In contrast to patents, trade secrets do not provide the same wealth of data for researchers. The overriding obstacle is that, by definition, trade secrets are secret. However, researchers have found ways to gather evidence via surveys and evidence from litigation. This final section presents the empirical papers examining trade secrets.
Surveys
The interest in trade secrets research from an economic perspective was
rekindled by surveys indicating that firms use trade secrecy more than had been previously expected. The first survey addressing a company’s use of trade secrecy, uncovered in this literature review, is that of Scherer (1977.) Scherer examines the impact that compulsory licensing has on company’s strategic use of IP. He finds that “companies subjected to antitrust mandatory licensing decrees were patenting fewer of their inventions and keeping relatively more of their new technology secret.”44 He further surmises that these companies were not changing their strategy formally, but experiencing a change in corporate culture that favoured secrecy.
The most widely cited contemporary survey is that by Cohen, Nelson and Walsh (2000.) In their groundbreaking paper, the authors conduct an extensive survey in 1994 of manufacturing firms and note that secrecy is much more heavily used than had been anticipated. In addition to preferring secrecy, the survey finds that exploitation of lead-‐time, moving rapidly down the learning curve and use of
44 Scherer (1977), p. 64
complementary sales and services are other popular methods of exploiting innovations.
Cohen, Nelson and Walsh argue further that firms choose not to patent due to difficulty in meeting patentability requirements, aversion to disclosure in patent applications, costs of legal defence and ease of inventing around. However, their study also indicates that firms patent in order to prevent copying, block other patents, earn licensing revenue, improve bargaining positions, prevent
infringement suits, measure internal performance and enhance the firm’s reputation.
In a similar paper, Arundel (2001) uses a 1993 European Community innovation survey to determine the relative importance of trade secrecy over patents.
Arundel finds that secrecy is relatively more valuable than patents, although this varies depending on firm market size. Small firms patent less because they prefer to have fewer, more valuable patents, and are perhaps less likely to develop patentable innovations. Larger firms, on the other hand, may prefer secrecy as it allows them to use lead-‐time advantages in the market. Arundel concludes that trade secrets and patents are often used together as
complementary protection measures.
The findings of these two papers caught the attention of the economic
community and spurred the growth in non-‐patent, IPR research. The surveys’
surprising results demonstrate that firms’ preference for trade secrecy over patenting is much higher than previously anticipated. However, these surveys suffer from a lack of empirical certainty in that, by their survey nature, they are subjective. Despite this, the results are surprising and open up a new area in IP and economics research.
While not strictly a survey, Friedman, Landes and Posner (1991), in their foundational paper on the economics of trade secrets law, propose a list of reasons why a firm would choose not to patent and might use trade secrets instead:
1. Time for competitors to independently invent the innovation is:
a. As long as the term of the patent and the innovation has modest value.
b. Longer that the term of the patent.
2. Innovation is non-‐patentable but difficult to independently invent.
3. Firm seeks to avoid disclosure in patenting process.
4. Patenting not cost-‐effective.
5. Innovation is a process innovation and therefore hard to detect infringement.
6. Innovation not easily licensed.
7. Innovation believed to be an early stage of a cumulative innovation.
The strength of their arguments has since been proven by other researchers and their paper directed later avenues of research. The first point, regarding patent term, has been extensively modelled and analysed for policy implications in both patenting and trade secrets terms (as noted in Gilbert and Shapiro, 1990;
Denicolo, 1996; and Scotchmer, 2005, among others.) Their last two points are those that are currently being examined. The suggestion that trade secrets are used because the innovation is not easily licensed is being approached from the licensing and choice of protection point of view, as in Beckerman-‐Rodau (2002.) Finally, the issue of cumulative innovation is under scrutiny from the perspective of trade secrets, as in Bhattacharya and Guriev (2006).
Evidence from Litigation
The empirical analysis of trade secrets using evidence from litigation is in its infancy. Lerner (2006) conducts a preliminary investigation into the empirical evidence found in litigation. Lerner examines approximately 500 litigation cases from two states known for their innovation; Massachusetts and California. His groundbreaking analysis provides some surprising conclusions and indicates a rich area for future research.
Lerner notes that trade secrecy cases face a cost not present in patent cases – the disclosure of the trade secret and subsequent loss of trade secrecy protection.
Whereas a patent gains its protection from its government issued, legally defined status, the same cannot be said of the trade secret. In trade secrecy litigation, the
plaintiff must prove that the protected intellectual property (the trade secret), exists; patents do not suffer from this same burden of proof. In order to prove its existence, the disclosure of the trade secret in the course of litigation may be required. This disclosure risks becoming public knowledge at which point the trade secrecy protection is lost. While the patent may risk invalidation via litigation, trade secrecy protection can be lost entirely, and, with it, market advantage.
A recent case involving the insurance company Allstate in the U.S. demonstrates the risk of loss of secrecy via litigation that Lerner highlights. In a lawsuit filed by an insurance claimant, Allstate was asked to turn over documents produced by a consulting firm, which spell out Allstate’s approach to handling claims. On appeal, an opinion details the extent to which Allstate claimed trade secrecy to prevent the admission of the documents to the court.45 Mauldin (2008) notes that Allstate declared that the “documents are valuable and that great lengths have been taken to ensure that such documents are not disseminated outside of the company.”46 Allstate ultimately won the lawsuit and claimed that the
documents should be sealed to preserve their trade secrecy. However, the judge ruled that the documents were disclosed at a public hearing and should,
therefore, remain public. 47 Thus, Allstate’s trade secrets ultimately were made public.
Lerner’s survey results in some interesting conclusions. First, he notes that trade secrets are often employed in industries, such as software, where patent protection has been limited. Additionally, the probability of winning a trade secrecy lawsuit is less than 40% and only 9% result in damages. This compares to a roughly two-‐thirds success rate in patent litigation. The damages awarded in the survey averaged $1.5 million, which is roughly one-‐third of the average patent litigation award.48
45 Allstate Floridian Insurance Company et al v. Office of Insurance Regulation, Case No 1D08-‐
0275, District Court of Appeal, First District, State of Florida, Opinion filed April 4, 2008.
46 Mauldin (2008), p. 163.
47 Ortiz, Brandon “Secret Allstate documents should be public, judge rules,” Lexington Herald-‐
leader, November 16, 2007 accessed from www.kentucky.com/181/story/233007.html.
48 See Lerner (2006), p. 13.
Lerner highlights the fact that this is an empirical area ready for exploration. At the conclusion of his paper, he indicates his future research will be to expand the database and investigate the litigation decisions themselves. However, Lerner’s paper provides an indication that this will prove to be a fruitful area of research.
In a law paper, Almeling et al (2010) present their descriptive analysis of a database constructed from trade secret litigation in federal courts. The authors gathered data from 394 cases with issued written opinions from 1950 to 2008.
The authors selected a random 25% sample of available cases from 1950 to 2007 (273 cases) and all cases in 2008 (121 cases.) As the authors note, the use of litigation remains fraught with selectivity bias. “The unit of analysis – written decisions that are available on WESTLAW – is only a small part of the complete universe of trade secret misappropriations.”49
The authors find that the number of cases of trade secrets litigation in federal courts doubled from 1988 to 1995, and again from 1995 to 2004. Additionally, they find that the laws of Illinois, California or New York were applied in 30% of the cases. This likely coincides with the fact that courts located in these states are the busiest and in areas of economic importance in the U.S. They also find that the trade secrets themselves are divided roughly between two general categories: internal business secrets and technical secrets.
Furthermore, the authors find that perpetrators of most trade secrets thefts are insiders (i.e. someone known to the owner of the trade secrets.) The
perpetrators are most commonly current or former employees or business partners. The authors report their descriptive findings in the table below.
49 Almeling et al (2010), p. 7.
Table 2-5: Identity of Alleged Misappropriator in Trade Secret Litigation (1950 - 2008)50
Percentage
Employee or former
employee 53%
Business partner 36%
Unrelated third party 4%
Other or unknown 6%
Almeling et al (2010) also address the legal burden of trade secrets’ owners to prove that the trade secret was reasonably protected. Using a logistic regression, whose details are not reported in the paper, they find that the employee
confidentiality agreements and partner confidentiality agreements increased the likelihood, by “more than a factor of 100”,51 that the court would find in favour of the trade secret owner. Additionally, the authors report the following success rates of these litigations:
Table 2-6: Outcome of Trade Secret Litigation in Federal Courts52
1950 – 2007 2008
Owner prevails 42% 52%
Owner does not prevail 53% 43%
Owner prevailed on some trade secret claims but not on others
5% 5%
These values are approximately similar to Lerner’s findings that the owner wins in only 40% and again is lower than the two-‐thirds win rate of patent litigation.
However, the 2008 outcome rates reported in Almeling, and the authors’
observation that the number of trade secret cases is increasing, could indicate that these outcome rates may shift further in favour of owners.
Evidence from Litigation: the EEA
50 Ibid., p. 9.
51 Ibid., p. 26.
52 Ibid., p. 17.
Finally, the literature begins to address the evidence gathered from prosecutions under the EEA. The data used in these papers are most closely related to the focus of this thesis. Two papers pay specific attention to evidence from the EEA:
Zwillinger and Genetski (1996), who focus on the valuation of the stolen trade secrets, and Carr and Gorman (2001), who examine the impact of the
announcement of the theft of trade secrets on the victim firm.
Zwillinger and Genetski (1996) present a legal view of the valuation of trade secrets within the EEA cases. The authors examine best practices for reaching fair and consistent calculations of loss in EEA cases. The authors present an argument that the Fair Market Value and Reasonable Royalty are appropriate methods for EEA cases. The authors highlight the “ad hoc” application of Fair Market Value and argue for sentencing to be more closely based on intended loss to victim and gain to defendant. These issues are further developed in Chapters 4 and 5 of this thesis.
In a paper that adopts a business and law approach, Carr and Gorman (2001) analyse the impact of the announcement of the theft of trade secrets on the victim firm’s stock price. They find that the victim firm’s stock price decreases after a trade secret theft is announced and dub this effect the “revictimization”
by the stock market. The authors model the announcement of the theft of trade secrets and use a regression analysis to determine the impact of the
announcement on the stock price. Their hypothesis is that the public disclosure of the theft has an immediate and negative impact on the stock returns of the victim firms, which is seen via a negative average abnormal return. Using the S&P 500 as a benchmark, they find that the announcement of the theft results in a statistically significant average abnormal return of -‐0.89% in the victim’s stock price. Thus, the authors find in favour of their revictimization hypothesis.
The body of empirical literature examining trade secrets is limited due to the lack of available data sources. This literature review has detailed the
background behind this situation and examined the current state of the
literature. The conclusion is that trade secrets are an area ripe for further research.