More sweatshops for Africa?

img_0536After six long years, my randomized trial of factory jobs is at last public. Here is today’s coverage in Vox:

In the past several decades, manufacturing jobs have fled the developed world for the developing world. Obviously, that’s profoundly reshaped the economies of developing countries like China and Bangladesh. But what does that mean for the ordinary people that are doing the work — often for incredibly low wages?

Answering this question can be tricky. Large-scale data — like a nation’s poverty rate or GDP — can help us give a general sense of trade’s effect on growth and the poor. The problem is it can often be tough to figure out what low-wage manufacturing, specifically, adds to a country’s economy. It’s even harder to drill down and measure the impact on those employed in the factories.

Enter economists Chris Blattman of the University of Chicago and Stefan Dercon of Oxford University. They came up with an interesting way of answering this question: Run a random, controlled experiment.

Normally, economists can’t do stuff like this: You can’t exactly run a lab test on an economy. But Blattman and Dercon convinced five companies in Ethiopia to hire people at random from a group of consenting participants, and then tracked the effects on their incomes and health. That way, you could pretty clearly figure out the effects of taking a low-wage manufacturing job on actual people.

The Vox article is a nice summary. So is the paper summary at IPA. Here is the academic paper.

To sum up the findings I’d say this:

  • Most people who applied for these factory jobs didn’t like them or intend to stay, rather the jobs were low paid and unpleasant and used as a safety net of sorts, while people looked for other entrepreneurial activities or less difficult wage work
  • Taking a factory job didn’t give you higher or more steadier incomes, because the firms gave steadier hours but at significantly lower wages that people’s other opportunities (it’s a relatively frictionless, competitive market)
  • But the health risks of industrial work were high (think chemicals and dirty air, for example) and there’s evidence that serious health problems doubled if you took the factory job: Chances of a chronic health issue went up 1 percentage point for every month in an industrial firm!

So these Ethiopian factory jobs offered a risky safety net, mostly for poor young women.

When you gave them $300 cash, meanwhile, they started a small business and earnings went up by a third (I couldn’t help myself–I had to do this comparison).

The last line of the Vox article illustrates why I love their coverage. Can you imagine many newspapers saying “it’s complicated” when writing about this subject, especially on something in Africa?

So perhaps the most fundamental takeaway is that we need to have a more nuanced picture of globalization’s effect on the global poor. Instead of thinking in binary terms, we need to separate out the ways globalization has benefited the poor versus the way it hurts them.

Something as complicated as globalization is never going to be just good or just bad. We need to divide the good and the bad, and figure out how to address the latter without eliminating the former.

8 thoughts on “More sweatshops for Africa?

  1. Cool paper! The result surprised me. It supports econ-101 style intuition behind the simplest labor models — that the benefits of opening a factory are all in the GE effects.

    I worry that the press and public will miss the point, though. Consider the Vox headline: “Is globalization bad for the global poor? This study ran an experiment to find out.” Your experiment measures something very interesting, but it is not whether globalization is bad for the poor (unless I’ve misunderstood something). The Vox article discusses this point but it is buried as caveats near the end.

  2. Interesting experiment. Were most people aware of the health risks when they started working at the factory?

  3. We don’t really know what people believed at the outset, but we know what they believed at endline, and working in a factory job didn’t increase people’s perceptions of health risks there. This suggests that they knew to some degree what they were getting into. Its also possible that everyone underestimates these risks, now or later. If they have a respiratory problem a year after quitting, it’s hard for an individual to know if the factory was responsible. Whereas our experiment can tell us exactly that.

  4. There may be a more cynical way to understand their pricing. This is general speculation on my part from other areas I know, since I’m unfamiliar with Ethiopia specifically.

    However, young women who desperately need money rarely have good options. They frequently involve prostitution or abusive work dynamics; let’s call the benefit from these jobs r. A quick google search shows prostitution is a problem in Ethiopia and often receives $1-2 a day. (e.g. ” In the capital Addis Ababa, around 130,000 girls support themselves by selling their bodies.” http://onlineethiopia.net/2016/02/ethiopia-prostitution-for-just-2/)

    This is in a country with 2.4% of the population having HIV/AIDS.

    If these are the main workers at these factories, they only have to offer the utility of a r+e, which is awful, has poor health outcomes, but should likely have better health outcomes than prostitution.

    I skimmed the paper and didn’t see this addressed. However, the researchers are smart and this isn’t my field, so probably I’m missing some context. The devil is probably in the details of their counterfactual group.

  5. I work in Ethiopia and I would be very happy if everyone could start a micro-business and earn more than they can at a factory. However, isn’t there a saturation point for micro-businesses and is everyone a successful entrepreneur? Don’t 80-90% of people just need jobs?

  6. Would you have published the paper if the entrepreneurial alternative failed? Seriously, you need to ask yourself that question. Would you have put your name on a paper which gave empirical data that the entrepreneurs squandered their funds and that the sweatshops were the best alternative?

  7. So to answer some more of the questions above:
    – Actually, ALL our funding proposals before the study suggested that we expected the sweatshops to pay higher and more steady wages than people’s bad alternatives in the formal sector. So this was actually the paper I expected to write, and would have done so with gusto. But in this case I was wrong. There are places where I think sweatshops are better than most people’s alternatives, and indeed I would characterize that as the majority view in economics. It was the advocacy do-gooders I expected to skewer. Egg on my face.

    – The women who were applying to these jobs were largely unmarried young women living with their families. At least at the time. They had lots of good outside employment options, which we discuss (albeit briefly). They probably would not have told us if they were engaging in transactional sex, but based on my familiarity with the demographic, I highly doubt it. Actually, I think prostitution is quite rare in much of Africa, whereas more casual transactional sex is borderline routine in some places. Where I worked once, near a highway in western Kenya, a study suggested no one was a prostitute in terms of an occupation, but more than a third engaged in occasional transactional sex, mainly to get through bad periods.

    – Last, I don’t think you could scale these effects. Everyone can’t start a business, though you could have many more factories. That size of change is what we would say would change the general equilibrium, and we could expect very different effects. We were investigating changes on the margin. Besides, we’re not recommending people get grants instead of jobs. That was just our way of simulating how things might go differently for factory jobs if people had better outside options.

  8. Something as complicated as globalization is never going to be just good or just bad. We need to divide the good and the bad, and figure out how to address the latter without eliminating the former.

    golu dolls
    golu dolls