The OECD counterfeiting report is now public, and only 18 months or so behind schedule! As you might recall, there was some pushback from OECD member states when they realized the report would peg global counterfeiting activity at only $200 billion per year, which is much lower than previous estimates (which were basically pulled out of thin air).

As a result, there's no indication in the executive summary that the magnitude of international trade might be substantially less than $200 billion: that's the only number one finds. But if you actually plough through the methodology in the 158-page Overall Assessment, it turns out that the real number is probably much, much lower. In fact, a quick-and-dirty back-of-the-envelope calculation puts the amount of international trade in counterfeits as low as $5 billion.

The OECD report is designed to make the counterfeiting problem seem as big as possible. Here's the language in the summary:

Quantitative analysis carried out by the OECD indicates that the volume of tangible counterfeit and pirated products in international trade could be up to USD 200 billion. This figure does not, however, include counterfeit and pirated products that are produced and consumed domestically, nor does it include the significant volume of pirated digital products that are being distributed via the Internet. If these items were added, the total magnitude of counterfeiting and piracy worldwide could well be several hundred billion dollars more.

The question of pirated digital products will be addressed in Phase 2 of the report, and no one has a clue when that might come out. But if we stick to the question of international trade in counterfeit goods, the "could be up to $200 billion" formulation is weak indeed, especially since the report itself concedes that any credible estimate would be much smaller. But will the OECD tell us what its best guess as to the magnitude of the problem is? No: "the approach taken by this exercise is to establish a credible 'ceiling', while addressing the many information deficiencies surrounding the subject," we're told.

Yes, really. All this work went not to estimating the magnitude of the problem, but only to establishing an upper bound for the magnitude of the problem. How pathetic.

Still, we can use the OECD's own data to try to come up with a realistic estimate. The OECD sent out a survey to try to find out how many counterfeit goods were seized at customs globally in 2005. The total for 35 economies came to $769 million, or 0.01% of total imports for the economies concerned.

Now the OECD carefully never reveals what percentage of imports are inspected at customs. But in the US, the number is about 8%. Certainly the number is big enough that counterfeiters spend quite a lot of time and ingenuity trying to avoid customs inspections:

Smugglers of counterfeit cigarettes produced in the Far East tried to deceive customs officials in the United Kingdom by concealing the cigarettes in a container of rice noodles and by hiding the cigarettes in consignments of pottery and ceramic items...
Traders are constantly altering shipping routes to avoid detection.

One imagines that they wouldn't bother if customs seized only 0.5% of counterfeit imports – but it turns out that in order to make the OECD's numbers work, that's the number which would have to be true. Because the OECD doesn't do the obvious calculation, which is to extrapolate from seizures and inspection rates what the total magnitude of the problem is. The reason they don't do that calculation is obvious: it comes up with a number which is far too small.

Let's start with that $769 million figure. Now it includes numbers such as $66 million from South Africa, which uses "legitimate item value" as the basis for calculating the value of its seizures. In other words, if South African customs seizes a fake Prada handbag which would sell on the street for $5, and the real Prada handbag costs $5,000, then South Africa reports that it has seized $5,000 of goods.

So to get a better idea of what the customs seizures are actually worth, let's chop the $769 million figure in half: call it $385 million. Now, customs doesn't inspect shipments at random: they carefully target imports which they think might be counterfeit. At the same time, however, they might sometimes inspect counterfeit goods without realizing that they're counterfeit, and let them through unseized despite their being inspected. Let's assume that these two factors cancel each other out, and that 8% of counterfeit imports are seized at customs. If 8% of counterfeit imports are worth $385 million, then the total value of counterfeit trade is $4.8 billion. A far cry from $200 billion, to be sure.

So how did the OECD arrive at its $200 billion number? Well, for one thing it first declared that arriving at any number would be all but impossible:

With respect to magnitude, the study notes that promising work has been done in a number of sectors to measure the extent of counterfeiting and piracy, but that much more can and should be done... To date, no rigorous quantitative analysis has been carried out to measure the overall magnitude of counterfeiting and piracy.

That's right, the OECD's 158-page report, complete with very complicated mathematical equations, still doesn't count as a "rigorous quantitative analysis". Actually, they're right, it doesn't. Because after sending out surveys to customs authorities globally, the OECD suddenly changes its mind and decides that it's not going use that $769 million figure after all. Instead, the only use they're going to make of that data is to get an idea of the relative propensities of different types of goods to be counterfeited: clothes are counterfeited more than cigarettes, for example. As for the absolute amount of counterfeiting going on, they – well, frankly, they pull it out of thin air, again. Here's what they say that they do:

This approach was taken by establishing an upper fix-point limit; that is, an upper limit of counterfeit exports (in percentages) of those products and economies where counterfeiting and piracy were most pronounced.

And when they say "establishing", they mean "guessing". Their best guess for this fix-point limit is 5%, but they give no indication where they get that number from. And then as soon as they alight on the 5% number, they immediately decide that it's too low, so they double it:

With respect to the implications given by the model, 5% could be a reasonable fix-point for establishing the magnitude for trade in counterfeit and pirated goods; however, it could be higher. As the approach taken by this exercise is to establish a credible “ceiling”, while addressing the many information deficiencies surrounding the subject, a fix-point of 10% was therefore considered.

I've been looking at counterfeiting statistics for some time now, and I have no idea where or how the 5% number comes from; it seems very high to me. But there's no reason to simply double it to 10% just for the sake of being sure you've hit your "ceiling".

Now, what happens if you use the 10% number? You get a result of $100 billion, which clearly is too low. So the OECD doubles it again:

Given a fix-point of 10%, the overall ceiling of counterfeit and pirated goods in world trade would in 2005 amount to about US$ 100 billion. This estimate is nevertheless derived from GTRIC at its baseline and does not address the statistical variability of the index. While taking this statistical uncertainty into account implies that the ceiling could be lower, it also suggests that the ceiling of counterfeit and pirated goods in international trade in 2005 could have been as high as US$ 200 billion.

And that, ladies and gentlemen, is where the $200 billion number comes from. You guess what the maximum amount of counterfeiting is in the countries where it's most prevalent, being careful to use no empirical data in the process. You then double that number, double it again, and apply it to the amount of world trade: presto, you've got $200 billion.

In other words, just like previous estimates, this one is top-down: it takes the total amount of world trade, and then it assumes that some percentage of that trade must be counterfeit. If you take a large enough percentage of a very large number, then you're bound to end up with something shockingly enormous, like $200 billion. But if you try to do a bottom-up calculation, and extrapolate the total amount of counterfeiting by actually measuring the amount of counterfeit goods which are found or seized, then you can't get to anywhere near that figure.

Here's a revealing chart in the OECD report (click for a bigger version):

In order to get to the $200 billion level, the customs interception rate would have to be 0.5%: as the chart says, 1 in 200 counterfeit shipments would be intercepted. Basically, the OECD is saying that its own customs agents are complete and utter morons, who intercept vastly fewer counterfeit shipments than they actually inspect. Look how that line drops off sharply when you get to much more realistic interception rates around the 4% to 5% level – and note how the chart doesn't even extend as far as the 8% level, which is the actual inspection rate in the US.

The OECD's interception rate of 0.5% just isn't plausible, and I wouldn't be at all surprised to find out that the reason this report took so long to be made public was that some very powerful lobbies didn't want it made obvious that counterfeiting is a much smaller problem than they would have you believe. So the headline number in the report was made as big as possible: $200 billion. But the best estimate in the report is just a quarter of that number, $50 billion. And even $50 billion could well be an order of magnitude too big.