The results of Liberia’s experiment to outsource education are in

Testing times.
Testing times.
Image: Reuters/Thierry Gouegnon
We may earn a commission from links on this page.

In 2016, reeling from a devastating Ebola outbreak and still dealing with the legacy of civil war, the Liberian government announced a dramatic experiment to improve its education system. The Partnership Schools for Liberia (PSL), now called the Liberian Education Advancement Partnership (LEAP), handed the operation of 93 public schools to a group of eight private organizations, including a mix of both for-profit companies and non-profit charities, as well as local and international organizations. The schools would remain free to kids and use government teachers, but the experiment would test whether a change in management improved the abysmal state of the country’s education system.

Notably, the government agreed to run a three-year randomized control trial, administered by the Center for Global Development (CGD) and Innovations for Poverty Action. The CGD just published its report, looking at learning gains, access, safety, and sustainability (that is, cost) over the three-year period.

LEAP schools raised test scores by 0.21 standard deviations in math and 0.16 standard deviations in English during the trial. For the group that started in first grade, that’s the equivalent, it said, to four words per minute additional reading fluency—from 11 to 15 words per minute.

According to the report, the program reduced corporal punishment by 4.6 percentage points, from a base of 51%, but didn’t improve the dropout rate or incidence of sexual abuse. “Results vary by provider: some produced uniformly positive results, while others present stark trade-offs between learning gains and other outcomes,” the authors wrote (pdf).

Justin Sandefur, who co-authored the study, said there was a lot of variation in the results, with some operators able to deliver cost-effective learning gains, some delivering learning gains by expelling unwanted students to reduce class sizes, and some not delivering much in the way of gains at all. “Beyond year one, learning gains plateaued,” he said.

Luis Crouch, senior economist in the International Development Group at RTI, agreed. “These are okay but not great results,” he said. “Recall that countries such as Liberia are 2 to 2.5 standard deviations ‘behind’ the high-income countries. So the global community ought to generally not take too seriously results in the 0.25 range.”

Unintended consequences

The project aimed to test a few things: the role of public-private partnerships and, specifically, whether poor countries with broken education systems could improve schools by outsourcing delivery to private operators.

When the partnership started, only 38% of kids were enrolled in primary school (pdf), and of adult women who had finished primary school, only one-quarter could actually read a sentence.

The experiment also became a high-profile test for Bridge International Academies, a highly controversial, for-profit chain of low-cost private schools which rely heavily on technology and scripted lessons. Its ambition is to educate 10 million poor children by 2025. In early 2018, it raised $100 million from investors and was losing about $12 million a year educating 100,000 students. It now educates 400,000; a co-founder said breakeven would be 500,000 kids. Bridge is backed by Bill Gates and Mark Zukerberg’s foundation.

Bridge heralded the results. “In Bridge-supported schools student learning increased by 0.62 SDs, which is the equivalent of 2.5 years of additional learning,” said Ben Rudd, a spokesman for the company.

It is hard to draw sweeping conclusions from just two and a half years in an education system with very poor overall performance. For example, in the control schools for Bridge’s sample, the kids who started in first grade in 2016 could read about four words per minute at baseline; after three academic years, they could read about 12 words per minute. In the schools assigned to Bridge, that cohort can now read about 18 words per minute. A fluent reader should score over 65 words per minute.

Observers pointed out that some gains came with unintended consequences, namely expulsions. The dropout rate was 15% in control schools, and it was seven percentage points higher in Bridge schools. No other operator had such a significant impact on dropouts, up or down.

“There’s no dropout story for the other operators; there’s only a dropout story for Bridge,” Sandefur said. 

Rudd disputes the way the CGD analyzed some of the outcomes, arguing expelled kids should not be included the sample (preferring to instead test the effects of those who attend program schools). Sandefur disagrees: if a program expels children to reduce class size and it benefits the remaining students, that is an approach a government school cannot replicate. (Bridge said the dropouts were due to pregnancies, a claim others did not make.)

Paul Skidmore, co-founder of Rising Academies, another operator in the study, said the story on enrollment was complicated. The choice for operators, he said, was simple: “If you believed that students were going to be better off in your schools than the alternative, there was a responsibility to do whatever it took to enroll them and keep them enrolled, even if that meant improvising and flexing your model to fit the situations you encountered.”

Sandefur said the results show that some providers, such as Street Child, were able to produce learning gains at low costs without kicking students out. It hasn’t received the same attention as Bridge, which is more expensive. Bridge, however, points out it is improving on this front: from $640 per student in its first year, the cost came down to $215 in the second and then $163 in the third. It aims to get to $103 next year. Street Child’s cost was $37 by the third year of the trial.

The Liberian government’s cost per pupil was $50 a year when the trial began, and it paid the program operators in the trial another $50 per student (which was funded externally).

The takeaway for the Liberian system as a whole is mixed. The program, according to the CGD’s report, “lavished additional resources on many of the schools with the best staffing and infrastructure in the country.” Spending more on schools across the country would be ideal, but that’s unrealistic. The report notes that management at the program schools was far better. And gains, while not as big as some hoped, were still gains.

Liberia has since expanded the program, with 50,000 kids now attending 194 schools run by private operators.