Let's track a statistic. On the wikipedia page for fair trade there is a claim that "According to Adam Smith Institute estimates, only 10% of the increase in price over a similar non fair trade product ends up in the hands of producers." So I want to see if that's true.

Searching the website of the institute, I find their page on unfair trade on which the claim that:

Only 10 percent of the premium they pay for their Fairtrade products actually gets to the producer. The rest goes to people further along the retail chain.

As this is the web cover page for a pdf report of theirs, lets look at the pdf for the source.

The pdf, Unfair Trade (Sidwell 2008) claims in its Executive Summary that:

Just 10% of the premium consumers pay for Fairtrade actually goes to the producer. Retailers pocket the rest.

On page 11, however, we get more detail:

Just 10% of the premium paid for Fairtrade coffee reaches the producer.
While it may still be true that the money reaching a farming cooperative via Fairtrade will be significant to them, Fairtrade is an inefficient way to transfer money, with 90% of the premium paid going to retailers. Given that the consumer very likely pays the large Fairtrade premium on the understanding that it is a direct charitable contribution, they would be willing to send far more to poor farmers than farmers receive through the Fairtrade certification process. The Fairtrade tendency to discourage individuals from donating directly to charities arguably draws them away from the most efficient way to give, in favour of Fairtrade, losing the producers money.

The only cited portion is the italicised claim at the top. The rest appears to be just elaboration added by the paper writer. So let's keep following the statistic.

The footnote refers to a 2006 Economist voting with your trolley which is unfortunately paywalled. It looks like someone has put up another copy for some course, though, so we can look up the source there. On page 3 of the report we have:

But perhaps the most cogent objection to Fairtrade is that it is an inefficient way to get money to poor producers. Retailers add their own enormous mark-ups to Fairtrade products and mislead consumers into thinking that all of the premium they are paying is passed on. Mr Harford calculates that only 10% of the premium paid for Fairtrade coffee in a coffee bar trickles down to the producer. Fairtrade coffee, like the organic produce sold in supermarkets, is used by retailers as a means of identifying price-insensitive consumers who will pay more, he says.

The Harford refered to is, as the article says earlier, the author The Undercover Economist, a 2005 book. It's not clear from 'voting with your trolley' whether the statistic comes from an interview with Harford or his book.

A bit more searching turns up an extract from the undercover economist that appears to have the passage where this statistic was calculated:

Customers who wished to support third-world farmers -- and such customers are apparently not uncommon in London -- were charged an extra 10p. They may have believed that the 10p went to the struggling coffee farmer. Almost none of it did.

Cafedirect paid farmers a premium of between 40p and 55p per pound of coffee, and that premium was reflected in the price they charged to Costa. That relatively small premium can nearly double the income of a farmer in Guatemala, where the average income is less that $2,000 a year. But since the typical cappuccino is made with just a quarter-ounce of coffee beans, the premium paid to the farmer should translate into a cost increase of less than a penny a cup.

Of the extra money that Costa charged, more than 90 per cent did not reach the farmer. Cafedirect did not benefit, so unless using the fair trade coffee somehow increased s costs hugely, the money was being added to profits. The truth is that fair trade coffee wholesalers could pay two, three or sometimes four times the market price for coffee in the developing world without adding anything noticeable to the production cost of a cappuccino. Because coffee beans make up such a small proportion of that cost customers might have concluded that the extra 10p was to cover the cost of the fair trade coffee, but they would have been wrong. A certain Undercover Economist made some inquiries and found that Costa worked out that the whole business gave the wrong impression, and at the end of 2004 began to offer fair trade coffee on request, without a price premium.

This means that the most grand claim, that "only 10 percent of the premium they pay for their Fairtrade products actually gets to the producer", traces back to a simple claim that this was true for coffee at one chain in london, under a pricing scheme given up in 2004 after acknowleging that it was misleading. If anything, the original source is a good argument for fair trade coffee: if it would only slightly increase their costs to switch to fair trade, that's a much better argument for doing so than if it would up their costs by 10p a cup.

This also means that the statistic is almost useless for what Julia and I are interested in. If we want to buy chocolate chips grown ethically, what does it matter to us that at an already overpriced (3L/cup) coffee chain they overcharged for fair trade coffee? Yet with the way the statistic presented as applying to all fair trade goods, as if they did a broad survey, we might have concluded it was applicable. I'm going to write to the adam smith institute and ask if they have some other source for their claim that applies to goods other than coffee purchased by the cup at expensive london coffee shops.

Update 2012-05-30: This is old, but the Adam Smith Institute never responded.