Library Research Signals, screening and ordeals
A Small Hassle Did the Sorting
In 400 Indonesian villages, households that had to travel and apply for a cash-transfer program turned out much poorer than households enrolled automatically. A little friction did the sorting, and adding more didn't help.
Vivi Alatas, Abhijit Banerjee, Rema Hanna, Benjamin Olken, Ririn Purnamasari and Matthew Wai-Poi, 2016
What they were looking at
In 2011 Indonesia expanded PKH, its conditional cash-transfer program, which pays poor households $130 a year for six years. Every program like this runs into the same question. How do you get the money to the households that need it most, when you can’t easily see who they are?
The researchers ran a randomized experiment across 400 villages in Lampung, South Sumatra and Central Java, and used the same asset test everywhere.
- In some villages, households had to travel to a site and apply, then take the test.
- In others, households were surveyed at home and enrolled automatically if they passed.
What they found
“requiring beneficiaries to apply for benefits results in substantially poorer beneficiaries than automatic enrollment using the same asset test” Alatas et al. (2016), abstract
Where did the difference come from? Mostly from who didn’t show up. Better-off households expected to fail the test, so they didn’t bother to apply, and the trip did the sorting before the test ever did.
“adding a small application cost to a transfer program can substantially improve targeting through self-selection” Same
They also tried making the trip harder, and that didn’t help.
“Marginally increasing application costs on an experimental basis does not further improve targeting.” Same
By the working paper’s estimate, it would take a far bigger ordeal, “e.g. tripling wait times to 9 hours or more,” to get noticeably more sorting.
What it means
So can friction sort? Yes. A small cost that the people who wouldn’t qualify won’t bother to pay does the screening for you. But the sorting comes from that first small step, and piling on more mostly adds cost, which lands on everyone.
Friction has a cost of its own, too. When the US closed Social Security field offices, the number of people receiving disability benefits fell by 16%, and the drops were largest among moderately severe, low-education applicants. So targeting got worse, because the extra friction screened out people the program was there to reach.
Where it shows up in a revenue team
- One qualifying step filters well. That’s a short call to book, one question that matters, or an appraisal.
- A ten-field form isn’t ten times better than a two-field one. Alatas and colleagues found that adding cost past the first ask didn’t improve the targeting any further, and Deshpande and Li’s study of disability offices found that added hassle drives away people you want too.
- Measure who a step removes, and how many. Every time you add or take away friction, check what happened to the mix, by tier.