Copyright is too long

The Tension Between Incentives to Create and Access

As reported in the New York Times there has been a rise of copyright enforcement lawsuits in music with million-dollar payouts. These suits, often based on flimsy evidence of musical theft, have found new life following the $5.3mn payout to the estate of Marvin Gaye for infringing copyright in the song “Blurred Lines.”

The economic argument for copyright is that there is insufficient incentive for artists to create because art can be cheaply copied. Granting a creator monopoly rights over their creation makes it profitable (or maybe less unprofitable) to make art.

Does it work? Do artists actually benefit from copyright? Sure. A 2013 survey of 5,000 musicians by Peter DiCola found the average share of revenues directly attributable to copyright was 12%, but that this was significantly higher for composers and for musicians in the top 1% of earnings.

More rigorously, there is a large literature that tries to tease out the impact of copyright by identifying dates when copyright status changed abruptly and comparing artistic works published right before and after this date. MacGarvie and Moser (2013) looks at the UK Copyright Act of 1814 and finds authors published between 1800 and 1830 earned nearly twice as much after copyright was extended in 1814.

More recently, the 1998 Sonny Bono Copyright Term Extension Act extended copyright by 20 years. Since the act only applied to content still under copyright, books published before 1923 were no longer copyrighted while those published after 1923 had an extra 20 years. Reimers (2019) gathers data on the top 10 best-selling novels each year from 1910-1936 and their price and availability today. Books published over 1923-1936 cost more and have fewer editions available compared to those published in 1910-1923.

So copyright surely raises the revenues of creators. But it also imposes a cost on society, since higher prices restricts access to the ideas in copyrighted work. That can have a negative effect on the creation of new art.

The best data I know of on this comes from baseball and science. In 1963, US copyright for magazines was 28 years, which could be renewed for an additional 28 years if the rights holder cared enough to file an application. In 1964, renewal was made automatic for works published in that year and after. Nagaraj (2018) exploits the fact that Baseball Digest apparently did not care enough to file for renewal, meaning issues published prior to 1964 had 28 years of copyright protection, expiring no later than 1991, while issues published after 1964 had 56 years of protection, expiring no earlier than 2020. Nagaraj shows wikipedia (over 2004-2012) cites issues of Baseball Digest published before 1964 135% more than those published after 1964. The access to better information (especially images - see below) seems to have made those wikipedia pages higher quality - they get visited 20% more. Meanwhile, those extra 28 years of protection couldn’t have been that valuable to Baseball Digest, since it didn’t bother to apply for them.

Lastly, Biasi and Moser (2018) exploit the 1942 Book Republication Program, which legalized piracy of German science books in the US for a short period during World War II. The subsequent wave of cheaper editions of these books diffused into libraries across the country, eventually raising the citations to these textbooks from books, articles, and patents, as compared to Swedish science textbooks (whose copyright was respected).

The above studies illustrate the tension in intellectual property rights. The longer the copyright, the higher the profits and (presumably) the larger the incentive to create. At the same time, long copyright also prevents reuse and improvement of ideas. This is important in science and technology, but also in art. So how long should copyright be?

Alex Tabarrok argues for a laffer curve in patent strength, where additional patent strength increases innovation at low levels and impedes it for high levels.

Patents last 20 years these days. There’s an argument for longer protection for copyright, since patents offer broader protection than copyright and revenues from art are probably smaller than for technology on average. So maybe 28 years, renewable once was reasonable, though my preference would be for less. But copyright today lasts for the life of the author plus 70 years. This is insane.

Suppose I write a book that will give me royalties of $10,000 per year forever. At a 5% interest rate, this has a present discounted value of $200,000 (if you invest $200,000 at 5%, you can earn $10,000 per year in perpetuity). If I have “only” 56 years of copyright, the present discounted value falls to $189,000, or 95% of the value I would get if I had a never-ending copyright. So, from the perspective of present value, the latest waves of copyright extension have increased the value of new works by a few percent at most. I don’t believe anyone is creating artwork because of that extra incentive. Of course, if the work already exists and copyright is about to expire, then the situation is different and extension is quite profitable - but at that point, it’s just rent-seeking.

Suggestive evidence that long copyright isn’t essential to a flourishing creative scene comes from Waldfogel (2014). Waldfogel attempts to measure the quality of music in the wake of Napster. The 1999 rise of Napter can be interpreted as a reduction in the power of modern copyright, since it enabled a non-negligible share of listeners to pirate music they would have otherwise purchased. But Waldfogel finds no evidence the quantity or quality of music declined (based on album sales and critics “best-of” lists). But as Waldfogel is careful to note, the evidence here is a bit muddled since the impact of piracy is mixed up with the effect of digital technology on the costs to record and distribute music.

Cleaner evidence comes from an era when technological change in music production was much slower. Giorcelli and Moser (2019) looks at the uneven rollout of copyright across Italian states in the 1800s (driven by unrelated political and military dynamics). They find going creating a basic copyright system (life of the author plus 10/12 years) led to about a 150% increase in the annual creation of new operas, compared to states without copyright. However, going from life + 10/12 years to life + 30 years and then to life + 40 years had no detectable positive effect. And in some instances these extensions had a definite negative effect on the creation of new work.

To wrap up: some copyright is better than no copyright, but beyond a point, it’s doubtful additional profits (realized in the far-distant future, if at all) do anything to increase innovation. Meanwhile, too much copyright allows the heirs and estates of deceased creators to block the next generation from riffing on, recombining, and generally advancing the music created by prior generations. We’re in the latter category.