IPA’s weekly links

by
Jeff Mosenkis (IPA)
My articles My siteAbout meMy booksMy videos
Follow on:TwitterLinkedIN

Guest post by Jeff Mosenkis of Innovations for Poverty Action.

A quick note, my posting frequency has slowed down in 2021, thanks for sticking with it. One reason has been that I’ve been co-authoring another set of links with my brilliant IPA colleagues, Luciana Debenedetti & Rachel Strohm, every other week focused on new research on COVID and social protection (this week’s is here). Among other, I think I also hit what I now realize was a quarantine burnout. If it’s helpful to anybody else, this article which colleagues shared with me helped me realize it was widespread phenomenon, and this one from Wharton’s Adam Grant, had some suggestions for combating it. In any case, I’ll be continuing with the links (even if at a slightly slower pace sometimes), and thanks for reading! 

• I haven’t mentioned it in a while, but I always owe a big thanks to my colleague Cara Vu, the busiest person I know at IPA in the U.S., who’s been editing these links and saves me from embarrassing mistakes in every post.
• A new study finds 71-77% efficacy of a new Malaria vaccine in a phase 2 trial with 450 children in Burkina Faso. The vaccine will now proceed to a larger trial.
• You may recall last fall there was an AER article (ungated version here) by Abel Brodeur, Nikolai Cook & Anthony Heyes estimating p-hacking prevalence across different methods in econ, using 21,000 hypotheses tests in 25 journals. They concluded that some methods (looking at you, IV) were more prone to p-hacking than others, but also that overall econ compared favorably to other disciplines. I didn’t see at the time, but one of the team that coined the term p-hacking and started the replicability/credibility movement in psych and was one of the reviewers of the article, Uri Simonsohn, who was one of the reviewers of the article, had a detailed post about why he disagrees with their conclusion.
• In this cool paper (from November) Karthik Muralidharan & Abhijeet Singh, looked at a school improvement program in India that sounded good (implementing management best practices), but an RCT showed had no effect on learning. Despite that, it was scaled up to 600,000 schools, and still appeared to have no effect on learning. What it did do though, they found in qualitative interviews, was give the appearance that schools were innovating and improving, and the authors suggest that may be where incentives really lie.
• This seems to parallel this accusation by UN aid coordinator Mark Lowcock, that aid agencies are failing because they’re not really listening to what the people suffering really need. They can send what they want/have handy/theorize would be good, even if refugees end up trying to sell what they’ve been sent, because of misaligned incentives:
  

  “Ultimately, organisations or decision-makers can choose to listen to people and be responsive, or they can choose not to. There are no real consequences for the choice they make. There are weak incentives to push them in the right direction.”

It’s unclear if he also means the agency he was in charge of, but Ilya Gridneff calls this pattern the development sector’s “self-licking ice cream cone”
Lowcock proposes an independent body to listen to what humanitarian beneficiaries actually need and grade aid agencies on whether they’re delivering it.

• A nicer ed story though, from Aker & Ksoll; a simple phone call to adults enrolled in a literacy program in Niger (weekly calls to the students and teachers in the class, along with village chiefs), increased learning and how long those learning gains lasted. They think the calls encouraged the teachers to be more prepared for class and reduced student dropouts.
• A nice look at how Vietnam kept COVID death rates so low, and also grew its economy in 2020 raises some uncomfortable points:
  • Strict border closures (particularly early with China), seemed to work, defying the mantra that viruses know no borders (one source points out that was an assumed truth not really tested). Karen Grépin says those restrictions work well early, when there are few cases and the response seems like overkill.
  • The single party communist government harnessed its existing surveillance system for disease monitoring and tracking. (Though countries like Taiwan and New Zealand also accomplished containment through different means.)
  • It’s possible that their cyber spying on China got them better data about the disease and earlier than was publicly available
• How a bad social science study ended up making the Vietnam war worse.
• An interesting NPR story on a new paper about disease surveillance that upends a popular myth. Many viruses like COVID-19 start in animals and jump to humans, but it’s not a one-time deal. The virus usually jumps several times, often over decades, before a version evolves that can spread from human-to human. (The other versions might be catchable from the animal and can make a person sick but don’t spread in the human population). Surveilling groups of sick people, which they’re doing in Malaysia, can find those sick with an early new virus, potentially before it’s transmissible between humans.
• The BBC “People Fixing the World” podcast had a couple of episodes on research-driven impact: satellite data for identifying needy cash beneficiaries in Togo (Apple) with Josh Blumenstock, and a nice story on World Bicycle Relief (Apple) with a lot of on the ground reporting from Zambia and Ana Garcia Hernandez discussing the RCT of giving bicycles to schoolgirls in rural areas. (For more, here are study summaries on cash & bikes.)
• Jobs:
• Funding available from IFPRI ($25,000-$200,000) for research on cash transfers and intimate partner violence in low- and middle-income countries (deadline July 9th)