Rachel Glennerster | Innovation and the Center for Global Development | Talks at Google

Hello, everybody. I'm Kent Walker, President of Global Affairs at Google, and delighted today to bring you a Talks at Google session featuring Rachel Glennerster, who is the President of the Center for Global Development in Washington, D.C. Rachel's had a very distinguished career working at the University of Chicago, working at— with the UK government on international development efforts. In senior roles, and then before that, serving as the executive director of J-PAL, one of the leading groups, labs around the world looking at the effectiveness of different kinds of development interventions. So without further ado, let me bring Rachel up to the stage, who has a presentation with regard to the ways that innovation can actually help in global development. Rachel.

Great. Thanks very much, Kent. So, thanks for joining. So, as you heard, I'm going to talk about innovation and development. So, I'll go through quickly why this is a new era for development and how R&D has been a major driver of improvements in people's lives around the world and whether AI can solve the innovation problem for low and middle income countries. So, as I think a lot of you know, we've had dramatic successes in development over the last 20, 25 years. And people are talking about now as a new era for development. They're mainly thinking when they say that about cuts in aid. But actually, while that's important, the much more important trend is that development has been a huge success, and therefore we need to think about it differently. So there's been dramatic falls in poverty, number of people living in poverty, in child mortality, even in the poorest of countries. And also importantly, countries have changed. So it's not just individual lives that have changed, but countries have changed in the sense that the number of countries who are low income status, and therefore really need a lot of help to to run, to be able to support the poor in their countries has fallen dramatically from 64 to 25 since 2000. And that means that a lot of the countries that we traditionally think of as developing countries are now middle income. They have a lot more money of their own to spend. And that means that development is a lot about influencing them. So, in this new era, it means two things. One is that the end of poverty is actually a doable task. Just people have calculated that 0.3% of world GDP would be enough to lift every single person above the poverty line in the world. Now, that's assuming you could target it perfectly, but it gives you a sense of the order of the challenge. It's also the case that middle-income countries, as I say, have got to be part of solving the problem. So everything from poverty, where they are currently spending much, much more money on relieving poverty than any aid programs. But they're also increasingly huge part of climate change. So at the moment, 69% of CO2 emissions come from low and middle income countries. Virtually none of that is from low income countries. It's mainly middle income countries. And even if you look cumulatively across, you know, since the Industrial Revolution, 38% of all carbon emissions come from low and middle income countries. So we can't solve these global problems without bringing in low and middle income countries. So many countries can self-fund their exit from poverty, but what they don't— what they're not doing themselves so far is investing a lot in the evidence, the R&D about what is effective. And also, they aren't investing in the technologies that can help. You know, we'll come on to some of those examples. And this table just shows you the percent of GDP that it would take to lift all the poor people in a given country above the poverty line. So, you see countries like India, you know, it's less than 1% would— is needed to lift everyone above poverty. So, you know, India has the funding to do that on its own, whereas somewhere like Nigeria, it's still still 8% of GDP. So I want to turn to the role of R&D in achieving the gains that we've just talked about. Now, it's not the only thing. So this is just a chart of growth in different kinds of countries. So the light blue is middle-income countries, and the dark is low-income countries. And you see this big, burst of growth from the 1990s to 2010s. And that's because of a lot of things other than technology. You had a lot of countries coming out of communism. You saw big expansions in trade, you know, joining the world economy, and that led to big improvements and, you know, getting rid of a lot of constraints in their policies. But technology has been a huge driver of improvements in life. So just to illustrate the magnitude of this, so yes, rising incomes have helped, help health, help improve health. But it's not just rising income. So like Vietnam's life expectancy today is the same as Italy's was in 1985. Right, so, when I was at college. And yet, Vietnam's income is 65% of what Italy's was then. So in other words, you're getting much, much better health for a given level of income.

Why?

Because of things, cheap, easy-to-use health technologies like vaccines and vitamin A and insecticide-treated bed nets. So I'm not talking necessarily about really complicated, you know, computers and AI, like these basic technologies were invented by someone, tested by someone, and really made a huge difference. But as we'll come on to, most of the R&D that discovers those technologies is done in— is funded by rich countries and And even if it's done in poorer countries, it's aimed at the rich country market because that's where the money is. And another reason these simple-to-use health technologies are so important is that even if you have highly trained doctors, the absenteeism rates in these countries are very high. For the higher trained doctors, and there's a massive no-do gap, by which I mean the doctors know what they ought to do but they don't do it. If you test them, someone with these symptoms, they know the right prescriptions, but when they actually face patients, uh, they don't follow those prescriptions. So that's why these simple, easy-to-deliver technologies that can be delivered by very low human, you know, less trained workers are so effective. Another kind of R&D that has been really important in driving these improvements is learning about what is effective, and this has really driven how governments spend their money. So this is just a chart of the number of new randomized trials in economics in, uh, that have been registered in economics in, um, in low and middle income countries. So a massive increase in the number of, of trials. And that's been really important in helping ensure that money gets spent wisely. So if you look at this picture of a classroom in Kenya, you might have an immediate reaction as to what is missing. What should we provide this class to improve outcomes? Any thoughts from the audience about what's missing in this classroom? Any technology is missing. Any technology is missing. Computers. More space. Yeah, they don't— it's very overcrowded. It's large classrooms. I've been in classrooms with 140 kids in them, for example. Paper and pen, like any equipment, any blocks or papers and pencil. I mean, often you have slates and chalk, like my dad used to have when he was at school in the UK. But yeah, lots of things that you could think are missing. And randomized trials were done on the impact of adding a lot of those things. Textbooks. No effect on learning. More teachers, no effect on learning. Computers in the classroom, no effect on learning. More equipment— I've got test tubes here, but, you know, could be blocks, Lego blocks, that kind of thing— no effect on learning. Just giving cash, just that you could use to spend on anything that you want in the classroom, No effect on learning. So doing these tests is really important because our intuition about what is missing can be wrong about what's important. And what turned out to be important here was changing how teachers teach. You've heard about the reading wars in the US. Unfortunately, that is true in many low- and middle-income countries, that they're They are using out-of-date pedagogies. And instead, structured pedagogy and teaching at the right level are two kinds of programs that have proved extremely effective. You train the teachers. Now, when you train the teachers, you will also give them books and some of these other things. But if you give the stuff without changing how they teach, there's no effect. And on edtech, some edtech works if you can get teachers to use it. But the biggest hurdle at the moment is teachers don't use it. So agriculture is an example where technology has been— again, relatively simple technology— has been really important at driving development. Most of the extreme poor live in rural areas. The Green Revolution brought in new ways, new seeds, new fertilizers, and you saw these big increases in yields in Asia and, and Latin America. Increased in yields, reduced food prices, improved health, and reduced fertility. And that combined led structural change. So there's some really important work done in this area where different seeds were introduced in different areas at different times, and so you can really follow the kind of macroeconomic impacts. So just a 10-year delay in the rollout of new seeds would have reduced GDP in 2010 by 17%. So huge impacts because it had all these knock-on effects. So we need these new technologies that are relevant. Is AI going to solve the problem? Maybe I'm biased here, but not without some economics. There's huge promise because it will reduce the cost of things like developing new seeds for Africa, new medicines. And so with lower costs of innovation, you would expect more. And indeed, I've talked to some people at Google, and they've said, well, we could use AlphaFold for some of these things, and then we'll put it out for free and give away the patents. But that's only the start of the process. Because if you're creating a new drug, someone has to take it through trials. You also have to take it through trials if you're doing a new seed. You have to get regulatory approval in all the different countries. It's actually easier for drugs because you can get the WHO to sign off and most countries will agree, whereas seeds you have to go country by country. You've got to set up distribution channels, do the marketing, and if the patent is being given away for free, then nobody has any incentive to do any of those things. So, you've got a risk there. So let me talk about the incentives for innovation and why they fail so dramatically, particularly for low and middle income. In general, they are really distortionary, but needed, but there's a solution. So the status quo on patents, so excuse me for a little bit of economics here. So imagine, as a lot of innovation is, low production cost, but a high innovation cost. It takes costs, you know, producing the first or the first, you know, 100 of something is quite expensive. But once you've got it going, you know, the production cost is quite low. A typical demand curve is, downward slope, you have few people who are willing to pay a lot, but most people will only buy it if the price is lower. So what you do as a monopolist with a patent, you set quite a high price, a little x on the demand curve, and you get a nice chunk of revenue that will allow you to cover all that innovation, that yellow innovation cost. But that misses out all those people down that demand curve. That whole triangle there is a lot of value to people who would have wanted to buy it at the cost of production, but don't get to buy it because you've set a high price in order to cover your innovation cost. So imagine an alternative, what we call an advanced market commitment. We have the same setup. Oh, sorry, let me just say that imagine if the top part of that curve, the people who are willing to pay a high price, are the people who live in rich countries. And a lot of the people who can only afford a lower price are in low and middle income countries. And then imagine that this is a product that is only useful in low and middle income countries. You know, it's a cassava or an improved, you know, sweet potato. Sweet potatoes in Africa are very different from our sweet potatoes. And then you might not get the product at all, right? Because nobody can afford to pay the really high price. Or it's a vaccine for, disease that is only prevalent in Africa. So what you can do is offer to pay a certain amount of top-up if something is invented. And then you set up— and then in exchange to that commitment to pay an innovator a top-up, in exchange for them charging a low price, and then they'll do a high quantity. So you're effectively, as a donor, paying the innovation cost, but only if people buy it and find it useful. And so you're making it cost-effective to do the innovation. People have an incentive to do the innovation, where do you get it out to lots of people, it solves the patent problem. Because you get all of that nice revenue of the AMC top-up, and you cover your production costs, and so you cover your innovation costs. So we've been working on how to do this, or what are some examples of the kind of innovations that we could do if we had that kind of incentive structure. So neonatal sepsis— you've probably heard of sepsis, bacterial infection in the blood. But it kills a lot of newborn babies. It's quite hard to tell whether they have sepsis, and they die really quickly. So in a matter of a day, you can die of sepsis. But the diagnostic takes 2 days to come back. So it's kind of useless. So at the moment— We both undertreat children who have sepsis, aren't picked up and die, and we overtreat people who don't have sepsis. And so you're getting a lot of antimicrobial drug-resistant bacteria that can affect— can come back and haunt, really, the whole world. And we're already— spending huge amounts of money trying to treat people with these drug-resistant diseases. And I met someone when I was talking about AMR, antimicrobial resistance, who had one of these. And she was 32, and there was nothing they could do because she couldn't— none of the drugs worked anymore. So what we're proposing is $60 million pool mechanism where you reward— you have $20 million milestone payment for someone who creates a diagnostic that works fast and can be done— used in clinics in India or Nigeria or South Africa. Then they get some support to scale it up. But then also they get paid, as in the previous slide, are in line with how many they actually sell. So that gives them an incentive to produce an innovation which is useful for everyone. So we estimate that will give— 12% of sepsis deaths would be averted if there was a proper diagnostic. And 54% fewer unnecessary antibiotic courses with a benefit-cost ratio of 78 to 1. So huge benefit-cost ratio. And people have the ability to do this, we think. There's just not an incentive. I mean, I've talked to a lot of innovators in India. And their response is, you know, I came into this area to work on diseases prevalent in India, but I can only make money if I work on cancer and things that the US— you know, I can only get money for my business if I'm working on something that is important to the US. So implications for development. Development has changed dramatically such that poverty in much of the world is an issue for domestic governments.. But those governments don't invest in R&D, both learning what works and new medicines and seeds and other technologies that are relevant to those environments. And I think it's 67% of R&D is done in rich countries and 27%— no, 63% is done in rich countries and 27% in China. So basically nobody is doing R&D outside those countries. So R&D investment, including on what works, is a key priority for philanthropy and multiplies the investments being made by middle-income countries. Now, we haven't talked that much about the poorest countries who are still left at the bottom. They are facing much, much tougher situations than the other countries. Like, why did they not get out of the, of the trap of poverty during that good time, uh, in the '90s and 2000s? Because they face real issues. Um, and so that's another priority. AI is reducing the cost of innovation, but it has to be paired with careful incentives, um, if we're gonna solve the most important problems in the world.

Thank you. Well, thank you all for coming. Thank you for joining online. It's been a real pleasure. Thank you, Rachel.

Thanks.