RoyaltyStat Blog

Effect of Text Redaction on Royalty Rates and Other Licensing Terms

Posted by Alan Kwan

Any participant in the process of commercializing technology knows that secrecy is important for many reasons. First, if one discloses a valuable technology or trade secret, this may invite competitors wanting to ape that technology. Second, even if the technology is protectable, confidentiality about transaction terms may protect counterparties in future negotiations. Third, secrecy may just be safest when there is risk that audiences may misinterpret that information. When commercializing innovation, what is most frequently kept secret?

Licensing data such as that collected by RoyaltyStat is practically the only setting in which to test this question and gain a window into what priorities for innovators are when it comes to secrecy. Usually, we do not observe what was not disclosed. However, publicly listed firms must file “material” updates, which include major licensing agreements. Despite mandatory disclosure, firms can choose to redact (not disclose) key details citing competitive concerns or other proprietary costs, which members of the public can identify and acquire through the Freedom of Information Act (FOIA) years later. Using these disclosed and redacted documents, we can see what firms disclosing licensing transactions (commercialization of innovation) hide. Do they hide the IP, a product, the commercial terms or a contractual term, such as how long the arrangement is for? Conceivably, each of these items has their own value.

For this research task, Ednaldo Silva of RoyaltyStat graciously provided us 1,049 originally redacted documents and their unredacted form. We analyzed these documents by using textual analysis to identify and extract redacted passages. We used computer-based rules to flag about ¼ of the approximately 60,000 redactions and hand-classified the rest. For simplicity, the redactions were classified into four categories: product, intellectual property (software, trademark, trade secret, patent), commercial term or other contractual detail, which includes the duration of contracts, conditions for termination and other legalese that is seemingly unrelated to either the product market or its value directly.


The median contract contains 20 redacted clauses, with the max redacting 636 clauses of text. There exists rich variation in what is redacted. Although it is clear payment-related terms are the most redacted 94%, it is not true in every case. IP-related information is almost as commonly redacted, and is concealed around 74% of the time, and some combination of IP- and product-related information is almost as frequently redacted as payment. Because licensing contracts are agreements of commercialization, payment terms being most redacted is natural because the payment terms are most heavily discussed (e.g., there are more redactions because there is more to redact). Imaginably, for startup companies further away from commercialization, the focus may shift even more so toward the protection of IP.

Answering this question itself is interesting but documenting what is redacted (and how frequently) enables us to better understand our previous findings. We may discuss this more on future blog posts. First, Gaurav Kankanhalli and I document that redacted contracts tend to have lower royalty rates. This suggests licensors choose to suppress price concessions. This is important for legal scholars due to patent litigation and accounting scholars due to transfer pricing. Second, Gaurav Kankanhalli, Ken Merkley and I find that redacted contracts tend to receive bullish investor reactions, whereas prior literature suggest secrecy has usually irked investors. A natural question is whether this bullish investor reaction is dominated by reactions to payment terms or IP. Preliminary analysis suggests it is strongly dominated by redactions of IP and the positive reaction of investors seems less driven by concealment of payment terms.

As we revise our papers to fully incorporate these new findings ahead of the journal publication process, we look forward to culminating our understanding of these important issues. We also hope that future research adopts our detailed approach, either deepening the content classifications of IP or by extending this content classification approach to other types of strategic redactions made by firms, for example in supply agreements.

Published on Feb 10, 2020 10:05:36 AM

Alan Kwan (Ph.D.) is Assistant Professor of Finance, University of Hong Kong. He can be contacted at:

RoyaltyStat provides premier online databases of royalty rates extracted from unredacted license agreements
and normalized company financials (income statement, balance sheet, cash flow). We provide high-quality data, built-in analytical tools, customer training and attentive technical support.

Schedule an Online Demo 

Topics: Text Redaction and Royalty Rates