Prof. Michael Kremer
Big Brains podcst

Fighting Poverty And Pandemics, with Nobel Economist Michael Kremer (Ep. 66)

Economist discusses how randomized control trials can improve poverty interventions

Prof. Michael Kremer
Big Brains podcst

Show Notes

The solutions to global poverty can appear obvious, even if they’re difficult to implement. But, as University of Chicago economist Michael Kremer has discovered, interventions that may seem like common sense can actually be wrong.

In 2019, Kremer won a Nobel Prize for his work studying ways to alleviate global poverty. A pioneer in the use of randomized control trials in economics, Kremer has examined poverty interventions like scientists do medical treatments—putting interventions through a trial to isolate effects. Kremer’s studies often reveal surprising and counterintuitive ways of fighting global poverty and have radically altered thousands of lives.

Subscribe to Big Brains on Apple PodcastsStitcher and Spotify.

(Episode published April 5, 2021)



Paul Rand: When you think about the people working on the ground to address poverty, you probably picture healthcare workers, activists, and maybe not-for-profit workers, but you probably don’t think of economists. Well, Prof. Michael Kremer wants to change that.

Michael Kremer: People often think of economists as often an ivory tower looking at data that they had no involvement in collecting.

Paul Rand: Kremer, who is now an economist at the University of Chicago, previously at Harvard, won a Nobel Prize in 2019 with two MIT scholars for their experimental approach to alleviating global poverty.

Tape: This year’s laureates have shown that it’s best to break down this daunting issue to smaller, more precise questions, questions you can credibly answer. And the most credible way of answering them is to try particular interventions in field experiments. This experimental approach has completely reshaped research in development economics.

Paul Rand: That approach involves getting out of the lab.

Michael Kremer: You’re spending time in the field, talking to teachers, to students, to farmers, to nonprofit organization workers, or government workers, and that gives you a richer sense of context.

Paul Rand: When you’re dealing with the problems of global poverty, it can be difficult to isolate real solutions. There are so many factors surrounding any policy or project that figuring out the causal relationship between an intervention and an improvement in people’s lives is not always obvious.

Michael Kremer: Things we think are common sense, sometimes they turn out to be right, sometimes they turn out to be wrong. And things that we thought would have minimal impact or no impact turn out to have a big impact. I think it’s just good to maintain some humility about the degree of confidence that we have about things that we believe.

Paul Rand: That’s why the experimental approach is crucial. Kremer conducts randomized controlled trials to isolate results.

Michael Kremer: Doing experiments forces the researcher, and I say forces because I think a lot of economists didn’t start out had to be forced to spend time in the field. But once you’re doing that, then you start to think more broadly and you realize other things that are going on.

Paul Rand: From the University of Chicago Podcast Network, this is Big Brains, a show about the pioneering research and pivotal breakthroughs that are reshaping our world. Today, innovative approaches to old problems in economics. I’m your host, Paul Rand. Michael Kremer’s approach to economics is unusual, or at least it was until he made it mainstream. That story starts all the way back before he was even in graduate school. He was a high school teacher in Kenya, and he would return years later to find a connection he had made at an NGO, the International Child Support Fund, telling him about an idea, they had to improve some schools that they worked with.

Michael Kremer: This was a rural area of Kenya, pretty poor. There were pretty few textbooks in the school. So I think on average that time in these rural schools, one textbook for every 17 pupils. So in that environment, it’s common sense, and I certainly believe that one of the most cost-effective things that could be done to promote learning was to provide textbooks for kids.

Paul Rand: Seems logical.

Michael Kremer: It does. And there was actually a debate in the economics of education about the impact of spending, and some people you want to improve the character and liberals would say, "Hey, you need to spend more money." And conservatives would say, ‘‘It’s really not about the money, it’s more about getting better incentives for teachers and for administrators.’’ Even the people who are generally pretty skeptical about funding believed that if you had a shortage of textbooks, remedying that shortage of textbooks would increase learning.

Paul Rand: More textbooks equals more learning, equals better students, equals less poverty. All that makes sense, right? But how could you know for sure? A randomized controlled trial, the experimental approach.

Michael Kremer: So they had a limited amount of money and they did something that I think was actually pretty brave. They thought, ‘‘Well, we’re going to try to carefully evaluate the impact of what we’re doing, rather than assuming we’ve got all the right answers. We’re going to try to test it.’’ So they chose twice as many schools and they introduced the program at least initially in a subset of those schools. That allowed us to compare the schools that got textbooks from those that didn’t. And having taught high school in this area, I was initially really surprised when we saw the results. So we looked at the impact on average test scores and we didn’t really see any impact.

Paul Rand: No impact, the assumptions were wrong. Now some economists might have said, ‘‘Well, there you go, textbooks don’t really matter for those students and funding isn’t really important.’’

Michael Kremer: When you get a paradoxical result from a randomized trial, I think there’s a tendency if you’re picking a particular framework to try and make everything fit inside that framework. If you’re in a situation where if you just saw a correlation of the data, you could come up with some story about how it’s all explained by expenditure and incentives. But when you get some results, like this textbook result, that you weren’t expecting, but you can’t really make go away and pretend isn’t there, you have to question, not just what the impact is a particular program, but you have to think about, ‘‘Well, what’s really going on here?’’

Paul Rand: And this is where the other factor of the experimental approach becomes important, being on the ground and truly understanding the context.

Michael Kremer: And then if you are familiar with the context, then you can say, ‘‘What are some other hypothesis?’’ So thought about, ‘‘What could be going on here?’’ And we were missing a really important question, which is, what’s the organization of the curriculum? It’s a curriculum set to the typical student, or is it set to correspond to some international standards, or to some colonial pattern in a way that’s actually, they leave the typical kid behind?

Paul Rand: Everyone had assumed that the problem was lack of textbooks, but by ruling that out with the trial, they could dig deeper to discover the real issue.

Michael Kremer: So in Kenya, as in many former British colonies, school is taught in English. More generally the school system is oriented towards the top performing students. So if you imagine being taught in your third language, and then these kids are very poor, they might need to hop out at home sometimes. There’s all sorts of pretty bad disease environment with malaria, HIV was pretty bad, intestinal worms, which used to be a big problem in the US but are no longer, so they miss school a lot. It’s pretty easy to fall behind. And the curriculum was set at a pretty-

Paul Rand: High level.

Michael Kremer: ... pretty high level. Basically I would say it’s pretty similar to what I experienced in the US. And so the hypothesis, and this is just a hypothesis, was that if you fall behind, it’s pretty hard to benefit from textbooks that are geared to the official curriculum and then in your third language. How do we try to get a hand on us? Well, partly we did that by looking at the data from this experiment. And we saw the kids who started out on the pretest and performed well in the pretest, hadn’t really fallen behind the official curriculum, they actually showed big gains than the schools that had the textbooks versus those where the textbooks haven’t come in yet. But for the typical kid there wasn’t that gate. So that suggested that maybe there was a larger issue here, it wasn’t the textbooks, should be able to work, still believe in common sense, but these particular textbooks and this particular environment weren’t doing it.

Paul Rand: It turned out they had been throwing a textbook shape solution at a curriculum-shaped problem.

Michael Kremer: And then what we found was that you can actually rapidly catch up to the official curriculum. There were huge gains to providing some, either remedial education or approach it’s been called the teaching of the right level. And this is now being used quite widely, and I think a lot of kids are gaining. I think this is an example where I don’t think the nonprofit organization did the wrong thing in giving textbooks, it was a reasonable thing to do, but they actually tested to try and look what was the impact of this? And then we learned, and then that’s something that can be adopted by many different governments in terms of how they organize their school system and can generate benefits for many countries.

Paul Rand: This wasn’t the only randomized controlled trial that Kremer did in Kenya. When the NGO he was working with came up with another intervention, Kremer saw another opportunity, this one surprised everybody even more, and it involved an issue that no one was focusing on, but they discovered what was actually holding students back, it was worms.

Michael Kremer: These are worms like hookworm, woodworm, roundworm. These are worms that enter your body through, for example, your feet, or your hands if your hands are exposed to contamination, and they’re very, very common in this part of Kenya. 90% of kids had them or [inaudible 00:09:09] they’re not the worst health problem in the world, they don’t typically kill people. There’s malaria, there’s HIV, lots of other things that will kill you, but you can understand why it might not have been top of the agenda of the health ministry, also why the education officials might’ve thought, this is really a health problem. And just as background, the treatment for deworming is extraordinarily inexpensive, two cents per for the most common worms. It’s two cents per dose.

Michael Kremer: Well, the World Health Organization had actually recommended that in the areas where words were common, although the medicine’s really cheap, it’s pretty expensive to test somebody. You need to take a stool sample to a lab, get the results back, find the... That’s just not cheap and it’s not easy, but what the World Health Organization was recommending was that environments where there are lots of worms, just go to the schools and treat everybody, and because the medicine is very safe. So this is what the nonprofit organization tried, and this was the opposite of the textbook case.

Paul Rand: Very much, right.

Michael Kremer: Exactly. The nonprofit organization, they started out very small, they started out working in seven schools. They then were going to expand, and eventually were going to try to introduce this program of treating the worms in 75 schools, but they started out in 25 and they extended to another 25 and saw, ‘‘The part we’re through, we could compare the schools where the deworming treatment had already been provided and to those were added.’’

Paul Rand: Basically, everyone was missing an intervention that was right under their nose.

Michael Kremer: What we found was we saw that school absences went way down by about a quarter.

Paul Rand: Just by de-worming people?

Michael Kremer: Yeah, exactly. And the kids were missing a lot of school. And there are a lot of factors. I mentioned before, malaria, economic issues like HIV from the parents. And then the kids have to stay home and take care of their siblings or work. So we didn’t think that treating worms alone would have such a big impact, but that was what the data said. We also saw that beyond that, it wasn’t just the kids who got deworming medicine, there was some evidence interrupted transmission as well, and that other kids in the same school, or even kids in the neighboring schools seem to benefit. We saw reductions in wormload and other kids in the same school, other kids in nearby neighboring schools. So when you add up all these effects, it looks like a very cost-effective way to boost school attendance.

Paul Rand: So you were saying earlier, Michael, that these kids were missing a lot of school, but actually they’re now having more time in school. And I’m assuming what you found is that their school performance perform because of that additional time in school, is that right?

Michael Kremer: It was quite interesting. We saw that everybody went to school more years. We’ve recently done a follow-up study and we followed up these same kids, really, now 20 years later.

Paul Rand: So 20 years later you’ve continued the experiment, that’s remarkable, isn’t it?

Michael Kremer: Yeah. There was a lot of work involved in finding these kids 20 years later. And we did work throughout the period to try to get interim results. But now we know what the impact is, not just on education, but actually long run outcomes like earnings and how much consumption people have. And what we find is, roughly about a 15% increase in earnings and [inaudible 00:12:33] which I don’t want to get the wrong idea, this is not going to turn Kenya into Singapore, but this is a very meaningful-

Paul Rand: 15% is quite meaningful, I would bet it.

Michael Kremer: Exactly. If you think about increasing your own income 15% that would be pretty nice. And this is medicine that costs two cents a dose, and if you’ve counted the delivery cost and so on, maybe 50 cents per person, so an incredible rate of return, that’s an investment.

Paul Rand: And why is it that that deworming had such an impact on income, do you think?

Michael Kremer: That’s a great question. And to be honest, I don’t think we fully understand that. There are a number of possible explanations. Partly could be the impact on education, but we saw a bigger impact on education for the girls than for the boys, and yet we saw a bigger impact on earnings and economic outcomes for the boys. I think it also could just be that if you have more energy, maybe you’re able to take on certain types of jobs that you wouldn’t be able to take on otherwise, and maybe that has a persistent effect.

Michael Kremer: One clue is in the early results, in the interim results, there was a change in the type of jobs that people had. For example, off of a pretty low baseline, there was an increase in the chance people would work in manufacturing. This wasn’t manufacturing as in working in an auto factory, but it was maybe working in a mill to process grain, local mill, something like that.

Paul Rand: Much like the textbooks, isolating dewormings effect on school attendance and income gives other countries and NGOs important data about what might work in their own communities. And Kremer says that the experiment gives us two big lessons for the field of development economics. First, health interventions should be preventative and free.

Michael Kremer: The nonprofit organization that did this work, they have a general philosophy as a lot of organizations did at the time, that it’s very important to charge. And then if you give stuff away for free, that’s going to create negative consequences and we need to charge at least something.

Paul Rand: Because people wouldn’t appreciate it or they wouldn’t take it serious [inaudible] or something like that.

Michael Kremer: Exactly, that was their feeling. And as I said, it was very widespread at the time. We talked to them and they decided they were going to actually, again, try to look and see if their assumptions were correct and they tried providing it for free in some schools. And what we saw was there was a huge drop off and take up in the schools where they put even a modest charge compared to the schools where it was preferred.

Michael Kremer: So it turns out that this is something that isn’t just the case for deworming medicine, it’s been found for using mosquito nets against malaria for many other preventive health behaviors. So I think this is something where now there’s actually been a big change, I think that the standard now is for governments and other organizations to provide at least the cheapest and most effective preventive health products for free.

Paul Rand: Second, it’s crucial to understand the communities and the people that you’re working with.

Michael Kremer: I think the reason why this has had a big impact on economics is not just because it allows us to more cleanly isolate the impact of a particular program, I think it’s also because it means the researchers really getting involved with practical, specific problems and is working with not just on their own analyzing the dataset, but they’re talking to whether it’s the people that they’re being interviewed to collect the data from, whether it’s the government officials, or the teachers, or in some cases to other social scientists or to healthcare experts. And having that bigger context allows you to think about questions, maybe [inaudible 00:16:31] differently than you were.

Paul Rand: Kremer’s worth challenging the assumptions of how we fight poverty is what he won the Nobel prize for in 2019. But it’s not the only work he does. After the break, Kremer’s work on founding new nonprofits and speeding up the development of vaccines. Another aspect of your work and your career that’s just so interesting to me is that you’re spending time, not only founding, but being very active in nonprofits. So you’re not only studying, but you’re applying. And I wonder if I can trouble you just to talk a little bit about how you came to some of those decisions specifically about one of the NGOs you founded called Precision Agriculture.

Michael Kremer: So one area where I done some research was on the impact of providing information to farmers. I grew up in Kansas, in the US there’s extensive systems for agricultural extension for giving [inaudible 00:17:32] information to farmers so they can adjust fertilizer or [inaudible]. That system works pretty well in the US, but in developing countries, there often aren’t strong systems in place for that. And we’ve seen a big change across the world through adoption of mobile phones, and that’s true, even in pretty low income countries.

Michael Kremer: So even in Kenya, most people have mobile phones now, and they might not be smart phones, but they’ve got a basic phone. And that means that it’s possible to deliver information to farmers that is based on their location. Certainly you can deliver information that’s tied to a particular time in the agricultural season, but also can be timed around outbreaks of new pests.

Michael Kremer: So we did some trials of what was the impact of providing information to farmers, and what we found was that it really did affect farmer behavior. Again, I don’t want to claim that this is a magical factor, I’d say it’s transforming things, but it’s some ways a little bit like the deworming medicine, it’s very cheap to some of these mobile phones [inaudible 00:18:36] to choose.

Michael Kremer: So even if some farmers changed their behavior in response and say adopt... So a big problem in this area was acidic soil loss, and you can address that using [inaudible 00:18:48] to restore the PH balance. And some farmers behavior in this way, the rate of return was economist, so think about that. The benefit per dollar spent can just be a minute. These are rough numbers, but we’d estimate 10 to one ratio. So this was something that could have a very big benefit cost ratio. And I’ve been doing work in Kenya, another researcher, Sean Cole had done similar work in India, found similar results. And so I was approached by a philanthropist and we decided to form an NGO that would try to work with... And the NGO acquisition development works with governments, with nonprofit organizations, with private firms.

Paul Rand: Did this surprise you that you all of a sudden were in the NGO business?

Michael Kremer: I think that the image of economists 30 years ago and my image of what I would do when I went to graduate school was not that I’d be involved by nonprofit organizations. I think in some ways, if you think about, say, biotech or computer science, obviously a lot of people... There are things coming out of those that are practically useful, and maybe we’re finally getting to the stage of economics.

Paul Rand: Kremer’s desire to find practical solutions goes beyond randomized controlled trials and founding NGOs. He’s also helped up in the economics of vaccines, which is obviously extremely relevant today. The model he’s helped develop is called the Advanced Market Commitment.

Michael Kremer: And was used for something called pneumococcus vaccine.

Paul Rand: There are some viruses and bacteria that companies don’t want to invest the R&D time in to create vaccines because they don’t see a way to make money on them. Advanced Market Commitments are essentially contracts usually offered by governments or donors that guarantee a viable market for a vaccine once it’s successfully developed.

Michael Kremer: So $1.5 billion was set aside to help finance the purchase of pneumococcus vaccine as a way to encourage companies to invest in that.

Paul Rand: These commitments create a financial incentive for the private sector to develop new vaccines for diseases that affect developing countries.

Michael Kremer: And that approach was successful, vaccines were developed, we’re reaching 700,000 that they’ve estimated to save 700,000 more lives.

Paul Rand: When COVID-19 hit, Kremer immediately recognized the same issues around incentives within the private sector.

Michael Kremer: There are some decisions which are risky for a private firm. When do you start the construction? Do you construct this big factory? And it’s pretty expensive to construct big factories. Do you start constructing the factory before you have the results of the trial, or do you wait till after the trial and then start the construction? Well, normally what the private sector does is they wait til afterwards. This is a case where [inaudible 00:21:45] slowed things down by three months, by six months, the benefits of speeding up access to a vaccine. Obviously there are huge benefits in human lives. The economic impact of this is huge. The estimates are that the world is losing $500 billion just in short run GDP losses alone. And if you take a more comprehensive measure that includes long run health costs, costs in education, you get much, much bigger numbers. So one study just for the US estimates $800 billion every month.

Paul Rand: Wow.

Michael Kremer: So that means just from an economic point of view alone, if we could do anything to accelerate the end of the pandemic, for example, by speeding up vaccination, it’s almost shortly going to be worth it. It’s worth putting huge amounts of resources into it. And so both getting larger scale capacity and making sure that capacity gets installed early on is super valuable from a societal point of view.

Michael Kremer: Now, it’s valuable for a private firm as well, but it’s a lot more profitable for a society [inaudible 00:22:52]. So the values of society greatly exceeds the value to a private firm. If we hadn’t put in enough factory capacity to produce one more course a year, that’s worth $1,700 to society we estimate.

Paul Rand: When did you you propose this, Michael?

Michael Kremer: I think it was just around the time the work speed was announced shortly thereafter. The work speed was trying to do this, we advocated actually for a program that would have been larger scale [inaudible 00:23:23] that would have taken investing even more.

Paul Rand: Kremer was arguing that we needed to commit to investing in vaccine production capacity, even before we had a viable vaccine, so we could shave months off the wait for vaccines. This largely didn’t end up happening, and Kremer says we could have ended up in a situation where we were waiting even longer than we are now to vaccinate everyone. We just got lucky.

Michael Kremer: I think we got really lucky. It turned out that a bunch of different vaccines work. When we did our modeling for this, we actually assumed a pretty low chance of their work, but it was even if you assumed a low chance, this was a no brainer as an investment decision. It turned out that we got lucky and a bunch of them worked. We hope we’ll be able to vaccinate the US by the end of the summer, but from a world point of view, we need a lot more production.

Paul Rand: Can I ask you, what was the reaction to the study and the proposal that you all made?

Michael Kremer: I think it’s glass half full, glass half empty.

Paul Rand: Okay.

Michael Kremer: There were a lot of investments. I think the US was one of the leaders in this, UK as well. Both the US and UK invested a lot in vaccines, and we’re seeing the benefits, that we call for a lot more investment than we had, than was actually made, and I think it would have been better if there was more.

Paul Rand: Kremer says that the lessons we learned here are vital and that in the future, like with Advanced Market Commitments, vaccine buyers should invest in manufacturing capacity early for the right to buy doses at marginal cost. But the issue of how to ramp up COVID-19 vaccine production today is still far from over. Although wealthy Western countries are expecting to be vaccinated by summer, this is not the case for developing countries. Kremer’s team is still working to think of ways to make production even more efficient.

Michael Kremer: Imagine you hire a contractor to do some work in your house. You can say, ‘‘Hey, I want the kitchen remodeled in the following way." Well, the contractor may come back and say, "Yeah, I was going to try and get it done by April 15th, but it’s not ready, I’m going to need six more weeks.’’

Paul Rand: That would never happen by the way.

Michael Kremer: Exactly. And six more weeks waiting while your house is not ready, it’s a pain, but six more weeks when 200,000 people are dying of COVID each month and where the economy is just in terrible shape because of it, that’s a disaster. So you really want to specify the contract in ways that are going to give the company incentives to invest early and to invest a large scale. We thought that the best way to do that was set the contracts explicitly include not just the doses, not just the dates, but speed. Actually, the capacity of installation-

Paul Rand: Got it.

Michael Kremer: ... whether it’s building or repurposed into the factories, because otherwise you might do this in a contract with a penalty clause. If you don’t get it done by such and such a date, there’s some penalty or bonus for early delivering. Problem is to match the incentives you need very big penalties or bonuses, and that’s just too risky and even for the government, they’re going to be reluctant to sign that. And there are a lot of factors that are outside of their control. The testing could be delayed by adverse events, say that shutdown with trials. So we thought the best way to do this was to explicitly include goals for installation of the capacity, which is not perfectly under the control [inaudible 00:27:03].

Michael Kremer: So if you ran a contract and you just say, ‘‘We want this many doses,’’ well, it’s an incentive of the firm if you just look at purely the commercial incentives of the firms. I’m not claiming that they’re not motivated by other factors, but from a purely commercial point of view, well, maybe don’t build the biggest factory, build a factory that you’re confident you can build, and the factory that you can build at a modest cost, but maybe don’t go shut down some other factory and entirely repurpose it to this, because that’s going to be very expensive and give up some other product that’s really valuable. They did go much beyond what they’d ever done historically, they should get credit for that. But really this is a national emergency and it’s a global emergency, and going even further to expand capacity, it would have been worth it from a society’s point of view.

Paul Rand: You’re newer to the University of Chicago and you’ve come into a Becker Friedman Institute here at the university, and you’re launching a brand new lab called the Development Innovation Lab. And I wonder if you can tell me a little bit about what you’re planning on doing.

Michael Kremer: The lab is very much along the lines we’ve been discussing is to use the tools of economics to try to take on practical problems. So of a product in low and middle income countries that will often be in ways that are in collaboration with other disciplines, whether that’s education, researchers, and health researchers, or agricultural researchers. But often I think the aim would be to have long-term collaborations with practitioners, whether that’s a government agency, whether that’s a nonprofit organization or a private firm to really try to iterate, try a variety of different approaches to adjust problems carefully and measure the impact, and then just keep continually improving, hopefully serving both the needs of our partner organizations, but also serve and provide generating knowledge, which would be trying to disseminate the world at large.

Episode List

What makes something memorable (or forgettable?) with Wilma Bainbridge (Ep. 138)

Psychologist discusses the Mandela effect and a new AI tool to predict what we remember—from art to people’s faces

Learning to speak to whales using AI, with David Gruber (Ep. 137)

Scientist discusses how Project CETI is deciphering the Morse code-like ‘alphabet’ of sperm whales

Storm warning: Why hurricanes are growing beyond measure, with Michael Wehner (Ep. 136)

Scientist proposes a new Category 6 for hurricanes, calculates the growing cost and impact of climate change

How to manifest your future using neuroscience, with James Doty (Ep. 135)

‘Mind Magic’ author explains the scientific research on how to train our brain to achieve our goals

Why we die—and how we can live longer, with Nobel laureate Venki Ramakrishnan (Ep. 134)

Nobel Prize-winning scientist explains how our quest to slow aging is becoming a reality

What dogs are teaching us about aging, with Daniel Promislow (Ep. 133)

World’s largest study of dogs finds clues in exercise, diet and loneliness

Where has Alzheimer’s research gone wrong? with Karl Herrup (Ep. 132)

Neurobiologist claims the leading definition of the disease may be flawed—and why we need to fix it to find a cure

Why breeding millions of mosquitoes could help save lives, with Scott O’Neill (Ep. 131)

Nonprofit's innovative approach uses the bacteria Wolbachia to combat mosquito-borne diseases

Why shaming other countries often backfires, with Rochelle Terman (Ep. 130)

Scholar examines the geopolitical impacts of confronting human rights violations

Can Trump legally be president?, with William Baude (Ep. 129)

Scholar who ignited debate over 14th Amendment argument for disqualification examines upcoming Supreme Court case

Course registration