When my book The Undercover Economist was published five years ago, I would occasionally be asked whether I was in favour of sweatshops in developing countries. Not at all, I would reply. But I could see where the question was coming from, because I was certainly worried as to whether campaigning against them would do any good.

My argument had a logic that will be familiar to economists. Unless sweatshop workers are literally slaves, they are presumably working long hours in horrible conditions for low pay only because the alternative ways of making a living are worse.

When a well-meaning group of activists launches a campaign against sweatshop labour among, say, Nike suppliers in Indonesia, the obvious risk is that the sweatshops are closed, workers are tossed out on to the street, and the work is shifted to computerised sewing machines in Osaka. This is surely not the aim. The only alternative is economic growth: while it may be frustratingly slow, it finishes off sweatshops by producing far more attractive jobs.

But while the logic is straightforward enough, it is not watertight. A successful multinational may be profitable enough to be able to afford wage increases, and may prefer to take wage increases on the chin rather than move its business around. Economic growth itself can increase the demand for child labour as well as reducing the supply.

So I was intrigued to discover two new pieces of research addressing these questions. One is an article in March's American Economic Review, written by Ann Harrison of the University of California, Berkeley, and Jason Scorse of the Monterey Institute. Harrison and Scorse study data from Indonesia. In the 1990s, Indonesia was the focus of anti-sweatshop campaigns that persuaded the US government to put pressure on its Indonesian counterpart, and encouraged US consumers to boycott companies such as Nike. (An influential study in 1989 had found that Nike's suppliers paid lower wages than other companies in the export sector.) Harrison and Scorse look at the footwear, textile and clothing sectors and compare regions with lots of brand-name suppliers to regions with lower-profile businesses.

If my argument is correct, Harrison and Scorse would have found a slump in employment in export factories in the brand-name regions. There is little sign of this. Profits do fall, and so does investment. Some small plants closed. But few, if any, jobs seem to have been lost.

The minimum wage in Indonesia more than doubled between 1989 and 1996, after inflation, and this did depress employment. But there seemed to be no additional effect in the districts with lots of brand-name suppliers, despite the fact that wages in those regions outpaced wage increases elsewhere by almost a third.

The second paper was presented in draft form at the Royal Economic Society meeting in Guildford at the end of March. This research, by Nigar Hashimzade and Uma Kambhampati of the University of Reading, shows that economic growth – at least in the short-term – is not enough to reduce child labour. Complementary policies to strengthen schools and the incentive to attend them seem to be necessary.

Neither piece of research is the last word, and neither discounts the long-term effectiveness of economic growth in improving working conditions. But I am having to think again about anti-sweatshop campaigns. At least I am in good company. John Maynard Keynes is reported to have quipped, “When the facts change, I change my mind. What do you do, sir?”

Tim Harford's latest book is ‘Dear Undercover Economist' (Little, Brown)