Our Divided Patent System John R. Allison University of Texas McCombs School of Business Mark A. Lemley Stanford Law School David L. Schwartz Northwestern University Law School

Empirical Studies of Patent Litigation Only cases that reach ruling on dispositive motion or trial John R. Allison, Mark A. Lemley & David L. Schwartz, Understanding the Realities of Modern Patent Litigation, 92 Texas L. Rev (2014) (available at John R. Allison, Mark A. Lemley & David L. Schwartz, Our Divided Patent System?, 82 U. Chi. L. Rev. (forthcoming 2015) (available at

Background Allison & Lemley studied patent validity in 1998 We update that study (now based on cases over 20 years old) We also extend it – Include infringement and enforceability as well as validity – Include all district court and appellate decisions, not just reported decisions

Our study All patent cases filed in in which there was a decision on the merits, whether SJ, trial, or appeal and whether grant or deny Lemley and Schwartz hand-coded outcomes; Allison hand- coded patents Each decision on a patent is the unit of observation 949 observations—that is, merits decisions on each patent

Our study 2 Coded for 30 different dependent variables, including various grounds of validity, infringement, and unenforceability as well as the procedural posture of the ruling, technology, industry, etc.

Our Independent Variables Foreign Origin of Invention-Residences of majority of inventors, assignee domicile as a tie breaker Adjusted Number of Citations Received Total Prior Art References Number of Inventors Time length of litigation from filing to termination Age of Patent at Current Litigation Filing in Years Number of Defendants Number of Asserted Patents Reissue Patent? (not yet used) Federal Districts--Top 13 & All others Primary Technology Areas and Industry Areas One or More Secondary Technology Areas Declaratory Judgment

Technology areas TechnologyFrequencyPercentage Mechanical % Electrical % Chemistry % Biotechnology505.3% Software % Optics373.9% Total % Patent Decisions by Technology

Industry categories IndustryFreq.Percent Computer and Other Electronics % Semiconductor293.1% Pharmaceutical % Medical Devices, Methods and Other Medical9910.5% Biotechnology303.2% Communications % Transportation (Including Automotive)434.6% Construction323.4% Energy212.2% Goods and Services for Consumer Uses % Goods and Services for Industrial and Business Uses % Total %

Litigated patents are likely different from all patents Patents with rulings on the merits aren’t necessarily representative of all litigated patents – Less than 10% of cases reach merits rulings

Statistics Summary judgment Trials Overall definitive winners

Summary judgment of invalidity

SJ of validity and invalidity

SJ of infringement and inequitable conduct

Trial outcomes

Outcomes

Invalidity results overall

Regressions We put definitive wins and summary judgments into a series of regression models – Notable results: Citation counts aren’t significant E.D. Texas, Delaware and S.D.N.Y. correlate with higher patentee success C.D. Cal. Correlates with lower patentee success DJ plaintiffs prevail more than other accused infringers, especially on invalidity – Note that these are after factoring in all other differences in the cases

Interesting Findings from multiple regressions 1 Definitive patent owner win rate—significant predictors of patentee win –Foreign origin of invention: p <.001 –Number of asserted patents per case: p <.001 SJ of invalidity—all grounds—Significant predictors – Foreign origin of invention: p <.001 Negative (i.e., SJ of invalidity much less likely) –Age of patent at this litigation filing: p <.01 SJ of Invalidity—sec. 112 Inadequate disclosure –Age of patent at this litigation filing: p <.05 No significant predictors of SJ’s of non-infringement

Distribution of Technologies

Definitive Win Rates by Technology

Invalidity Rates by Technology

Infringement Rate by Technology

Top row = Coefficient; * = p<.10, ** = p<.05, *** = p<.01; Bottom row = Std. error Patent Owner Definitive Winner Foreign Origin of Patent0.601** (0.0230) Adjusted Number of Citations Received (0.270) Total Prior Art References * (0.0532) Number of Claims * (0.0563) Age of Patent at Current Litigation Filing (0.202) Number of Defendants (0.221) Number of Asserted Patents (0.826) TX ED1.336*** ( ) DE D0.144 (0.690) CA ND (0.922)

Mechanical (Primary)-0.863*** ( ) Electrical (Primary)-0.851** (0.0239) Biotechnology (Primary)-3.444*** (4.99e-05) Software BM (Subset of Primary)-2.307*** ( ) Software NBM (Subset of Primary)-2.176*** (2.20e-09) Optics (Primary)-1.490** (0.0156) Comparison Dummy = Chemistry F-Test for joint technology effects53.34*** ( ) Observations616

Distribution of Industries

Definitive Win Rates by Industry

Invalidity by Industry

Infringement by Industry

Top row = Coefficient; * = p<.10, ** = p<.05, *** = p<.01; Bottom row = p- value Patent Owner Definitive Winner Foreign Origin of Patent 0.621** (0.0295) Adjusted Number of Citations Received (0.551) Total Prior Art References * (0.0820) Number of Claims (0.130) Age of Patent at Current Litigation Filing (0.432) Number of Defendants (0.165) Number of Asserted Patents

TX ED1.473*** (3.14e-07) DE D0.241 (0.440) CA ND (0.727) Computer and Other Electronics (0.858) Semiconductor1.157 (0.111) Pharmaceutical1.755*** ( ) Medical Devices, Methods, and Other Medical0.934* (0.0685) Biotechnology (industry) (0.760) Communication (0.499) Transportation (Including Automotive)1.439** (0.0117) Construction0.433 (0.578) Energy1.289** (0.0235) Goods and Services for Industrial and Business Uses0.369 (0.421) Comparison Dummy = Consumer Goods and Services F-Test for joint industry effects41.03*** (1.12e-05) Observations632

Potential Implications If our results are representative of all litigated patents (or all patents) – Fits the traditional narrative Pharma patents – Appear strong – Industry needs strong patents Software patents – Appear weak – Industry doesn’t need strong patents – But biotech patents? Appear weak Conventional wisdom is that industry needs strong patents