Economics & Marginalia: November 4, 2022

Hi all,

I try to keep a generally optimistic worldview, but some weeks it really is difficult: we’ve got naked racism in the French Parliament, the UK running strong competition to the bottom of the barrel with its rhetoric on immigration and asylum seekers, from the backbenches, junior Ministers and the Home Secretary herself (remember this one? She took an L in a fight against tofu but bounced back, like a piece of morally-vacant flubber). And in Pakistan, Imran Khan got shot—which is very on brand, and whether or not you like him, once again depressing. I’m going to have to re-read Getting Better just to stop me falling into a depression. And take a short holiday, on which note there will be no links next week: I know they’ve only just come back from an extended break, but with the state of UK and global politics, can you blame me?

1. There are a few things that keep researchers up at night: the fear that your code has stopped running; your co-authors emailing at 1am asking for one more table; nightmares that Andrew Gelman has written a blog about your latest working paper. A new addition to the list: Data Colada running a blog about a method you’ve used in a paper you’re still writing (this happened to me recently, and I’ve only just worked up the courage to read it). This week, they take aim at meta-analyses: the practice of running an analysis on many previously published analyses to see what they collectively tell you about a problem. I’ve long been selectively sceptical of meta-analyses (selectively because I do think there are people and institutions who do them well, including my own colleague Dave Evans—I recall having a conversation in my previous job where we dismissed the prospect of commissioning one because we didn’t think we could get him to do it), and Data Colada pick up one of the problems I’ve always worried about and articulate one I’ve not thought about as much. The one I (and I think most people who do meta analyses) worry about most is that not all analyses are equal: some are better than others, and should probably be given more weight. There are ways of dealing with this problem, with varying levels of imperfection. The other is the concern that what we compare across multiple studies is often—sometimes subtly—incomparable. Just because two things are both cash transfers, for example, does not mean their effect sizes should be compared, because the design features may make them at some level very different interventions. Worth reading, even if you don’t share their fears to the same extent. Part one, and part two.
2. I have to admit: when I heard Liz Truss say that she was a “fighter, not a quitter”, the first thing that came to my mind was that bizarre Michael Jackson and Paul McCartney duet, The Girl is Mine (“I already told you Paul, I’m a lover not a fighter”). Tim Harford had rather more thoughtful impressions, specifically that quitting has a terrible reputation for something that is very often welfare enhancing. When asked for career advice by colleagues who are weighing up two options, my stock advice is ‘if you have a hard time deciding, then based on the information you have, there’s not much in it; picking either is likely to be about as good as the other’. I should probably update that with the rider that ‘if one of your options is the status quo, you should probably switch’. As Harford points out, between sunk cost bias and status quo bias, the path we’re already exerts a stronger gravitational pull on us than it should.
3. $1.90 is dead! Long live $2.15! If, like me, you’ve worked in this field long enough to remember when “a dollar a day” was shorthand for extreme poverty, the latest revisions to the international poverty line are very discombobulating, not least because they never settled on the rhetorically convenient $2 a day. The nitty gritty behind the poverty lines and the impact on our estimates of poverty are interesting and Our World in Data have a primer here (you may have strong objections to the method the Bank use, but it’s worth understanding).
4. I thought this personal, slightly roundabout reflection on the #metoo movement, and in particular the rationalising instinct among observers or third parties, by Lizzie Wolkovich, was excellent, and should be required reading for all of us.
5. They say never to kick a field of study when it’s down, so following the macro-centric Nobel in economics just awarded (a #MeToo link there too), I feel its ok to post this: most of the macroeconomic models we use are hot trash. They are so bad that even if you swap series’ around (so you enter inflation data in the investment bit, and vice versa) they do not substantially change their outcomes, despite being fed manifest nonsense. I have not dug into the details: if there are macroeconomists reading this who can convince me that the trash is, in fact, this criticism, please write to me. Genuinely: I want to know why this criticism is either wrong, meaningless or pointless.  
6. Superb stuff from Paul Blanchard, Doug Gollin and Martina Kirchberger on exactly how mobile smartphone users in Africa are (spoiler: very). Very interesting, with some amazing visualizations.
7. And lastly, last week I opened with Earnest Jackson and his song Inflation; this week I’ll (almost) close with it: part 2 of NPRs foray into the music business (transcript), and it’s completely mad business structure here, culminating with the song—I’m listening to it on Apple Music right now. And lastly, to leave you with a real brain-twister: a new way of measuring time, using—as far as I understand it, which is probably not very far at all—quantum excitement of atoms. A nice easy entry to the weekend.

Have a great weekend, everyone!

R