Hi all,
Well, that was a downer. After switching leaders over the summer, and an opening full of promise and dynamism they’ve fallen into the same old patterns, beaten badly and looking ineffectual in the process—dashing my hopes and prompting mildly panicked contingency planning in California. I am, of course, referring to the Lakers 1-4 run after a glorious, golden 3-0 start against tough competition, led by the young, dynamic new coach JJ Redick. What else would I be talking about? Oh yeah, the other thing. Not being an expert on American politics, I’m going to keep my half-baked political theories to myself (and a few unfortunates on whatsapp group chats), and I recommend many others do so too. Two adjacent observations: even at the time, I thought people piling into Nate Silver because he can be a bit abrasive (and who wouldn’t be under such constant and usually poorly-conceived criticism) and confidently declaring that he was a hack who knew nothing about elections or polling was spectacularly ill-conceived. And secondly, the US has an economy that most economists consider to be really very strong. There is much work to do in understanding exactly why perceptions and expert consensus so diverge. It’s not enough to point at economic indicators and demand that mood follow the direction of GDP or inflation measures. The answer may be about partisanship, or about how people experience level vs change effects, or about inadequacies of the summary measures we are using, or simply about the power of narratives; all of these, including the last, are things that economists can and should think about carefully.
-
There is, of course, going to be a great temptation to blame polling or forecast models for their inability to predict the scale of the Trump victory or Republican swing in the US. But the polls were incredibly tight; the forecast models transparent that very small shifts in polling averages (and thus very small errors in the polls) would have large effects on the outcome given this. Our ability to compute and interpret these fairly subtle points in real time, when dealing with the emotion of a highly charged election and its accompanying online arguments, was very poor. Nate (and the central thesis of his new book, which is largely about the ability of some people to do this computation and interpretation under non-ideal circumstances better than others) comes out of this looking pretty good. Andrew Gelman’s sober and careful analysis, both before and after the election does too.
-
Last week I led with Ken Opalo discussing what the election might mean for Africa; of particular concern for him was a relationship that went beyond simple ‘values evangelism’ and into a substantive relationship of partners focused on the policy issues that matter most for welfare and economic development in Africa. In a very similar vein, he considers China-Africa relations this week, both in terms of what African governments can learn from China’s experience (and the wrong lessons they too often draw) and in terms of their actual relationships.
-
November doesn’t just bring elections, cold weather and poor cheer (and my birthday); it also brings one of the great annual traditions of the development economics community, the Development Impact Job Market Papers series. Every years this series unearths brilliant papers and scholars; a number whom I now follow closely first came to my attention through their blogs on the forum. This year, they kick off with a superb piece by Michelle Rao at the LSE. Her paper looks at the link between the generation of new evidence about the effectiveness of a policy (in this case conditional cash transfers) and spending on it by funders and governments. The results are sobering: there is virtually no link between the two, either at the level of the individual study or as studies accrete into a body of evidence. The one situation in which the link is statistically distinguishable from zero is when the studies are made available to policymakers very quickly after the events being evaluated (though the definition of ‘very quickly’ here is still measured in years). This is excellent work on an important topic. One thing I wonder about, though, is the extent to which the findings of research are communicated throughout the process of evaluation, and so incorporated into policymaker decisions before or unrelatedly to the timing of formal release of a paper. Nevertheless, another excellent addition to a growing and improving literature.
-
Alex Miller (ex of Development Initiatives, which is sadly no longer) has revived his blog to release work that could not be completed under DI’s banner but which he doesn’t want to abandon yet. This blog, on using a natural language processing model to classify World Bank projects according to how much of it was devoted to climate adaptation, is really impressive.
-
This week in my priors being challenged: I have often assumed that the quite incredible lack of competitiveness in a lot of African manufacturing was down to structural factors, including cartel-like behaviour and capture of lucrative protected markets by politically-connected firms. That may well be part of the story, but in this impressive new work by Fabrizio Leone, Rocco Macchiavello and Tristan Reed, they show that a substantial part of the story may simply be market size. The minimum efficient scale for production in cement, at least, is simply rather large: most African markets remain rather small, given relatively low construction rates. But as those rates have increased the price of cement has fallen rather dramatically. This is really important: if they are right, then the best strategy to drive down costs may not be deregulation, but market integration—and by making markets larger, creating more space for competition between producers.
-
This one hit close to home: Tim Harford on whether wearing a fitness watch helps people lose weight, and what the exact question randomized control trials investigating this are asking. He makes the point that an RCT asks: if I gave a person a fitness watch, would they get fitter?; but the relevant question for most people is ‘if I want to get fitter, will a fitness watch help me?’ They are not exactly the same. The reason this one struck a chord with me is that I got a fitness watch for Christmas a couple of years ago (by request), because I wanted to track my outdoor running more effectively. But interestingly, I found that while I obsessed over the data a lot, I didn’t actually lose much weight or become much fitter. The constraining factors for me were not the data, but injuries and frequent toddler illnesses keeping me from exercise. I recently stopped wearing the fitness watch (it was making my skin look like an ancient parchment) and started swimming as well as running and I’m much, much fitter than I have been for a while. The question the RCTs on fitness watches ask is if adding a fitness watch at random might induce people to do more exercise; the question relevant to me was if giving me a fitness watch would help me achieve my (already-decided) goal of doing more exercise. Being very precise about what the exact question you’re asking, and what the exact question your evidence answers, really matters.
-
Quincy Jones passed away this week. Even if you don’t realise it, Quincy probably shaped some of your musical reference points. He was a genius, as the stories about him attest. He helped shape Ray Charles’s career, and was behind that incredible three-album sequence from Michael Jackson, but he also did some of the best TV and movie music out there, like the iconic intro to Sanford and Son (this was a different time: Miles Davis was making live arrangements for Louis Malle films). I feel like the last few years have taken far too many of my musical heroes away: Mark Lanegan, David Bowie, Prince [googling it, I realise they both died almost a decade ago, which tells you something about how badly their loss hit me], DMX… someone keep an eye on Nick Cave please.
Have a great weekend, everyone!
R
Disclaimer
CGD blog posts reflect the views of the authors, drawing on prior research and experience in their areas of expertise. CGD is a nonpartisan, independent organization and does not take institutional positions.
