I’ve written on a couple previous occasions about how our approach to copyright policy is badly distorted by wildly inflated estimates of what online piracy “costs” the U.S. economy. The true figure, as most serious analysts admit, is likely unknowable, but the content industries have discovered that no figure is too ludicrous to be parroted with a straight face by well‐​meaning politicians. The higher the fabricated number, the easier it becomes to claim that even the most expensive and draconian antipiracy measures, however questionably effective, can pass a cost‐​benefit test. Some recent news involving the video streaming site Hulu reminds me of yet another reason to be wary of those figures.

According to press reports, free access to Hulu content may soon be limited to users who already subscribe to a traditional cable package. The incumbent cable companies hope this will entice viewers to buy or maintain more profitable cable subscriptions rather than “cutting the cord” and shifting entirely to online viewing. Some may, of course, but others will predictably turn to piracy: Tech reporter Ryan Singel of Wired joked on Twitter that the Pirate Bay was probably purchasing new servers in response to the announcement. Regardless of whether Hulu ultimately opts for this approach, there’s a more serious general point to be teased out there, however.

Once digital content is produced, the difference in distribution cost between selling a hundred copies and a million is negligible. That creates enormous flexibility in pricing strategies, ranging from the pay‐​per‐​view “premium” content model (high price, small but devoted audience) to the free ad‐​supported model of broadcast television (much smaller profit per viewer, but made up for by a much larger audience). In practice, movies employ all of the above sequentially—or even simultaneously in different markets or formats. This has various benefits for both consumers and the studios, but also makes it that much more difficult to objectively estimate the “losses” attributable to piracy.

To illustrate, let’s imagine television show that initially streams online for free with advertising, garnering a million viewers per episode and earning $1 per viewer in ad revenues, for a total of $1 million. A small number who really dislike ads, or have connections too slow for streaming, let’s say 5,000, download pirate copies anyway—but the vast majority watch legally. After building an audience and generating some good word of mouth, the accountants suggest that it might be more profitable to stop the free streaming and instead sell ad‐​free episodes for $4, in hopes that enough dedicated fans will pony up to compensate for the predictable drop in viewership once the program is no longer free to watch. The paying audience does indeed drop to 255,000, which still leaves the company slightly better off for the switch, but 100,000 viewers decide to keep up with the show (at least initially) by downloading pirated copies. A subsequent price hike to $10, however, turns out to be a money loser. Now the show has only 80,000 paying viewers, while 150,000 are engaged in piracy.

Undoubtedly that piracy is costing the show’s producers something: If piracy were impossible, some unknown fraction of those who download illegally would be willing to pay the asking price. But just crudely using the actual market price at each stage—even if modified by some constant “displacement rate” to acknowledge that not every illicit download represents a lost sale at that price—yields some perverse results. As the pricing strategy for the show changes, the “cost” of piracy rises from $5,000 to $400,000 (even as revenue rises) to $1.5 million (while revenues drop by $20,000). Obviously, something is wrong here.

It’s no great mystery what: The problem is that the rate of piracy, the price of a digital good, and the “displacement rate” (the percentage of the pirates who’d buy at that price in a world of perfect copyright enforcement) are not independent variables. And, of course, the interdependency runs both ways: Pricing decisions are influenced by the knowledge that we don’t live in a world of perfect enforcement, and you can tell plausible stories according to which this might keep prices higher or lower than they’d be under perfect enforcement, depending on your assumptions about the conditions under which a particular audience will substitute the pirate for the legal good.

This does not, of course, mean that content producers are somehow at fault when piracy increases after they raise prices (either directly or, as in the Hulu case, indirectly, by tacking on the cost of a cable subscription). They’re at liberty to charge what the please. But it does mean we’ve got to exercise a bit of critical scrutiny when deciding whether the “costs” of piracy justify the “costs” of enforcement.

Returning to our imaginary program, suppose that under perfect enforcement—a zero piracy world—there would be 110,000 paying viewers at $10 per episode, netting the creators an additional $80,000 over what they’d make with their revenue maximizing strategy ($4 per episode) in the world of imperfect enforcement. That’s great for them, if not for consumers, but we haven’t factored in the costs of enforcement. Some of these are likely to be borne by the creators themselves—hiring lawyers to hunt down pirate copies circulating online and the like—but in practice they’re often shifted to taxpayers, in the form of direct enforcement expenditures, or to other parts of the economy, in the form of DMCA compliance costs or innovative services that are deterred entirely. It’s possible that, in this hypothetical scenario, the revenue maximizing strategy for the producers is to charge $10 while externalizing the costs of perfect enforcement, but the socially efficient outcome is to accept imperfect enforcement and let the producers revenue maximize against that background at a $4 price point.

Obviously, these particular numbers shouldn’t be taken too seriously—I pulled them out of a hat for the purposes of illustration—and I could’ve told an alternative story where perfect enforcement induces the producers to maximize revenue by lowering prices to lure in the highly price sensitive ex‐​pirates. In that case, you’d have to count the benefit to the producers and to the legal consumers who now enjoy lower prices in assessing whether the enforcement was worth the cost. It’s an empirical question which story would more often match the facts.

The point here is not to argue for more copyright enforcement or less. It’s that we can’t say anything meaningful about the net social costs of piracy—and the costs we should incur to reduce it—without taking into account the fact that those “costs” depend on pricing strategies chosen in partial response to piracy. Probably that’s not possible in practice—but the impracticality of generating a realistic estimate is no reason to take our existing unrealistic estimates seriously.