citations
Random sample within that group
• Final sample is a 1-to-1 matched sample (>10,000
litigated patents)
• Conditional logit: Is the incidence of litigation correlated
with the examination characteristics?
5
Variables included in the analyses
Application characteristics
•
•
•
•
small entity status
foreign priority & PCT
number of US parent applications
family pendency
Examination characteristics
•
•
•
•
•
•
Claims characteristics
•
• independent claim count
• independent claim length (shortest) •
examiner seniority
examination pendency
continuances (RCEs)
interviews
appealed?
first-action allowance?
applicant disclosures (IDS)
backward citations
6
Main Results
• Largest impact on litigation is from application
characteristics
• Incidence of litigation is impacted by characteristics
of examination to a much lesser extent
• Characteristics of the independent claims are
important
7
Conditional Logit Results
Odds Ratio
Point Estimate
Examiner Signatory Authority
None**
Partial*
Missing
Small Entity**
Continuation History
Number of Domestic Parents**
Pendency from Earliest Parent to Docketing
Foreign Applications
Foreign Priority Claim**
National Stage Entry**
Examination Variables
Number of IDS Filings**
Number of Recorded Interviews**
Pendency Before Examiner**
Number of RCEs**
At Least One Appeal**
First-Action Allowance**
Patent Claims Variables
Number of Independent Claims**
Length of Shortest Independent Claim**
Backward Citations
To US Patents and Applications**
To Foreign Patents or Applications*
To Non-Patent Literature**
** Coefficient is statistically significant at the 1-percent level.
* Coefficient is statistically significant at the 5-percent level.
Coefficient
95% CI
Point Estimate
Standard Error
0.821
0.843
1.176
2.474
0.763
0.722
0.948
2.312
0.884
0.984
1.459
2.648
-0.197
-0.171
0.162
0.906
0.038
0.079
0.110
0.035
1.238
1.210
1.266
0.213
0.012
1.000
0.998
1.001
0.000
0.001
0.414
0.779
0.378
0.662
0.452
0.917
-0.883
-0.249
0.046
0.083
1.043
1.034
1.052
0.042
0.004
1.215
1.162
1.270
0.195
0.023
0.898
1.130
1.973
0.853
0.874
1.075
1.515
0.774
0.924
1.187
2.570
0.939
-0.107
0.122
0.680
-0.159
0.014
0.025
0.135
0.049
1.068
1.052
1.083
0.065
0.007
0.999
0.998
0.999
-0.001
0.000
1.001
1.000
1.002
0.001
0.000
1.004
1.001
1.007
0.004
0.002
1.002
1.001
1.003
0.002
0.001
8
Application characteristics
3
Odds Ratio
2.5
Note: effects calculated for a change
from the 25th percentile to the 75th
percentile for continuous variables
2
1.5
1
0.5
0
9
Examination characteristics
3
Odds Ratio
2.5
Note: effects calculated for a change
from the 25th percentile to the 75th
percentile for continuous variables
2
1.5
1
0.5
0
10
Claim characteristics
3
Odds Ratio
2.5
Note: effects calculated for a change
from the 25th percentile to the 75th
percentile for continuous variables
2
1.5
1
0.5
0
11
Goodness of fit (ROC Curve):
Application characteristics drive the model
12