Episode 19: Dr. Tavneet Suri on Basic Income and Why Evidence and Facts Matter
This week on AI at Work, Rob May discussed the concept of basic income with Dr. Tavneet Suri from MIT Sloan School of Management. At its most base case, basic income means that the government is paying a flat fee every year to every citizen. Tune in to learn more about what this means and about Dr. Tavneet Suri’s research in sub-Saharan Africa.
Rob May, CEO and Co-Founder,
Dr. Tavneet Suri, Associate Professor,
Rob May: Hello everyone, and welcome to the latest edition of AI at Work. I'm Rob May, the co-founder and CEO of Talla, and we bring you this podcast so that if you're thinking about how AI might impact your job and your industry, we hope to have some guests on from time to time that can help you think through that.
Our guest today is Dr. Tavneet Suri from MIT Sloan School of Management. She is an expert on basic income. This podcast is going to be a little bit different in that we're actually going to talk about this concept of basic income, which is very popular in AI circles particularly as people think about what may happen if AI takes a lot of jobs. Would this be a possible solution, providing a basic income to people? Welcome Tavneet, can you give us a little bit about your background and how you got to Sloan?
Tavneet Suri: Sure. Thanks, Rob. It's really nice to be here on a freezing cold afternoon. I got to Sloan about 12 years ago, having finished grad school. I've got a PhD in Economics from Yale, so I finished grad school out in New Haven, Connecticut and then moved up here.
RM: Cool, and where and when did your interest in basic income start or come from?
TS: We started thinking about this a few years ago. I have to admit I don't work on the United States at all. I work mostly in sub-Saharan Africa. One of the things that we've been thinking about around basic income was whether it can help solve poverty and issues of poverty in developing countries. Together with a partner called Give Directly, we decided we'd try and do a study around basic income.
There's lots of stories out there about whether basic income means people will stop working because they have no incentive to work, and will it affect labor supply, and will it affect all these other things. I think the ultimate answer is you have to test it to see before you can understand that. Because as you probably know, with 100 people there's 100 different opinions about what to do. The list of things it could potentially do that we could justify could be quite long. We think people might invest more. We think people might stop working. We decided it would be useful to try and assemble a piece of evidence around whether this would be a good way to alleviate poverty in developing countries.
Give Directly raised a bunch of money for the actual basic income, and then we've been fund-raising for the study, which is now in progress. No results yet I'm afraid, but soon.
RM: For people that aren't familiar with the concept of a basic income. At its most base case, the government is paying a flat fee every year to every citizen. What are some of the other key features? When people talk about basic income, what are some of the other ideas or parts of basic income that come up that our listeners should be familiar with?
TS: Of course, the value of it matters. What's basic? How basic is basic? One question is around the quantity. One question is around how long, what frequency, and what form. I think in the developed world, of course, it's not supposed to be in addition to existing welfare programs. You're going to also take out those welfare programs. How you organize and manage that starts to show up. How do I reduce those bits and pieces as I bring the basic income in. In developing countries, of course, there's not a lot of existing welfare programs, so we don't need to deal with that piece of it. We had to decide what amount, what frequency, things like that.
I think the other piece is how you deliver it. In the US, this is probably easy. Everybody has a bank account. Not everybody, but a lot of people have bank accounts or some version of a prepaid credit card I can put money onto and stuff like that. In developing countries, delivering it is often extremely costly. I have to go hand people money. It turns out when you're handing small bits of money, the cost of delivering it is quite high. With mobile payments coming into the system in developing countries, you can now deliver money very cheaply over the mobile phone. That's the other piece, how are you actually going to implement it and get it to people.
RM: From a moral and ethical perspective, what's the number one argument that people like to make for basic income, and what's the number one concern that people might have about a basic income?
TS: I don't know that I have a good sense of everybody's moral and ethical concerns. I can tell you what excites me the most and what we might be my biggest worry on the downside. The reason we want to test is we don't know what the effects will be. We really want to understand what the effects are before we try to implement it at scale in a bunch of countries.
For me the biggest upside is that we think of basic income as covering your basic needs. If you're in a developing country --I think this is pretty translatable, though-- I'd want to cover food and health and education maybe. Not private school education, but basic can I get you through school kind of stuff, and rent . That's what I think of as your basic necessities, can I cover those? For me, the biggest upside is that if I can cover that, there's a huge potential for people to have upside risk benefits.
This is true in developing countries. I don't know if there's true evidence of this in the US, but poor people tend to heavily invest in things that alleviate downside risk. When I'm really poor, I don't care about upside risk. I really care about downside risk because I need to feed my kids. I need to send them to school. I need to pay for health care. I need to do all these things. So, they overvalue anything that prevents downside risk, which means you might make inefficient investments to protect yourself against downside risk, because you can't smooth out that risk yourself.
The biggest upside I'm excited about is that if I cover your basic needs, I've covered your downside risk. Now, people can go “Oh I can go do this high risk, high return thing that I could never do because I was really worried if it fails, I'm not going to be able to feed my family and my kids.” I feel like if we cover that, you allow people the opportunity to go do whatever they wanted to do in the world, right? They don't have to worry that if it fails, they're not going to be able to cover very basic things.
For me, that's the biggest upside. It feels like it's an opportunity for people to have a chance at something they wanted to do without worrying about the failure risk. As you go down the income chain for poor people, they have less access to credit markets and insurance markets and all these other pieces. Downside risk just becomes more and more important for this part of the population. That for me is the big upside, can we give people an opportunity to everything they've always wanted to do?
On the downside, at least in developing countries, if we give a whole bunch of people a basic income, they're going to go buy stuff, and prices might go up in local markets. You can imagine that.
RM: Yeah, that's the inflation.
TS: There might be an inflationary effect. I think that's one piece I'm worried about, and we will measure. The other piece is we flooded a bunch of communities with a bunch of money. If anybody outside the community hears, you'll worry there'll be a rise in crime or something. I think those are the downsides.
In developing countries people have talked a lot about the labor supply effects. Will people stop working? Will it make them in quotes "lazy"? Or will they just stop working because now they have this income? There's a suite of small literature in economics on cash transfers, where I give people cash transfers. It's not a basic income, it's not as regular, it's not any of that. But there, we don't see any labor supply effects. We don't see that when we give people cash, they stop working. But again it's not quite a basic income.
I know some people have concerns that there'd be effects on motivation and labor supply and all that sort of stuff. I think we have to test it to see.
RM: Yeah I agree. Tell us a little bit about this specific test that you're doing in Kenya, as much as you can. How is it structured and what are you testing and how long does it run?
TS: It's structured as a randomized control trial. Do you know what that is? It's basically using the methodology in clinical trials for drugs for policy. It's structured as a large randomized controlled trial. It is in approximately 300 villages in Kenya across two different counties. And in the communities in the study, we're doing it universally. Everybody gets it in those villages if they're allocated to a village that gets it. Which means we don't screen on income, nothing. It is truly universal in the community.
And then, we have different arms. We're testing different types of UBIs. Just because a lot of the issues around UBI are how do I design it? In one group, we're giving them a UBI for two years, and it's given out at pretty regular intervals. In one group, we're doing it for 12 years. And in one group, we're doing the two year amount, but as a lump sum.
You can imagine these design changes matter, right? A lump sum might matter if I want to go to start a business. I need a bunch of capital to go do it, as opposed to getting it in regular intervals. Then, I have to figure out how to save it and not spend it et cetera. We've got kind of three different types running against a control. Usually what we do is we do a baseline survey of a bunch of outcomes, and then we implement the randomized control trial and then we're going to go back. It is 12 years in one of the arms, so it's can be a very long running study.
RM: How far into that 12 years are you?
TS: We're very early, less than six months. If it was December, I would say six months. It's very early. We started implementation in the spring and rolled out. Like I said, in developing countries, one of the cost sides that you have to think about is how I deliver the money. This is all done on mobile money accounts. So basically, we handed out a whole bunch of phones to every adult in these communities and gave them a mobile money account so they can just get the money on their phone. We don't have to worry about driving around with a pot of cash to hand out.
That enrollment happened at the end of the spring into the summer. We're going to do the first follow up next spring and then another one two years later. The two year will be really interesting, because if you think of the two year and the 12 year arm, we told the 12 year ARM we're guaranteeing you a basic income for 12 years. Every adult in the house gets a guaranteed basic income for 12 years. We actually took a bunch of teenagers in the house, and we'll bring them onto the UBI over the next couple of years. We didn't go down too far because it's hard to know where you stop, but we took a bunch of teenagers and enrolled them. As they become 18, they'll get mobile money accounts and get onto this.
The difference between the two year and the 12 year, if you think of it, when I get to the two year point, they both got the same amount of money. The difference is, the 12 year arm knows they're getting it for 12 years. So if we see any differences, it's something about the fact that you told people they're guaranteed it for a long time. We want to understand, is it just the money changes? Or, do these expectations that I'm going to have it for 12 years change my behavior at year one because I'm like, “Oh I can do even riskier things. Or, I can go do something else.”
RM: Long term.
TS: Exactly. So, does the long term commitment matter? Even at two years? That's the nice side of doing a survey at two years to understand this is you'll get a sense of how much committing to a long term UBI matters for people. Then we're measuring, of course, every outcome under the sun. A bunch of wealth, and savings, and employment, and occupation, time use. Then, we're doing community surveys.
You can imagine, in poor countries, the schools aren't great, and the government doesn't spend enough on schools. But if we have a UBI, we all get together and say, “Hey, why don't we all put in a little bit of extra money and hire a new teacher or build some infrastructure at the school?” We're going to do a bunch of community surveys around how do the schools and clinics look and crime, and then measure prices as well in the market.
RM: How did you choose Kenya? And then, are you working with the Kenyan government pretty closely on this? What's that relationship like?
TS: Good question. We're working with this organization called Give Directly to raise the money for the transfers. This is like, we get approximate numbers, because right down to the single number I don't have at the top of my head. It's something like 20,000-odd people getting some sort of transfer. It's about $25 million of transfers over 12 years. It's quite a large sum of money. Give Directly raised that money for the transfers.
We're working with them in Kenya, and they've been working in East Africa. We need a community where mobile money is pretty prevalent because we want to distribute it over mobile phones, so that's one of the constraints. We have talked to the government a lot and work with them, but most of the money being disbursed is from donors that we've raised. They're not the most flush-in-cash government in general. For the test, we raised the money. We're hoping that if we find good positive effects, that there'll be at least an avenue to talk to the government about how to implement something like this at scale, potentially.
RM: When you started thinking about this and you did your initial research on universal basic income and what had been done to date, was there anything really interesting or surprising or did anything jump out at you, like, “Wow I didn't expect to find that!” Or, “Wow, these studies are all flawed in the same way.” Was there anything you could take from that initial survey of the existing literature?
TS: That's a good question. I think in general, the existing literature has been small. There were these experiments in the US, I want to say, circa the 70s. Which has been a while, and we've changed a lot since the 1970s. I think we just felt like there was a ton of discussion and a ton of debate and a ton of here's the upside and here's the downside and here's my favorite upside and my favorite downside without really trying to build an evidence base. And we felt like it was time to start to build an evidence base before we try to implement something that some of us for whatever gut reason believe is good and some of us for whatever gut reason feel is bad.
The 1970s experiments, I think, were well done, and had a range of effects. But it's been a while...
RM: Different world.
TS: It's a different world. We've changed a lot in demographics and in all sorts of stuff. I think it's not clear that we should try to decide this question based on those sorts of studies. For the developed world, there's few and far between. There are a couple of studies running in the US, or about to start in the US. I think it'll be really nice to try and pull together these different studies. There's a couple in Europe happening too. I think all of us are chatting to each other so that at least there's some set of outcomes we're all looking at so that we can aggregate the evidence a bit.
RM: Your study is relatively young, but is there anything that you can already take away from it or anything that you've learned? Have you learned anything about the response that people had to it?
TS: In general, there's nobody who refuses it. This notion that a bunch of people would refuse it in general doesn't seem to be true. Now of course, this is a much poorer population than in the US. We took a couple of the poorer counties in Kenya. It's not even the average Kenyan, it's the poorest of the Kenyans. There isn't really this notion that people would say no. I think in that sense, we've got everybody to get on accounts and things like that.
Do they truly believe we're going to give it to them for 12 years? I don't know. I can imagine if someone came to me, being skeptical. Really, 12 years? Do I think you're going to stick around that long? In that sense, we haven't seen a lot of refusal. I think the best set of evidence will be next summer when we've got our first survey under our belt with some outcomes.
RM: UBI has really become a big focus of the AI community and the tech community, and you hear a lot about it in Silicon Valley. When you hear people like Elon Musk and tech leaders out there talking about it and how we should be looking at it, do you have an opinion on whether those types of endorsements are a good thing for UBI or bad thing? Is it good that they're raising awareness? Is it bad that they're pushing people to do something that we don't fully understand yet? Do you think about that at all?
TS: Absolutely. To be honest, I think you should not be pushing something without evidence. That said, I'm an academic, so that's always going to be my position no matter what. I selected into having that position. I do think it's tough to see one side of the story without data and without evidence. I think evidence, in fact, is the best way to make these decisions. In the US, the UBI is going to be expensive. It's not going to be our UBI, because if we gave you the Kenyan UBI here, you'd laugh at us. It's not going to be a basic income, it's going to be a day or something. Not quite a day, but you get the idea.
I think figuring out how we're going to fund that is going to be important. Part of that is going to be it's going to substitute for other things. It almost has to. And there's a bunch of questions around whether you can make it cheaper by doing a little bit of targeting or a little bit of exclusion.
There's some literature in economics, again, in poor countries, where if you ask people to go to an office to pick up a cash transfer, the rich people like “Oh, I can't be bothered to do that.” There could be a bit of self targeting opt-out because it's not a big deal for you. If you make $1 million bucks a year-- or maybe more here, if you make $5 million bucks a year-- do you really need to go do this and get it? There's issues like that in the US, and I think we really need evidence before we try to design how to implement something like this.
Even in our case, we didn't want to just test a UBI against nothing, because well, what is that? Is it permanent? Forever? I don't know, is it for 10 years? Is it for 20 years? How do we give it to you? Do we give it to you every month? Or lump sum? So there's a bunch of these design issues that come up that I don't think are trivial and might have effects on how effective it is. All the more reason to kind of test different pieces of this and variance of those.
RM: Sure, although I would argue that in the United States, we don't exactly practice evidence-based government in general, in terms of how we set up our laws and policies. We'll see.
TS: It's something I would love to see more of, to be honest. I think it's really important that policy be that. But then of course, the politics does sometimes take over. I think the only way we kind of get back to that is to keep saying evidence matters, and facts matter, and really truly having good evidence about policies matter. I think there are some people who do appreciate that as policymakers. But yeah, I understand that evidence is tricky these days. Not just here, also in Europe.
RM: A lot of the tech community people fall into different camps. We were talking before the show a little bit about this idea that there may not be jobs for humans down the road. Feel free to opine on that piece if you want. But my question is, if that were to happen and AI takes over, AI does all the work, and puts humans out of work, is UBI something we should be looking at as a potential solution to that?
TS: I am not going to opine about whether tech will reduce jobs. I would leave that to the people who work on those pieces. There is evidence on automation, and happy to put you to some folks who could tell you more about that. In fact, MIT has a new task force on work of the future and what work in the future might look like. There's a bunch of faculty kind of trying to put together something. I'm on this task force, but I provide international perspective rather than the US.
I always think a UBI needs to be thought about for any country wanting to do some welfare. Whether that's because there's mass unemployment, or for any other reason, a UBI should be on the table as a design. When you do welfare systems, one of the things you worry about is something we call targeting. Even if I want to just reach poor people and not UBI, what does it cost me to reach them? What are the error rates?
We have these error rates. Some of what I hand out might be given to people who are not poor, which is kind of bad targeting. I might not reach some poor people, which is also bad targeting. The question is, even if you're doing pure welfare based targeted to people, if the targeting is expensive and imperfect, you might be better off doing something universal. I think it should always be on the table, irrespective of whether there is mass unemployment or not. And yes, if there is.
We're always doing something in the welfare space. I don't think we've gotten to a world where we're not doing any. Some European countries are doing a ton. We're always going to be somewhere in that space. I can't imagine that we ever go to a society that does none of that. Maybe we go to a society that does a lot more because there's lots of unemployment and things like that. We will always have to think about, do we want to target it, or do we want to do something universal? That all depends on the costs of the targeting, how expensive is it to do, and how good is it? Do I actually reach the people? Does that make sense?
RM: Let's say I'm a tech worker. I'm interested in this concept. I want to be an educated voter, and I think man, in the next decade, this is an issue that is going to be debated, maybe even for the United States. I want to be smart on this. What should I do? What are the best resources to learn about basic income if I'm not an academic? I don't want to read a bunch of papers with charts and math that I don't understand. For example, are there good books out there? Are there news outlets that cover this from time to time? Are there people I should follow? Do you have any ideas?
I should know the answer to this question, shouldn't I? In 10 years, the good thing is, we'll have results. We will write our crazy horrible 60 page academic papers. We will also have a lot of non-academic stuff, because a lot of our donors are in the space of wanting to have an impact on policy. It turns out, much as you don't like reading those 60 page papers, neither does any policymaker. They don't have the time. They have less time than you do, probably. I think there will be lots of consumable stuff coming out from our study.
There are a couple of studies in the US, and I think they will be in the same space where they will produce a bunch of stuff that is easily consumable. I would say watch out for the news. I'm sure the news will go after this, I have no doubt. We've already done a bunch of press around the study which doesn't even have results, like with you. I have no doubt that there will be plenty of stuff to read. I think at the moment, it's largely opinions.
The concept of a basic income is pretty easy to understand. I would just say make sure you check up all the bits and pieces of studies that are going on. They're non-study stuff, they're better communicated stuff.
RM: Great. Well, Dr. Tavneet Suri, thank you for coming on and we look forward to seeing the results of your work. For those of you listening, thank you for listening. If you have guests you'd like to see or questions you'd like us to ask in the future, please email us firstname.lastname@example.org and we'll see you next week. Thanks for listening.