An Accidental Health Economist Talks Education, Bill Gates, and why Impact Evaluation isn’t Enough

Dean Jamison is one of the world’s leading global health economists, but the way he tells it, it was something of an accident. He started as an education economist at the World Bank in 1976, and he was a pioneer in using impact evaluations to study the effectiveness of distributing textbooks in Nicaragua and the Philippines.  Later on, Jamison received an invitation to join the World Bank’s first economic mission to China, where he was tasked with analyzing health issues. “I was eager to see the country and didn’t want to tell anyone how little I knew about health,” he admitted recently.  “But from that point on, I was a health economist.”

It’s a good thing too. Jamison went on to become lead author of the World Development Report 1993: Investing in Health, which Bill Gates called “mind-blowing” and said it inspired him to focus his philanthropy on global health. Since then, Jamison served as co-chair of the Lancet Commission on Investing in Health with former World Bank chief economist Larry Summers and led work on the World Bank’s nine-volume  Disease Control Priorities series, which  are widely used in low-income countries as a guide for spending on health care.

In a recent interview with the World Bank’s Strategic Impact Evaluation Fund, Jamison, professor emeritus at the University of California, San Francisco, talked about why the World Bank’s first chief economist thought impact evaluation wasn’t really economics, how the field has changed in the decades since, and why you don’t need to start worrying about a flu pandemic—yet.

In 1981, you published a paper about the impact evaluation of an education project in Nicaragua. Impact evaluations weren’t being done at the World Bank at the time - so how did you get the idea?

I had just come from working with the Institute for Mathematical Studies in the Social Science at Stanford, and at the time, their work on education was 20 years ahead of economics in terms of using randomized control trials. I was an economist, but I had the benefit of exposure to education’s different intellectual tradition.

Was there support for impact evaluations?

To get funding for the Nicaragua study, I had to speak with Hollis Chenery, the World Bank’s first chief economist. After I explained my idea, he looked at me and said, “This sure isn’t economics, but you seem like a reasonable guy, so I will support you.”  Impact evaluations in education were the last thing an economist with PhD from Harvard was expected to do. Chenery didn’t understand why I would want to demean myself by doing that type of work, but he figured if I wanted to do it, he would support it. My manager’s reaction was very different.  Timothy King, who was the chief of the World Bank’s Population Division, very much supported the notion of understanding what works and what doesn’t, and how cost-effective various programs were. But it was definitely not fashionable, like it is now.

Were people open to the results?

The evaluation in Nicaragua found that giving children access to textbooks improved their learning, and so the results confirmed prior expectations

and were widely cited and used. At that time, classrooms were totally barren of textbooks, which was seen as a barrier to learning. It was similar in the Philippines: an evaluation of a World Bank operation showed a substantial increase in learning when the textbook ratio was reduced from 10 students for every textbook to two students for every textbook.  The results also showed that further reducing the ratio to one student for every textbook—which would have been twice as expensive—wasn’t any more effective. Everyone, including the government, was interested in that research and pleased with the results. But results that didn’t conform to expectations – such as our findings in the Nicaragua evaluation that math instruction delivered via radio had a dramatic impact on student learning – weren’t used.
 
You were publishing results of impact evaluations in the early 1980s, but the field didn’t really take off until much later. Why is that?

Many of the reasons are practical. Impact evaluations of World Bank project investments are expensive, require data that aren’t routinely collected, and can be labor intensive. Also, there need to be incentives. You may require a 10-year period to get results, which bites a whole decade away. Researchers without tenure need to get their work published in journals quickly—or they don’t have a job. The professional rewards structure for that kind of work was out of whack and so there wasn’t a lot of take up. Another factor is that randomized control trials are so simple that anyone can understand the results, they looked a little intellectually lower-class compared to garden variety econometrics. Randomized control trials in the pharmaceutical business are very different: drug trials can end after six-months and they have an end point that doesn’t require a 15-year wait. 

Impact evaluations are much more popular now. Do you think that’s a positive development?

Randomized control trials serve an important purpose, but it can be like looking under a street light for lost keys. You shed good light on a short-term and well-defined question, but the world’s most important questions might be ones that require longer-term horizons to unfold, and may require a complex intervention. These are questions that a randomized control trial can’t answer.

Do you miss your time at the World Bank?

The Bank was the high point of my career in terms of being intellectually engaged. I published a lot before I joined the Bank, but I wasn’t used to having actually people read it. As an assistant professor, I was my own free man, but at the World Bank, I got thoughtful feedback from peers and a manager that I really respected.

You recently published an article with Larry Summers and others on global preparedness for pandemics. It seems every year now that reports warn of a particularly bad flu season. Is there a pandemic on the horizon?

Ebola scares people, but the risk of an influenza outbreak is much larger. Having said that, it’s unlikely that something as severe as the 1918 flu outbreak will come back this year. Our modeling says to expect an outbreak like that every 100 to 200 years, so there’s about a one percent chance it will happen this year. But if it does, we are nowhere close to ready.

Find out more about World Bank Group Education on our  website  and on  Twitter .