Measuring the costs and benefits of copyright re-use for follow-on creators: evidence from a crowdfunding marketplace

Kristofer Erickson (University of Glasgow, CREATe), Fabian Homberg & Martin Kretschmer, “Measuring the costs and benefits of copyright re-use for follow-on creators: evidence from a crowdfunding marketplace”

This paper exploits the patronage-like features of an online crowdfunding market to study the effect of transaction and licensing costs on the ability of follow-on creators to successfully re-use existing works. We study the reported costs of production of 1,993 projects seeking funding on the Kickstarter platform from January to March 2014 to investigate if re-use of copyright material has a measurable effect on the cost of creation under four different conditions: (1) the creator has obtained a copyright license; (2) the creator vows to seek a license if funding is successful; (3) the creator borrows from a copyright work but does not discuss licensing; and (4) the creator borrows from a work in the public domain.

Previous experimental research has explored the effect of copyright protection on the ability of follow-on creators to in-license and transform existing works. Crosetto (2010) conducted experiments on sequential creativity in which players were given the opportunity to exclude others from a solution to a word puzzle in exchange for a set royalty fee. When the royalty fee condition was higher, participants tended to protect their innovations, leading to anti-commons conditions and a failure to transact. In a low fee condition, players obtained higher payoffs by opening initial innovations and extending other players’ open solutions. Bechtold et al (2016) employ a similar experimental approach to study creators’ ability to judge whether borrowing or originating is the most beneficial strategy under different licensing cost conditions, on the assumption that copyright imposes costs on sequential innovators:
‘According to rational choice theory, a would-be creator, faced with this borrow/innovate decision, should compare the costs and benefits of borrowing with the costs and benefits of innovating. Borrowing entails a variety of costs including, primarily, licensing fees and transaction costs. Innovating, on the other hand, may involve substantial investments in research and experimentation that borrowing does not.’ (2016: 16).

In order to simulate the cost of borrowing in their experiment, the authors of the study paid a bonus to players who opted not to use solutions discovered by an earlier player, and reduced the points awarded to those who did. Bechtold et al find that follow-on creators do not behave rationally. That is, they borrow when they should innovate and they innovate when they should borrow, leading to reduced individual and social welfare (2016: 26). While interesting evidence of cognitive bias in valuing creativity, the results obtained from these experimental tests of follow-on creation are of limited value to policy. These results are incomplete without supporting evidence about (1) the actual transaction costs encountered in licensing environments and (2) the actual benefits of engaging in sequential vs. follow-on creativity in real world creative industries markets. 
Research on real-world market performance of transformative creative goods is limited, but recent studies have yielded compelling initial results. A series of studies by Hill et al (2010) and Hill & Monroy-Hernandez (2013) examined the quality of remixed work on a collaborative online media platform, Scratch. The authors employed the number of ‘likes’ as a proxy for the quality of follow-on transformative works. The study found that projects which re-used media elements were rated significantly less favorably than original productions by the platform audience, when controlling for number of views and popularity (exposure) of the creator. In film production, research suggests that adaptations and sequels do perform better than original treatments at the US box office (Economist, 2016). However, as pointed out by Hesmondhalgh (2007), this observation may reflect a selection bias introduced by risk-aversion by studios (e.g. production companies supply more sequels and adaptations and market them more aggressively because they represent less risky investments).

Building upon the observations obtained in experimental studies of follow-on creativity, we attempt to measure the presence and effect of copyright licensing costs on follow-on creativity by comparing transformative projects with licenses against projects where re-use has proceeded without a license. Rather than artificially manipulating the payoff for innovators and borrowers, we rely on the mechanism of crowdfunded patronage to assess whether it is optimal to borrow or innovate. In our research setting, if the cost of creation is raised too much due to the presence of licensing costs, this should negatively impact the ability of follow-on creators to secure funding, effectively preventing in-licensing. In the second part of the analysis, we attempt to determine whether the payoffs associated with in-licensing offset any costs associated with obtaining permission to re-use a work. 

Research methods and data:

Data collection consisted of computer-assisted content scraping and human coding of product pitches on the crowdfunding platform Kickstarter. This platform was chosen due to specific characteristics which permit comparison of products using different types of IP inputs. First, by requiring that funded products achieve a minimum required threshold before receiving any funds at all, Kickstarter incentivises producers to minimise the costs of production and honestly report them. Consequently, we use the amount of money that creators initially ask for as a proxy for production costs. Second, because the Kickstarter database contains failed as well as successful projects, it enables researchers to assess the outcome of different IP licensing strategies on success or failure, an improvement on studies of traditional retail markets containing only those products that made it to successful commercialisation (see Heald et al, 2016). 

In this study, we are interested in discerning the costs and benefits of sequential innovation in three possible configurations. 1) a follow-on creator in-licenses a copyright work negotiating a license ex ante; 2) a follow-on creator uses a copyright work without permission, taking on the risk of ex post costs (although avoiding transaction costs of initial permission search) and 3) a creator builds upon a work in the public domain which carries no permission or licensing requirements.

The authors analysed 1,993 product pitches on the crowdfunding platform with funding periods concluding in the first quarter (January – March) of 2014. The sample included successful, unsuccessful, and cancelled or suspended projects in the categories of publishing, video games, comics, theatre and film. These categories were selected because they consist of expressions which can be protected by copyright. Other categories include products which can be protected with patent or design right, which have different licensing requirements. Variables collected from each project included funding level requested, funding level received, quantity of backers, status and previous experience of pitch creator, and the presence of re-use in products which followed on from an existing work. A project was categorised as a re-use when the creators explicitly discussed taking inspiration from another work in the product pitch. Re-use was further scrutinised to see if it pertained to inputs from the public domain, inputs where a license was obtained prior to the product pitch, inputs where the creator vowed to obtain a license subsequent to successful funding and inputs where the pitch creator did not explicitly seek to obtain a license.   

Findings and implications:
 
First, the authors assessed whether transaction costs associated with seeking permission to re-use a work inhibits re-use by producers on Kickstarter. The authors estimate an Ordinary Least Squares (OLS) regression model on the reported cost of production for media products in five categories. Controlling for popularity, status of pitch creator, previous experience and other factors which may influence reported costs, we find a moderately significant negative effect on the cost of production for in-licensing of copyright works compared to ‘pure’ original works. The authors interpret this to indicate that in-licensing has a cost-saving effect for creators who borrow. We further seek to identify whether there are costs associated with seeking a license. We do this by comparing works for which permission was sought against follow-on uses where permission was not required (public domain inputs), and we find no significant difference in reported costs. One possible interpretation is that in-licensing of third-party IP confers additional cost-savings benefits to creators beyond simple permission to use an expression (such as valuable tools or assets related to the license). This result does not support the hypothesis that copyright transaction costs will be reflected in pricing behavior of follow-on creators on Kickstarter. 

To test for the presence of increased payoff for original creators, the authors use a logistic regression employing the outcome of success or failure for product pitches as dependent variable. Holding constant the amount of funding initially pitched for (which is shown to significantly impact the outcome), we find a strongly significant positive effect for transformative use in both in-licensing and public domain adaptation, on the likelihood of a project’s success. This initial result suggests that there is a payoff incentive to engage in follow-on creativity based on copyright inputs, in addition to the cost savings incentive.

Our paper points to reconsideration of the directions for future empirical and experimental studies of copyright’s effect on sequential innovation. One of the shortcomings of experimental approaches to testing copyright incentives has been that due to the small labour investments expected of laboratory participants, the transaction and licensing costs measured in experimental settings make up a high proportion of the overall costs of ‘production’. In our real-world study, production costs ranged from under £1000 to over £100,000 GBP. Even excluding larger projects of >£20,000 we could not detect the presence of licensing transaction costs. 

Second, since experimental approaches to sequential creativity have focused on tasks with relatively straightforward rules and limited solutions, the advantage of in-licensing for sequential creators has been re-cast as an access problem. Our study cannot assess whether access to copyright works limits creativity on Kickstarter since those refused a license may be deterred from pitching in the first place. However, our results suggest that a copyright license in practice includes more than simple access to a work, since public domain adaptations and unauthorised re-uses were not associated with lower costs of production. Instead, the savings effect was found only where producers explicitly obtained a license to re-use a copyright work. This suggests that there is more to a licensing transaction than simple permission – the value of a license may include raw material and know-how inputs to production which have an importance independent of copyright.