No sooner than I speculated on what a self-quarantine period might look like than I received actual data – a friend of mine has had to self-isolate her entire family on their return from a holiday. In what will no doubt be shocking news for you all, it turns out locking two very small children into a small London dwelling with two increasingly stir-crazy adults turns out to be rather stressful. Blissfully ignorant of the demands of parenthood, I suggested she had 14 days of yoga, boxsets and depleting her accumulated hoard of wine (especially if you’re following the cricket); judging by the tone of her response, it’s probably a good thing she has another 10 days to cool off before she can escape and give a well-deserved thump. For the rest of you: wash your hands, stop hoarding things like face masks and take solace in the data, if you can. This week’s links include a slight dose of econometrics, so some self-isolation with wine may be recommended.
- Should we get the econometrics out first? Two things, both of which I would use to hammer home the message: no matter how clever your measurement or your study design, you must always think through the model you’re testing. First, Marc Bellemare, whose blog I have been enjoying more and more as I dig deeper and deeper into it, considers the SUTVA. Thought it sounds a little bit like non-violent resistance, it’s not (and indeed, SUTVA violations have been known to induce acts of violence against Stata); rather it’s an assumption that is made in econometrics when you want to infer causality. Essentially, the assumption (which is stronger than most people seem to think) requires that there are no spillovers from a change in the causal variable for me on either anyone else, or for me over time. This is actually a really difficult thing to prove to be true, so most of the time, we have to assume it or just argue that we think it holds. The problem is, very often, studies don’t manage to make this case very carefully, and thus report effects that are either stronger or not as strong as they should. And related: David McKenzie reports on a way to infer the long-term results of a short-term study. The best bit is at the end, where he describes a case where the approach may break down. Without really thinking through the ‘why’ of your result it can be difficult to even know the ‘what’ that you think you’re finding. I remind you of my advice from earlier.
- Something lighter, if not less important: a really nice piece from CGD setting out how they went about selecting programmes that have made a dent in learning programmes at large scale. One programme from DFID makes the cut, along with a bunch from USAID, with none run primarily by Governments. It’s well written and thoughtful and comes highly recommended.
- I really liked this paper from Stefano Caria. He digs into Ethiopia’s push towards industrialisation and documents that its cost advantage in labour is more than offset by lower productivity vs. important peers. He does a bunch of digging into this and argues that a big chunk of this may be driven by worse management practices, and especially worse labour management. Really good diagnostic work like this is rarer than it should be.
- VoxEU have produced a new e-book looking at the position of women in economics, including minority women (summary here). Meanwhile, Marginal Revolution have produced a video about Janet Yellen, part of their own women in economics series. Bonus gender link: also from VoxEU, research that shows that men benefit more from larger venture capital boards than women do. They find the result is not obviously (first order) discrimination, but that men seem much more likely to reach out to and contact members of the VC board than women. Once again, the findings suggest that gaps stem from deep roots.
- I’ve said in other settings that the mystery of this world is not that there is so much migration, but rather that there is so little. Germany provides an interesting case study here. There remain substantial wage gaps between the East and West, even once you account for the cost and quality of living. Why do East Germans not simply move over until wages are equalised? If I were speculating on this blindly, I might have guessed that productivity is just higher in the West, but absorptive capacity not sufficient to eliminate this difference through arbitrage. This piece suggests it’s much simpler than that: people like home, and moving away from it requires a premium.
- If you’ve been following Coronavirus in the UK, you cannot have escaped Chris Whitty, illustrious alumnus of DFID. He has very much been at the public face of advice on the new virus, and the evidence suggests this is rather a good thing. In the US, at least, trust in politicians is so low that any intervention that makes use of them, even well-intentioned, is likely to have the perverse effect of leading more people to ignore their advice.
- And lastly, this week has been a bit low on frivolity for me, but I did rather love these tweets pointing out exactly how unsanitary most pop music wants us to be. Sweet Caroline: vector of disease.
Have a great weekend, everyone!