Syndicate content

Recent comments

  • Reply to: What’s the latest in development economics research? A round-up of 140+ papers from NEUDC 2017   3 days 23 hours ago

    Thanks for the summaries. Great effort

  • Reply to: IE analytics: introducing ietoolkit   5 days 5 hours ago

    Thanks for sharing this, Paul.


    Exporting customized tables is a problem we often run into. What we usually do is use Stata's file write function to create a tex table, with locals inputing the regression outputs.


    Yours seems like an inventive solution, however. I do have some questions about it -- I'll just ask them in your repo, if that's ok.

  • Reply to: IE analytics: introducing ietoolkit   6 days 2 hours ago

    This is pretty neat.

    We have also put together a package with a similar purpose of making it very easy to generate complex tables from Stata:

    Ours is more focused on difficult tables. First you create a latex template for your table with placeholders for all the estimates or numbers you want, and then link that to an estimates file generated by Stata, with a program that combines these two.

    It's definitely more work than the one-line solutions you're describing here, but it allows infinite customization.

  • Reply to: What’s the latest in development economics research? A round-up of 140+ papers from NEUDC 2017   6 days 11 hours ago

    Very useful economic research report. There are many reputed organizations like Art of Living foundation doing noteworthy work in the field of development in rural areas.

  • Reply to: How big data helped us estimate the impact of an intervention with 0.8% take-up   6 days 19 hours ago

    Hi David and Claudia, thanks so much for this post- it is incredibly timely. We at IDinsight have been experimenting with these techniques to extract LATE estimates in conditions with low take-up, and although I knew others must have been doing similar things this is the first paper I've seen. Just last week I finished up a matching algorithm that started as the equivalent of your "nearest neighbor LASSO" approach, though then deviated for reasons that aren't interesting to get into here.

    I'd be really interested to see (or write perhaps?) some papers that stress-test this approach. For instance, would it give an estimate similar to IV in conditions where IV is appropriate (and confidence intervals are tighter)? Should we be using LASSO or one of the myriad of other machine-learning-for-prediction techniques? How do we know if the data is rich enough to give reliable estimates? Can we figure all this out with simulation? Etc.

    If anyone knows any papers dealing with these topics, I'd love to see them.