IPA’s weekly links

(From the video at the end of the post)

Guest post by Jeff Mosenkis of Innovations for Poverty Action

  • Oxfam releases a report around the same time as Davos every year on who owns what portion of global wealth. Their spin on it is designed to make headlines, but Dylan Matthews explains why it’s really hard to measure.
  • Also in Vox, Stephanie Wykstra provides a nice plain-language summary of what the research says about microloans.
  • A very cool very cross randomized experiment (more than 50 treatment combinations!) from Busara on sending SMS surveys in Nairobi and response rates.
  • About sixty percent of people in sub-Saharan Africa live in rural areas and many of them are dependent on farming, but the low amount of food those farmers get out of the ground compared to other regions of the world has been a particular puzzle. A all-star team in VoxDev summarizes what we know about improving agricultural output through experiments with ag extension workers trying to help farmers be more productive.
  • But not so fast, economists – Gollin & Udry have a new paper using panel data from Uganda and Tanzania suggesting that even measuring how much farmers are producing (or can) is harder than we think:

We find that measurement error and [unobserved] heterogeneity [in inputs] together account for a large fraction – as much as ninety percent — of the dispersion in measured productivity. In contrast to some previous estimates, we suggest that the potential for efficiency gains through reallocation of land across farms and farmers may be relatively modest.

  • An all-star team of behavioral scientists, led by Katy Milkman and Angela Duckworth, are running a nationwide experiment on sticking to exercise regimens. Members of the 24-Hour Fitness chain can join here by Jan 31.
  • I’m usually a bit tempered in excitement about “nudge” style research – often the behavior change is statistically significant but small or short-lived (in my experience). The exception seems changing the default option – that can be huge, and a new meta-analysis quantifies the effects over 58 studies and find they can be substantial, but not all defaults are equal:

Our analysis reveals two factors that partially account for the variability in defaults’ effectiveness. First, we find that defaults in consumer domains are more effective and in environmental domains are less effective. Second, we find that defaults are more effective when they operate through endorsement (defaults that are seen as conveying what the choice architect thinks the decision-maker should do) or endowment (defaults that are seen as reflecting the status quo). 

And enjoy this, from the Mongolian heavy metal band The Hu, whose two YouTube videos have racked up over 10 million views (skip to 1 minute to see why), here’s the Goats and Soda piece about them.