Thread by @DinaPomeranz

Now being presented at NBER DEV summer institute: "Outsourcing Service Delivery in a Fragile State: Experimental Evidence from Liberia"

Important impact evaluation of a large & controversial education policy.

By @marome1, @JustinSandefur &
@wsandholtz

nber.org/conf_papers/f1…

At baseline, learning in Liberian schools was extremely low. Even after 6 years of schooling, less than 20% were able to read (compared to over 90% in Burundi).

Students in the treatment group were learning about 60% more per year than those in the control group.

There were several different providers. BRAC, Bridge Academies, and 6 others. Their effectiveness varied a lot. (But standard errors are large)

The contract details mattered a lot. In the case on Bridge Academies, the rules were less strict, and the payment was not per student.

This provider shifted pupils from
oversubscribed schools and underperforming teachers to other government schools.

Lot of questions still open. Can't make broad conclusions about Public Private Partnerships. Contract details matter.

Outsourcing also requires state capacity to monitor & manage this process.

These schools weren't just outsourced, they also got more $ than regular schools.

Comments by Michael Kremer.

First: big compliment to authors. This program was rolled out in an extremely fragile setting over a short amount of time. Impressive how this was implemented and evaluated.

In this very low capacity, post conflict environment, most of education budget came from aid and donors didn't want to fund the schools through the very low capacity government.

This is different from charter schools, as teachers remained civil servants.

Per pupil budget was double the regular schools. But extremely low (100$ per pupil).

The test score impacts were actually extremely high compared to many other school interventions.

Other positive outcomes: more student and parental satisfaction.

Lower household expenditures. Better ethnic relations.

Very large reduction in both student and teacher absenteeism.

Lots of improvement in management. The management seems key. Other research shows more resources without improved management has more limited effect.

A lot of press coverage and controversy about the program, in particular about Bridge.

Bridge had a different contract and also a very different pedagogical methodology. They use very strict lesson plans via tablets.

The fact that the effects had large heterogeneity between providers is not a negative. It means government can dynamically choose the better providers, so the average effect can improve.

Some final comments:

Final, final comments :)

Overall Kremer is impressed both with the impacts of the program and with the ability of Liberia to pull this off in this very difficult context.