Thanks to the authors for coming on here and clearing up some of the confusion.
Huge thanks to Berk for his illuminating and educational series of posts on this topic.
1. Berk on this -
"These two estimands are not legitimate alternatives to each other in the three-year impacts paper: when they are not equal to each other, only one of them is the ITT effect, but the place to adjudicate that is not the abstract."
The one that IS the ITT effect is the across-village, correct?
2. I think "yes." And thus if Berk if we take your summary (shared on Twitter) at face value -
a. There is no stat. sig. treatment effect (other than on assets), and
b. There may be negative spillovers (with a wide confidence interval)
How do you reconcile that with -
"Even if one accepts the least rosy interpretation, the results do not say that cash transfers aren’t “good.” Too much solid evidence shows positive impacts on consumption, nutrition, happiness, etc."
i.e. What is your most updated view on the efficacy of unconditional cash transfers? Meaning - the results from HS18 do not say that cash transfers AREN'T "good," per se. But they also don't say they ARE "good." And so I'm curious about what your take is there.