Guest post by Jeff Mosenkis of Innovations for Poverty Action.

Pardon our remodeling!

Chris is migrating his site to new servers, our apologies for any recent downtime. FYI that we’re still getting SSL worked out so your browser might warn you the website is insecure (just offering http not https). I don’t really know what that means for a blog, but just in case, don’t put your bank account number in the comments section until it’s worked out?

In the meantime I’m reposting last week’s links that got lost in the transition.

• By now you’ve probably seen the big Africa headlines – Kenya Airways will start flying directly to the U.S.! (h/t Anne Healy)
• And The Daily Show will have an Africa correspondent, South African comedian Loyisa Madinga! Great week for U.S.-Africa relations, right?
• An addendum/correction to last week’s link about an op-ed on North Carolina rating below some autocracies in election integrity. Andrew Gelman looked into the underlying methodology and thinks it doesn’t hold up (the researchers defended their methodology, even if it yielded strange results).
  • (By coincidence this week, a court threw out North Carolina’s gerrymandered electoral map. Last year the Supreme court invalidated two NC districts for the race-based gerrymandering.)
• Gelman also talks about his response to Deaton and Cartwright’s critique of randomized controlled trials in econ. He suggests that people put too much stock in them because of the empirical methodology despite other weaknesses that could affect the conclusions. But honestly I don’t think there’s much disagreement here. Here’s Rachel Glennerster:

“Economists have now settled down into RCTs as just one tool,” Glennerster told Devex. Among academics, the kind J-PAL works to connect with the world’s policymakers, she said, “the trend toward using RCTs is simply part of this bigger movement in economics to care more about where we can really pin down what is causing what we see.”

• Any critique I’ve seen of RCTs as a method apply in one way or another to any empirical study – the results and conclusions are limited, or as Pam Jakiela put it:

Nice insight by Deaton and Cartwright, but for some reason they keep spelling "study" as R-C-T. https://t.co/vBVRFj4M09

— Pamela Jakiela (@PJakiela) November 11, 2016

• Chris Blattman presented some RCT results to Deaton among others at Princeton soon after the paper first circulated, and said they didn’t disagree on much. Deaton seemed more concerned with people just putting too much stock in RCTs because of the method.
  • I’d add that I’ve never met an economist or policy professional in the development RCT world these days who’d copy and paste a program based on one RCT. I think the way the world is going is more like Mushfiq Mobarak’s No Lean Season – slow-scale up, with multiple years of testing as a program expands, looking for externalities or changes in effects at larger scale, and very careful expansion into other locales where the same intervention may work differently.
  • Or multiple simultaneous evaluations across countries like the Ultra Poor Graduation model or Metaketa Initiative.
  • Most RCTs I see these days are part of a body of research all trying to get at the mechanism behind a larger question and test solutions, and any important insight is usually based on a body of research.
• Tim Ogden has some nice papers from AEA in his newsletter, the faiV
• After SpaceX (South African Elon Musk’s company) appeared to have lost a reported billion-dollar U.S. spy satellite code-named “Zuma,” the value of the South African Rand on currency markets briefly spiked:

News-reading algorithmic traders may have been further confused by reports on the wires of a US congressional aide saying that Zuma was lost.