Do trade secrets matter? It is not at all clear if you ask the stock market

What happens when trade secrets meet the stock market? Emeritus Katonomist Nicola Searle, and her colleague, Andy Vivian, in addressing this question, suggest some potentially surprising insights.

IP matters. Or, at least, IP should matter to companies. But what if markets didn’t really care about IP? What if IP doesn’t really matter? My co-author Professor Andy Vivian and I are trying to better understand IP’s role by analysing the market reaction to the theft of a company’s trade secrets. Plot twist: the market is remarkably nonplussed.

To address these questions, we use what is known as an ‘event study.’ Event studies take an event, in this case the announcement of case of trade secret theft, and analyse its impact. Our analysis models the firm’s stock market performance before and after the event, while controlling for any fluctuations in the market as a whole. Differences between the market and the firm’s stock price are subjected to statistical analysis, and those differences that survive the harrowing experience of stats are considered ‘abnormal returns.’

The basis of event studies using stock market prices is the ‘efficient market hypothesis.’ This hypothesis, common to financial economics, argues markets efficiently ‘price-in’ changes. The market adjusts prices to account for the present and future financial impact of information revealed by the event. The consequence is that stock market prices reflect the fair market value of the firm, including the impact of the trade secret theft.

Event studies have been used before in IP, in particular regarding patent infringement disputes and patent oppositions. Generally, the very instance of an infringement case conveys positive returns for the patent owner and negative returns for the alleged infringer, irrespective of the outcome. This is so, despite the significant costs for all parties, and the high levels of uncertainty.

One study finds that Apple suffers negative returns, regardless if they are plaintiff or defendant (Nam, Nam & Kim, 2015), suggesting that Apple is particularly sensitive to patent disputes. However, the start of opposition proceedings involves negative returns for the patent owner and, on average, a modest positive return when the case resolves, whatever the outcome, as the uncertainty dissipates. (Nature may abhor a vacuum, but finance particularly abhors uncertainty.)

The cybersecurity literature can also shed some clues as trade secret theft often involves cybercrime. As expected, many studies find a firm suffering a cyber breach also experiences a decrease in its share price. However, a surprising amount of cyber breaches have no real impact on a firm’s stock.

There are a number of hypotheses as to why this might be the case. The frontrunners are that cyber breaches have become so commonplace that they are already priced in by the market, consumers are surprisingly forgiving to companies who have lost their personal data, or cybersecurity does not really matter.

Only two event studies seem to have examined trade secret cases (Carr and Gorman, 2011 and Gupta, 2016.). Both studies find negative returns from the trade secret owner, but both studies relied on very small sample sizes (11 and one, respectively).

We build on this work by using 100 trade secret criminal cases under the US Defend Trade Secrets Act, and its predecessor, the Economic Espionage Act. How does the announcement of a criminal trade secrets case affect the stock price of the victim firm?

Not much.

Overall, we find no statistically significant returns associated with a trade secret theft announcement. The market shrugs ¯\_(ツ)_/¯ and says, ‘no big deal.’ The findings from our model hold even when controlling (adjusting) for effects from the following variables: parallel civil actions, insider crime, corporate defendants, economic or industrial espionage, the firm’s sector, the type and alleged value of the trade secret, financial aspects of the firm and the sophistication of the crime (such as targeting particular trade secrets or hacking)

When we look at the variables individually (meaning we analyse them separately from the complete model described above), we do find some interesting results: negative returns are associated with more severe crimes, outsider crimes, high value crimes, high growth firms and R&D intensive firms. But the overall picture is a surprisingly muted response.

This is not to say that individual cases are not disruptive. Take the 2011 case of AMSC and Sinovel. AMSC sold turbines to Sinovel, its largest customer. In early 2011, Sinovel stopping accepting shipments from AMSC. It then transpired Sinovel had paid an AMSC employee for AMSC trade secrets, suggesting Sinovel was preparing to manufacture its own turbines. The announcement of the trade secrets case in September resulted in the stock market price of AMSC dropping by 22%, which translates to USD$1B of shareholder equity.

The Yahoo Finance graph shows the impact of this theft. In the middle (blue cross) is the date of the announcement of the theft, September 14, 2011. The price drops precipitously following the announcement (yellow arrow); the stock price of AMSC has never reached its pre-April 2011 heights.

Stock Price of AMSC 15/07/11 – 14/11/11 from Yahoo Finance. Blue cross indicates theft announcement.

What does this tell us about trade secrets? It is possible that trade secret disputes have already been priced in by the market and, as such, they may simply be considered a cost of doing business. But it also raises questions as to the true value of trade secrets. If trade secrets are so important, as we think they are, then why is the impact of trade secret theft so minor? Do trade secrets really matter?

There is an alternative explanation that sends chills down my spine - our model is wrong. It is possible our dates are incorrect, which would mean we are not analysing the time when we would expect the market to react. Using the wrong dates could also explain the lack of a statistically significant result.

We have used, to the best of our knowledge, the earlier of the date the criminal case became public knowledge or the date that a civil dispute became public knowledge. We have looked at the impact up to ten days before and after the date.

If any IPKat readers have thoughts that might shed further light on why we do not find any impact, we would be most keen to hear them.

This post is based on our working paper, here.

Picture on lower left is by Benoît Prieur and is licensed under the Creative Commons Attribution-Share Alike 3.0 Unported license.