At the end of May, the US Congress passed the Stop Counterfeiting in Manufactured Goods Act. Faced with a huge and growing problem, the Congress acted decisively: simple possession of counterfeit labels can now mean the confiscation of an entire factory, even if the factory is mainly used for entirely legitimate purposes.

But just how huge is this problem? Look at the act itself. Right at the beginning, under "Findings", we're told that

(2) the U.S. Customs Service and Border Protection estimates that counterfeiting costs the U.S. $200 billion annually.

When I saw this, I was surprised. That $200 billion number looked suspiciously similar to the basis for New York Comptroller Bill Thompson's bogus statistics. This time, however, there was a cited source for it, so I called up the Customs Service.

I spoke to a very friendly spokeswoman called Erlinda Byrd, who told me that Customs didn't compile such numbers at all: she had no idea how their name wound up getting attached to that figure in federal legislation.

So I determined to find out how big the counterfeiting problem really is. Virtually any article you read on the subject will have some language about how counterfeiting "is estimated to cost" or "is believed to be" $200 billion or $500 billion or some other number each year. Most of the time, there's no indication at all of who is doing the estimating or the believing. I made it my job to find out where these numbers came from, and whether they made any sense.

There's certainly no shortage of experts in this field. I spoke to Peter Lowe, the assistant director of the Counterfeiting Intelligence Bureau, which is part of the International Chamber of Commerce. He reckoned that the numbers probably came from an estimate that counterfeiting accounted for between 5% and 7% of world trade; the source of the 5-7% number itself, however, he said, was "lost in the mists of time". Like most people I spoke to, Lowe was wholly upfront with me: "I don't know what the source of that was, to be honest," he said, "but it's widely bandied around: it's passed into anti-counterfeiting folklore."

I spoke to Steven Gursky, a lawyer who is a board member of the International Anti-Counterfeiting Coalition, and who runs the intellectual property practice at the law firm of Dreier LLP. "People seem to have gotten very comfortable with the number of $200 billion: I've heard it over and over again," he said. Gursky was a source of anecdotal numbers. "If one company can sell $30 million of fake goods to one customer in one year, that's why the number doesn't shock me," he said. After all, the $30 million was just in the clothing industry: there's also auto parts and shaving blades and batteries, and airplane parts and baby formula and pharmaceuticals and all manner of other things which are commonly faked.

I spoke to Tim Trainer, who used to run the International Anti-Counterfeiting Coalition, and who now runs the Global Intellectual Property Strategy Center. He said that "trying to gauge illegal conduct is impossible," but that you could at least make a stab at getting some numbers by looking at customs seizures and extrapolating from there.

I spoke to Ruth Orchard, director general of the Anti-Counterfeiting Group, which actually went to far as to commission a couple of reports from the CEBR looking at the economic impact of counterfeiting.

And yet, after talking to all of these experts and reading (or at least skimming) hundreds of pages of reports in a search for any kind of hard data or even backed-up estimates, I found pretty much nothing.

There's lots of propaganda out there, of course. The IACC helped to sponsor a Harpers Bazaar event on counterfeiting, and gave attendees a handout entitled "Fakes by the Numbers". It started by saying that estimated annual sales in counterfeit products worldwide are $500 billion, and even borrowed from Bill Thompson's report, juicing it up even further by saying that $1 billion is "the minimum estimated annual loss in tax revenues in New York City due to counterfeiting".

The ACG has its own factsheet, which says that "in the US, the FBI estimates losses to counterfeiting at US$200-250 billion a year", and even gets industry-specific: "The Motor & Equipment Manufacturers Association (MEMA) conducted a survey indicating that the global automotive industry loses US$12 billion to counterfeiting, thereby resulting in the loss of 750,000 jobs". But if you look at the MEMA publication, you'll see that there was no survey: instead, it cites a 1997 Federal Trade Commission report to back up the $12 billion number. The 750,000 number is nowhere to be found – which is unsurprising, since it's ludicrous on its face.

This is a pattern: every time you think you're getting one step closer to a real survey, or real data, the purported source of the information turns out simply to be citing someone else. For instance, the International Chamber of Commerce has a pretty chart on page 144 of its International Anti-Counterfeiting Directory, showing the value of trade in counterfeits exceeding $350 billion as world trade in general increases. But really it's just a chart of world trade, shown once at the 100% level and once at the 6% level. Reverting back to that annoying passive voice, the caption for the chart simply says that "the overall level of counterfeiting in the world today is generally estimated at 5-7 per cent of world trade".

Insofar as anybody ever cites a source for the 5-7% figure, it's an OECD report from 1998 entitled "The Economic Impact of Counterfeiting". This is a very comprehensive report, but it never made an attempt to quantify the size of the problem. Instead, it reverted the same old passive voice:

The overall costs of counterfeiting in the world today are normally estimated to be 5-7 per cent of world trade. There is no substantial aggregated data to support the high percentages, but the figures are now accepted and used to illustrate the extent of the counterfeiting problem.

In other words, a report which admits there's no data behind the percentages, and even goes so far as to call them high, is the most commonly cited source for those very percentages.

The report then gets positively funny:

In 1997, the Counterfeiting Intelligence Bureau (CIB) of the International Chamber of Commerce (ICC) calculated the nominal value of the estimated share of counterfeit goods as a percentage of world trade. They used aggregated data on total world trade provided by the World Trade Organisation and took the general assumption that counterfeiting has increased from 3 per cent in 1990 to more than 5 per cent in 1995. (My emphasis.)

In other words, multiplying one number by 3% and another number by 5% counts as a "calculation". Well, I just calculated, in that case, that US chocolate consumption has doubled in the past year, since it was 2% of GDP a year ago and is 4% of GDP now. The point is that unless those percentages have any kind of factual basis, they're at best meaningless and at worst outright lies.

The most footnote-rich report comes from the IACC, in a white paper (I have a revised version, which I can't find online) entitled "The negative consequences of international intellectual property theft: Economic harm, threats to the public health and safety, and links to organized crime and terrorist organizations". This report takes the OECD technique one further:

In 1982, the International Trade Commission estimated losses from counterfeiting and piracy at $5.5 billion. In 1988, losses were estimated at $60 billion. In 1996, damage to the United States economy was estimated at $200 billion. Bank robberies, by contrast, generally involve less than $70 million per year, but seem to garner more public attention and law enforcement resources.

This astonishing rise from $5.5 billion in 1982 to $200 billion in 1996 says nothing about an increase in counterfeiting, of course: it simply shows an increase in the kind of numbers that pressure groups like the IACC are willing to put out with a straight face in an attempt to get attention. But let's unpack what the IACC is saying here: basically, take all the money that US banks lose in bank robberies in one year. Multiply that by a factor of 50, and that's what US businesses lose to counterfeiting in a week. We're told that the average bank robbery garners $4,587: you'd need to pull 21,000 of those an hour, eight hours a day, five days a week, to get up to $200 billion a year. Is anybody scratching their head here and saying "that can't be true"? Not that I can see.

The good thing about the IACC report is that it's footnote-heavy; the bad thing is that it seems to be entirely indiscrimiate in what it footnotes. There's lots of unsupported Congressional testimony, alongside press releases and magazine articles: it's very hard to find solid research amongst the fluff. But look hard enough and you will find one glimmer of hope: a CEBR report written for the European Commission entitled "Counting Counterfeits".

When you read the report, you realise that it's actually a report about how one might go about quantifying the extent of counterfeiting, were one so inclined, and how much such an exercise would cost. Nevertheless, the report does cite another CEBR report, which found that "the trade in counterfeits in just four industries reduces EU gross domestic product by €8 billion per annum and costs 17,000 jobs." Note that the €8 billion number, which seems much more solidly-grounded than most other numbers we've been looking at, never seems to get cited. Could that be because it's so much lower than the other numbers we've been looking at?

I very much wanted to read this report with the €8 billion figure. Finally, I thought, I could point to some real statistics. Even if €8 billion is a lot lower than $200 billion, it's still a very large number, and I could at least get a vague idea of how much that $200 billion number might be exaggerated.

The CEBR, however, didn't want to send me the report: it was privately commissioned by the ACG in London. So I phoned up the ACG, and they were kind enough to send me a copy. (It's not available online.) And indeed, at first glance, this was exactly the kind of thing I was looking for. It's full of lots of mathematical equations: here, for example is the page where the CEBR shows how it's calculating revenue loss.

The conclusions of the study are, therefore, quite precise. The clothing and footwear industry, it finds, loses €7.581 billion each year in revenues and €1.266 billion each year in profits; similar calculations are made for the perfumes and cosmetics industry, the toys and sports equipment industry, and the pharmaceutical industry.

And yet. Hidden behind all that four-significant-figures accuracy, what do we find?

Using data obtained on the likely scale of counterfeiting in each industry from the ACG for the UK and Association des Industries de Marque (AIM) for the remainder of the EU, the model was used to estimate...

In other words, the CEBR built a hugely complex model, and then what did they feed in? Data from the very people who were commissioning the reports. Every number in the report is derived from Table A1.1 in the report, which I reproduce in full below:

Table A1.1 ACG and AIM Estimates of the Proportion of Counterfeit Goods

That's it. Four numbers, representing estimates of the proportion of each industry which is represented by counterfeit goods. Four numbers, it's worth noting, which seem, on their face, to be absurdly high. This report was written five years ago; even today, with people ordering drugs off the internet in enormous quantities, no one estimates pharmaceutical counterfeiting in developed economies to be more than 1% of the market. Yet the CEBR, in estimating the cost of counterfeit pharmaceuticals to the EU economy, unblinkingly swallows an assertion that counterfeits make up fully 6% of the European market.

Nowhere in the report is there any hint of how these numbers were arrived at, and nowhere do the authors ask themselves whether the numbers are plausible. Nowhere are we given any plausible reason to believe – or, indeed, any reason to believe at all – that counterfeits account for fully 12% of the European market in toys and sports equipment. As they say in computing, GIGO: garbage in, garbage out. The €8 billion figure which looked so promising turns out to be based on estimated numbers which are simply implausible on their face, numbers which no one seems to want to defend.

What else is there? If we look across the pond, we find the UK Patent Office's Annual Enforcement Report. Scroll down to page 51, and you'll find a table showing the value of counterfeit goods seized by the Police Service of Northern Ireland. According to the PSNI, they seized 40,000 videos and DVDs worth £2,000,000, 80,500 music CDs and cassettes worth £1,800,000, and 5,000 pieces of computer software worth £450,000. Do you detect a pattern here? Apparently counterfeit CDs are worth on average £22.36 (over $40); counterfeit DVDs and videos are worth on average £50 (over $90); and counterfeit pieces of software are worth on average £90 (over $160). Clearly, no one ever pays $40 for a counterfeit CD or $90 for a bootleg DVD – nor, I hazard, are people likely to pay $550 for a counterfeit power tool. So the PSNI valuations are way out of whack – I can't imagine how they were arrived at. No wonder that, later in the report, we're told that "we require a settled classification system for products and more robust valuation procedures".

Indeed, if you keep on going to page 110, you'll find a table of seizures alongside what is more reasonably put as their "retail value" – which is clearly an estimate not of what the counterfeits are worth on the street, but rather of the retail value of the things being copied. Some things, however, aren't sold at retail, like rolls of counterfeit labels and tags. Those simply get valued at £10 each – that's per label, not per roll. These kind of valuations are great at generating enormous and scary headline numbers, but it's hard to see how they reflect any kind of reality.

So where does that leave us? Back where we started, with Erlinda Byrd at US Customs. She emailed me some numbers which can also be found here, showing that US customs seized $138,767,885 worth of counterfeit goods in 2004. A lot of people I spoke to thought that the counterfeiting statistics were extrapolated from these customs figures – which I do trust, or at least trust more than any other numbers out there.

According to another Customs spokesman, Barry Morrissey, Customs carefully examines the paperwork and manifests for every cargo shipment into the US. If there's anything suspicious, the shipments can be x-rayed (finding, perhaps, that what is meant to be golf clubs is actually polo shirts), or physically inspected. Nationally, about 8% of cargo shipments coming into the country are physically inspected. If we assume that customs seizures account for 8% of all counterfeits entering the country, then that would put the value of counterfeits successfully imported into the US each year at about $1.6 billion.

Compare that extrapolated $1.6 billion figure to the $287 billion that New York's Thompson values the counterfeit market at nationally: there's simply no way that home-grown counterfeits can make up the difference. It's entirely plausible that Thompson's estimate is two full orders of magnitude too large – you could take just 1% of his number and it could still be an exaggeration of the truth. Of course, we can't know for sure, because there simply aren't any remotely reliable counterfeiting statistics out there.

The underlying problem is that anti-counterfeiting groups know full well that they can put out any numbers they like with utter impunity: there aren't any pro-counterfeiting groups who are going to take issue with their numbers. Meanwhile, journalists and politicians are sheep who will uncritically parrot any number that is thrown at them.

More to the point, however, it's now too late to start generating fact-based, reliable numbers. The OECD, it is rumoured, may or may not be embarking on a survey trying to quantify the effects of counterfeiting. But if it does, and the numbers bear any relation to reality, they're hardly going to be trumpeted by groups such as the IACC and the ACG. All their rhetoric is tied up in the idea that counterfeiting is growing fast – something which I don't doubt. But world trade probably reached $7.9 trillion in 2004, and 7% of that is $475 billion. A lot of people have been using the 7% of world trade number as the basis for their calculations, which means that if the OECD comes up with a value for global trade in counterfeits which isn't in the hundreds of billions of dollars, it will look as though the number has fallen substantially. And that's a message which nobody wants to send.

The upshot is that a lie has circled the world hundreds of times before the truth has even found its boots, let alone thought about putting them on. The contest between the truth and the lie is so incredibly unequal that the truth will never win: it's now far to late for that. This isn't a sexy story, and it's hard to see how the lie is causing much, if any, harm. But just bear it in mind next time you see seemingly authoritative statistics being bandied around by journalists or politicians. There's a good chance they're utter bullshit.