Piracy in Japan: What the 10 400 Billion Yen in "Losses" Actually Measures

The figure has everything needed to become a perfect headline: 10 400 billion yen in damage caused by piracy and counterfeiting of Japanese content in a single year.

It genuinely comes from the Japanese government. On January 26, 2026, the Ministry of Economy, Trade and Industry (METI) announced that a study conducted on its behalf by the Content Overseas Distribution Association (CODA), then subcontracted to PwC Consulting, estimated 5 700 billion yen in damage from illicit digital copies in 2025 and 4 700 billion for counterfeit character merchandise. Total: 10 400 billion.

But this number does not mean that Japanese studios, publishers, musicians, and developers would have collected an additional 10 400 billion yen if piracy had disappeared.

It is precisely here that the study becomes far more interesting than its headline.

The full report shows that the 5 700 billion digital figure is already a corrected estimate that attempts to exclude a large portion of consumption that would never have become purchases. By contrast, the 4 700 billion for counterfeit merchandise follows a different logic: it essentially corresponds to an estimate of money spent on counterfeits.

In other words, the 10 400 billion is a useful aggregate for measuring the scale of a phenomenon, but far trickier to interpret as a genuine “economic loss.”

Where Exactly Do the 10 400 Billion Yen Come From?

METI did not merely monitor a few major pirate sites. The study rests in particular on consumer surveys in six countries: Japan, China, Vietnam, France, the United States, and Brazil, before extrapolating globally. It covers four digital categories and, for the first time in this edition, counterfeit merchandise tied to Japanese characters.

The official rounded figures are as follows:

Category	2025 Estimate
Video	2 300 billion ¥
Publication	2 600 billion ¥
Music	300 billion ¥
Video Games	500 billion ¥
Digital Total	5 700 billion ¥
Counterfeit Merchandise	4 700 billion ¥
Reported Total	10 400 billion ¥

In PwC’s detailed tables, the unrounded digital total used for the main estimate is 5 682 billion yen. The report also produces a second estimate based more on existing legal markets, which reaches 6 523 billion. The public figure of 5 700 billion therefore corresponds to the so-called “user-based” approach, chosen notably because it allows comparison with the 2022 study.

This difference between 5 682 and 6 523 billion, nearly 15%, is already instructive: there is no global counter capable of directly observing the revenues destroyed by piracy. The result depends on the model used.

This does not make the study useless. It simply defines what it is: an estimate.

The 5 700 Billion Does Not Assume “One Piracy Act = One Lost Sale”

This is probably the most important correction to make to the simplified narrative that accompanies this figure.

PwC distinguishes two concepts.

The first is damage in the broad sense: it estimates the monetary value of all Japanese content consumed in pirated form.

The second is damage in the narrow sense, which the report describes as an amount of prevented sales. To move from the first to the second, PwC applies a conversion rate corresponding to the proportion of pirated consumption that users declare could have been replaced by a legal purchase if the pirated version had not existed. The report explicitly defines this rate as the share for which there is a possibility of “conversion to a legitimate purchase in the absence of piracy.”

The result changes radically.

Under the user-based approach adopted for the official figure, the broad-sense value reaches 18 914 billion yen. After applying conversion rates, 5 682 billion remains.

By dividing the narrow estimates by the broad estimates published by PwC, one can reconstruct the order of magnitude of the filter applied:

Category	Broad Value	Narrow Damage	Implied Ratio
Video	8 494 Bln ¥	2 323 Bln ¥	≈ 27.3 %
Publication	8 069 Bln ¥	2 584 Bln ¥	≈ 32.0 %
Music	952 Bln ¥	323 Bln ¥	≈ 33.9 %
Games	1 398 Bln ¥	452 Bln ¥	≈ 32.3 %
Total	18 914 Bln ¥	5 682 Bln ¥	≈ 30.0 %

This last percentage is my calculation based on PwC’s tables; the report actually performs its conversions at more detailed levels before aggregation.

But the order of magnitude is clear: roughly 70% of the theoretical value of pirated consumption is already excluded from the reported “digital damage” figure.

METI therefore does not reason at all as if each illegally watched episode, each downloaded manga, or each pirated game automatically constituted a full lost sale.

CODA had moreover explained this very clearly as early as the previous study. Before a committee of the Japanese Agency for Cultural Affairs, the organization specified that a user consuming a work solely because it is free should not, in principle, be counted in the damage. CODA also insisted that it was an estimate and that the figure should not be considered absolute.

This is a considerable nuance.

The Real Weak Point Shifts: How to Know What a Pirate Would Actually Have Bought?

Discarding the “one copy = one sale” hypothesis solves one problem but opens another.

The conversion rate rests on a counterfactual: one essentially asks the consumer what they think they would have done in a world where the pirated offer did not exist.

Yet this world is not observable.

Someone may sincerely answer that they would have bought a manga, then never buy it when 8 or 10 euros must actually come out of their wallet. Conversely, some users may underestimate what they would have spent. Local availability, price, release delays, the presence of a translation, existing subscriptions, or the existence of free legal offers also alter the decision.

This is the difference between stated preference and revealed behavior: saying what one would have done does not offer the same level of causal proof as observing what consumers actually do when a pirate offer appears or disappears.

The report nonetheless tries to avoid another source of distortion: extreme responses. After an initial filtering of invalid responses by the questionnaire provider, PwC performs a second data cleaning and notably excludes certain values exceeding the 80th percentile of monthly consumption or expenditure. For the digital portion, the indicated valid sample sizes become 243 respondents in Japan, 194 in China, 182 in Vietnam, 188 in France, 205 in the United States, and 158 in Brazil.

This choice is defensible: a few individuals declaring thousands of consumptions can blow up a global extrapolation.

But it also produces a trade-off. The most intensive users may be aberrant responses… or genuine very heavy consumers of pirated content. Removing them reduces sensitivity to anomalies while risking amputating a real part of the distribution.

The global estimate then depends on other factors such as the connected population, the proportion of Japanese content users, prices, and exchange rates.

This explains a result that might seem contradictory: the number of pirated content items consumed per person decreased between the two surveys, METI indicates, while the estimated digital damage rose from around 2 000 to 5 700 billion yen. The ministry attributes this increase to several factors, notably price and exchange rate variations, the increase in the number of concerned users and the connected population, and the growing global penetration of Japanese content.

Saying that “Japanese piracy has tripled” would therefore be misleading.

It is the monetary damage estimate that has nearly tripled, not necessarily the individual intensity of piracy.

The 4 700 Billion in Counterfeit Merchandise Poses a Different Problem

This is probably the easiest part to miss when mechanically adding 5 700 and 4 700 billion.

For counterfeit merchandise, PwC does not use exactly the same substitution mechanism as for digital content.

The official estimate of approximately 4 681 billion yen comes from an approach based on distribution channels. The survey seeks to identify expenditures made on different categories of e-commerce sites: channels considered legitimate, mixed platforms where genuine and fake products can coexist, and channels considered dedicated to counterfeits. Coefficients are then applied before global extrapolation.

There is a fundamental conceptual difference here.

For digital:

valued pirated consumption → conversion rate → potentially prevented sales.

For merchandise:

estimated expenditure on counterfeit products → estimated damage.

The report even specifies that it does not calculate, for merchandise, the exact equivalent of digital damage “in the broad sense”, notably because the questionnaire does not allow obtaining the number of units and a reference legal price in a comparable way.

PwC also performs a verification with another method: starting from products that consumers themselves declare recognizing as counterfeits. This approach yields 6 429.3 billion yen, against 4 680.7 billion for the channel-based approach ultimately retained, about 37% more.

This is not proof that either of the two methods is false. It is a demonstration of the result’s sensitivity to how one defines and identifies a counterfeit.

And above all, spending 2 000 yen on a counterfeit figurine does not necessarily mean that an official 8 000-yen figurine would have been purchased otherwise.

Some buyers specifically seek the very low price of a copy. Some know they are buying a fake; others do not. Some counterfeit products may correspond to official items unavailable locally. Some consumers would have switched to the original; others would have bought nothing.

A genuine estimate of lost official sales would therefore require, here too, a substitution counterfactual.

This is why the final sum of 10 400 billion must be handled with caution: under a single “damage” label, it brings together a model of potentially prevented digital sales and an estimate of expenditures devoted to counterfeit merchandise.

Mathematically, the addition is correct.

Economically, the two terms do not mean exactly the same thing.

Independent Research Confirms That the Substitution Rate Is Never Universal

Economists have debated the relationship between piracy and sales for over twenty years. And the literature supports neither the simplistic idea that every copy is a lost sale, nor its opposite that piracy has no economic cost.

A famous study by Felix Oberholzer-Gee and Koleman Strumpf published in 2007 in the Journal of Political Economy pitted actual P2P download data against US album sales. Its estimate of the effect of file-sharing on sales was statistically indistinguishable from zero and incompatible with the idea that file-sharing explained the bulk of the sales decline during the studied period.

It would nevertheless be abusive to conclude that “piracy causes no lost sales.” It concerned music, 2000s P2P, and a commercial environment far removed from today’s Crunchyroll, Netflix, Steam, or manga platforms.

Other experiments find precisely the opposite effect.

Brett Danaher, Michael Smith, Rahul Telang, and Siwen Chen studied the arrival of the graduated response Hadopi in France by comparing sales evolution with other European countries. They estimate that awareness of Hadopi increased French music sales on iTunes by approximately 22 to 25%, with larger effects in genres previously more heavily pirated.

This constitutes an indication much more compatible with a genuine substitution between illegal consumption and purchase.

But here again, the result is not “1 for 1.”

A study by Benedikt Herz and Kamil Kiljański covering nearly 30 000 respondents in six European countries estimates that a first free and unauthorized viewing of a film displaces on average approximately 0.37 paid legal viewings. Their calculation implies a sales decline of about 4.4% over the studied period, with very large differences across countries. Even more remarkable: 94% of estimated lost sales came from just 20% of consumers.

A vast research project by the University of Amsterdam’s Institute for Information Law, conducted in thirteen countries including Japan, also finds highly variable rates depending on content type and consumption mode. For hit films, its temporal analysis yields an aggregated range of approximately 0.20 to 0.45 displaced legal viewings per first illegal viewing after various controls.

This study nonetheless deserves a clarification: it was funded by Google, although the researchers state they worked under a regime of scientific independence. This funding does not invalidate its results, but is part of the relevant context for evaluating them.

The interesting conclusion is therefore not that a particular rate of 0.30, 0.37, or 0.45 would be the “right” one.

It is exactly the opposite: there probably does not exist a universal conversion rate from piracy to lost sales.

It depends on the work, the country, the price, income, age, ease of legal access, release chronology, medium, and consumer profile.

Pirates Are Also Often Among the Best Customers

An additional difficulty appears in several consumption studies: the “pirate” and “buyer” populations do not necessarily form two distinct groups.

The international IViR study observed, for example, that the vast majority of people illegally consuming content also consumed it legally, and that their median legal consumption could be significantly higher than that of non-pirates.

This phenomenon is important for causal interpretation.

Suppose a very active fan buys fifteen manga per month and reads five more on pirate sites. A simple correlation could show that the pirate buys more than the non-pirate. This obviously does not prove that piracy increases their purchases: they may simply be a much more passionate consumer.

Conversely, observing five additional pirated readings does not allow concluding that five purchases have disappeared.

One must isolate what would have happened without illegal access.

This is precisely the counterfactual problem that PwC tries to approach with its conversion rate. Its approach is therefore economically more defensible than a naive multiplication of the number of copies by the retail price.

The weakness lies elsewhere: this counterfactual is primarily declared by consumers rather than observed through a causal experiment.

The 5 700 Billion Figure Is Therefore More Serious Than It Looks and Less Precise Than It Looks

These two propositions can be true simultaneously.

It would be unfair to present the METI study as an operation consisting of attributing a work’s price to each pirated copy. The report explicitly avoids this error. By reducing roughly 18 900 billion yen of valued digital pirated consumption to roughly 5 700 billion in narrow damage, it acknowledges that a majority of pirated uses would presumably not have been converted into purchases.

But it would also be excessive to treat the 5 700 billion as an objectively measured accounting loss.

No bank statement shows 5 682 billion yen missing.

It is an estimate of the revenue that could potentially have been captured in a counterfactual world without piracy, built from surveys, prices, volumes, populations, conversion rates, and extrapolations.

This distinction may seem academic, but it completely changes how the figure should be communicated.

Even CODA Distinguishes Pirated Value from Actually Lost Sales

A recent Japanese document provides a particularly enlightening verification.

ABJ, the organization that monitors Japanese publication piracy among other things, estimates that in 2025 approximately 688.8 billion yen worth of manga and other publications were read for free on the ten main pirate sites targeting the Japanese market. But the document explicitly warns that this is the value of free readings and not an equivalent decline in legal sales.

It adds that reusing a conversion rate near 30%, consistent with the METI/CODA study, the order of magnitude of the commercial loss would be rather around 230 billion yen.

This is exactly the distinction that the public debate should retain:

value of what is pirated ≠ money that would necessarily have been spent legally.

So, What Does the 10 400 Billion Figure Actually Mean?

The most rigorous formulation would be the following.

Japanese authorities estimate that in 2025, the unauthorized exploitation of Japanese content online represented a gigantic amount of economic consumption. For digital, their model evaluates at approximately 5 700 billion yen the fraction likely to have replaced legal transactions. For counterfeit merchandise, they simultaneously estimate at approximately 4 700 billion yen the global expenditures made on counterfeits identified following their channel-based method.

This suffices to show that the problem is colossal.

This does not demonstrate that Japanese companies would have realized 10 400 billion yen in additional revenue in a world without piracy.

The distinction is all the more necessary because this figure has already entered Japanese political discourse on international intellectual property protection and content industry development. The shift from a model estimate to a memorable political number is precisely the moment when the hypotheses that produce it risk disappearing.

The Figure Japan Should Publish in the Future

METI’s work would paradoxically gain strength if it resisted the temptation of a single total.

Three separate indicators would be more informative: the total value of digital pirated consumption, the amount of potentially displaced digital sales after conversion, and expenditures devoted to counterfeit merchandise. Alongside these, publishing multiple substitution rate scenarios would immediately show the result’s economic sensitivity.

This would prevent a figure like 10 400 billion from being interpreted now as the size of piracy, now as stolen revenue, now as a certain shortfall.

The PwC study already contains a large part of the necessary nuances. The problem appears especially when they are compressed into a single sentence.

The final finding is therefore more subtle than “Japan exaggerates its losses” or “piracy costs nothing.”

Piracy very likely has a real, sometimes significant economic cost, and independent research shows it can indeed displace sales. But an illegal consumption is not a disappeared sale, and the 10 400 billion yen are not 10 400 billion yen in observed revenue that would have been taken from Japanese creators.

It is a composite indicator of economic exposure and potential shortfall.

And to understand what Japan truly loses, this difference is worth several thousand billion yen.